Macro ReviewHow to read
Cross Asset · Drivers · Risk sentiment

Beta cost 27.0pp a year per unit across , inside its band

Is the market paying for risk, and what does risk cost to hold?

Weeks to 13 Nov 2026Prices to 16 Nov 2026Read on FridaysUpdated 18 Nov 2026
Reward for beta, four tenors−27.0ppS&P 500 members · 90% band −86.8 to +32.7 · 9th 2014 · 13 Nov 2026inside its band
48mean percentile of 9 readings priced in advance · high is dear · 13 Nov 2026priced cheaper than usual
spreads28mean percentile of eight spreads, 63 sessions · two above zero · 13 Nov 2026
3of 6 · price of risk, co-movement, positioning · 13 Nov 2026
Equity positioning+45.3%asset managers' net E-mini S&P, % of · 91st pct since 2010 · 10 Nov 2026
Consumer sentiment48.1Michigan index · 0th pct since 1978 · Sep 2026

How this page reads risk sentiment

From the line's pricing to how many families read , each part with its question and today's reading at the page's default settings

Exhibit 2 · The page's parts in reading order and what each one answers

Is the market paying for risk, and how is it priced? Each step feeds the next, and each box links to its section.
Source: Yahoo Finance; U.S. Treasury; Federal Reserve Board (H.10); Macro Review calculationscross.sentiment-15

Exhibit 3 · What each measure asks and how it is made and read

One row a measure, in the page's order; the name links to its section. The settings below change only the cards that say so.
MeasureQuestionHow it is madeHow to read it
The line's slopeWhat did a unit of beta earn?Each asset's return over the tenor, against the beta it had when the tenor began. A beta-2 asset earning 8% beside a beta-1 asset earning 4% is a slope of 4pp.Above the CAPM line, beta paid more than the market's own move implies.
The CAPM lineWhat should a unit of beta have earned?The market's return over cash across the same tenor.The benchmark the slope is read against.
The gap to the CAPM lineWhat did beta pay beyond the market's own move?The slope less the CAPM line. It is also the slope of each asset's alpha on its beta.Positive when high-beta assets beat what the market's move implies.
The interceptWhat did a zero-beta asset earn over cash?Where the line meets zero beta. Among stocks it moves against the slope.Drawn as a view and never read on its own.
The 10th, median and 90th percentile linesDo outcomes fan out at high beta?A line through each of three percentiles of returns at each beta. The spread is the 90th line's slope less the 10th's.A wider spread means outcomes fan out faster as beta rises.
Rank correlationDid alpha line up in beta order?Beta and alpha ranked across the assets, then correlated.+1 when alphas line up in beta order, −1 when reversed. One extreme asset cannot carry it.
The four tenors togetherWhat has beta paid across horizons?Each measure over one, three, six and twelve months, recent days weighing more within each. The four are then averaged so that each tenor has an equal say.Read with its 90% band. While the band holds zero, the sign is not settled.
Out of line with their betaWho moved against their beta?Assets ranked into quarters by beta and by alpha. With the market up, high beta in the bottom alpha quarter and low beta in the top are listed.A description of the window, not a signal.
The order gaugeDid the most volatile assets turn first?The rank correlation of each fund's own volatility with the sessions since it turned.Negative when the least volatile turned first.
The other five familiesDo spreads, prices, co-movement, positioning and surveys agree?Each is ranked against its own history; the co-movement index is read against zero.Counted with the reward for beta: how many of six sit on the risk-on side.

The settings

SettingOptionsWhat it changes
Tenor1, 3, 6 or 12 monthsThe cross-section and the leadership sections. The composite always draws all four.
UniverseThe S&P 500's members by default, the 18 sector funds, or funds across asset classesEvery section from the cross-section on.
AssetsAll, or one of the universe's presets: ex energy, ex information technology, ex commodities and energy, ex crypto, or one group of fundsEvery section from the cross-section on. Each preset is its own fit.
HistoryFive years, or since 2014How far back the time series draw. Nothing is refitted.

The headline, the findings, today's table and the count describe the page's default settings: the S&P 500's members, all assets, and three months for the cross-section.

Source: Yahoo Finance; U.S. Treasury; Federal Reserve Board (H.10); Macro Review calculationscross.sentiment-16

Exhibit 4 · Each block's reading at the default settings and the family count

The page's default settings: the S&P 500's members, three months for the cross-section, the four tenors with an equal say; week to 13 Nov 2026.
BlockMeasureTodayBand or percentileReadsWhere
Core pricingThe line's slope less the CAPM line, three months, pp−5.3−17.9 to +7.3inside its band→ Exhibit 5
Core pricingRank correlation of beta with alpha, three months−0.29−0.58 to ±0.00inside its band→ Exhibit 5
The four tenors togetherSlope, pp a year−27.0−86.8 to +32.79th percentileinside its band→ Exhibit 1
The four tenors togethercounted in the familiesGap to the CAPM line, pp a year−23.5−83.2 to +36.310th percentilerisk-off→ Exhibit 1
The four tenors togetherRank correlation of beta with alpha−0.23−0.48 to +0.0212th percentileinside its band→ Exhibit 1
LeadershipMembers out of line with their beta, three months, of 49937–22 high beta, 15 low beta→ Exhibit 10
LeadershipRank correlation of own volatility with the time since turning−0.26–the least volatile turned first→ Exhibit 11
Across asset classesRank correlation of beta with alpha, four tenors−0.15−0.36 to +0.0638th percentileinside its band→ Exhibit 1
The other familiesPrice of riskthe mean percentile of the readings priced in advance48th percentile13 Nov 2026–risk-onrisk-on when below 50→ Exhibit 14
The other familiesRisk-on spreadsthe mean percentile of the eight spreads28th percentile13 Nov 2026–risk-offrisk-on when at or above 50→ Exhibit 12
The other familiesCo-movementthe co-movement index, the latest complete week−0.50 sd13 Nov 2026–risk-onrisk-on when below zero→ Exhibit 16
The other familiesPositioningasset managers' net E-mini S&P position, its percentile91st percentile10 Nov 2026–risk-onrisk-on when at or above 50→ Exhibit 15
The other familiesSurveysMichigan consumer sentiment, its percentile0th percentileSep 2026–risk-offrisk-on when at or above 50→ Exhibit 15
The countFamilies reading risk-on, of six3 of 6–3 on the measure used until 25 Oct 2026→ Exhibit 13
About this exhibit

A band is the 90% band of the reading; while it holds zero the sign is not settled. A percentile is the reading's place among the weeks since 2014. Each family reads risk-on on its own side of its own line, and the count treats the six as equals, which they are not. The reward for beta is counted through its four-tenor gap.

Source: State Street Global Advisors; Yahoo Finance; U.S. Treasury; Cboe Global Indices; University of Michigan; CFTC; Macro Review calculationscross.sentiment-14

What the evidence says

What a unit of beta paid against the CAPM line, who carried it, and how many of the six families read
  • Across to 13 Nov 2026, beta cost 27.0pp a year per unit, inside its band. The 90% band runs from −86.8 to +32.7pp a year. The reading is the 9th percentile of the weeks since 2014. It has stayed below zero since 16 Oct 2026, five weeks. All four tenors read below zero, from −56.3 over one month to −4.9 over twelve. Recent days weigh more within each tenor, and each tenor has .→ Exhibit 1
  • Beyond the CAPM line beta paid −23.5pp a year across four tenors (10th 2014). That is the reward for beta's reading in the , so it reads risk-off. Three of the six families read risk-on on 13 Nov 2026. On the measure used until 25 Oct 2026, the count is three.→ Exhibit 4
  • Across 44 funds in every asset class, beta and alpha ranked at −0.15 over four tenors. Its 90% band, −0.36 to +0.06, holds zero. Across the same funds the slope read −11.7pp a year. Without commodities and energy it read −19.4pp.→ Exhibit 1
  • Over the three months to 13 Nov 2026, the members' line paid −8.9pp per unit of beta, against −3.7pp on the CAPM line. The gap of −5.3pp sits inside its band, −17.9 to +7.3pp. Beta explained 15.7% of the spread of returns. The was −0.29, its band −0.58 to ±0.00.→ Exhibit 5
  • Over three months one of the 18 sector funds moved against their beta, with the market down. High beta that held up: Consumer discretionary. Among the members, 37 of 499 sit in those corners. The list describes the window; it is not a signal.→ Exhibit 10
What would change this view
What would change this viewTriggerTodayDecided by
The composite's band clears zerothe members' four-tenor slope's 90% band wholly above or below zero−86.8 to +32.7pp (13 Nov 2026)not crossedeach Friday's composite
The rank band leaves zerothe three-month rank correlation's 90% band wholly on one side of zero−0.58 to ±0.00 (13 Nov 2026)not crossedeach Friday's line
The families tipfamilies reading risk-on ≥ 4 or ≤ 2 of six3 (13 Nov 2026)not crossedthe week to 20 Nov 2026
The rises through 50the price of risk ≥ 5048 (13 Nov 2026)not crossedthe week to 20 Nov 2026
The reasoning behind each point
  • Four tenors
    Across four tenors to 13 Nov 2026, beta cost 27.0pp a year per unit, inside its band.

    The 90% band runs from −86.8 to +32.7pp a year. The reading is the 9th percentile of the weeks since 2014. It has stayed below zero since 16 Oct 2026, five weeks. All four tenors read below zero, from −56.3 over one month to −4.9 over twelve. Recent days weigh more within each tenor, and each tenor has an equal say.

  • The count
    Beyond the CAPM line beta paid −23.5pp a year across four tenors (10th percentile since 2014).

    That is the reward for beta's reading in the family count, so it reads risk-off. Three of the six families read risk-on on 13 Nov 2026. On the measure used until 25 Oct 2026, the count is three.

  • Across assets
    Across 44 funds in every asset class, beta and alpha ranked at −0.15 over four tenors.

    Its 90% band, −0.36 to +0.06, holds zero. Across the same funds the slope read −11.7pp a year. Without commodities and energy it read −19.4pp.

  • Three months
    Over the three months to 13 Nov 2026, the members' line paid −8.9pp per unit of beta, against −3.7pp on the CAPM line.

    The gap of −5.3pp sits inside its band, −17.9 to +7.3pp. Beta explained 15.7% of the spread of returns. The rank correlation of beta with alpha was −0.29, its band −0.58 to ±0.00.

  • Out of line
    Over three months one of the 18 sector funds moved against their beta, with the market down.

    High beta that held up: Consumer discretionary. Among the members, 37 of 499 sit in those corners. The list describes the window; it is not a signal.

  • Order
    The most volatile funds turned last in the current move: a rank correlation of −0.26 on 16 Nov 2026.

    33 of the 43 funds have turned. In past high-beta sell-offs the median reading was +0.23.

  • Price of risk
    Risk is priced cheaper than usual: the readings priced in advance average the 48th percentile of their own histories (13 Nov 2026).

    Risk-on spreads averaged the 28th percentile over three months. Two of the eight spreads paid.

  • Holdings
    Asset managers held 45.3% of E-mini S&P open interest net long on 10 Nov 2026, the 91st percentile since 2010.

    Michigan consumer sentiment read 48.1 in Sep 2026, the 0th percentile of the months since Jan 1978. Margin debt was +37.2% on the year in Aug 2026 (89th percentile). AAII bulls less bears is not shown: the survey's history is for members, and its results page sits behind a bot wall the site does not work around. NAAIM exposure index is not shown: subscription only since 2026. ICI weekly equity fund flows is not shown: the publisher refuses the site's requests.

Next: Factor analysis — which factors drove one asset's return

Source: Yahoo Finance; U.S. Treasury; Federal Reserve Board (H.10); Macro Review calculationscross.sentiment-read

Core pricing: what a unit of beta paid

Every asset's return against its beta at the tenor, the line through them and its band, the CAPM line, the gap between them, and how beta and alpha rank
Tenor
Universe
AssetsHistory
  • Beta explained 15.7% of the members' returns over three months. The line paid −8.9pp per unit of beta, against −3.7pp on the CAPM line. Its 90% band runs from −21.5 to +3.7pp. → Exhibit 5
  • At high beta outcomes fanned out faster: the 90th percentile line's slope ran 6.5pp above the 10th's. The spread compares the best and the worst outcomes at each beta, not the average. → Exhibit 5

Exhibit 5 · Each member's three-month return against its beta

Loading the readings…

Source: Macro Review calculationscross.sml-3

Exhibit 6 · Each tenor's line in numbers, with the four tenors together

Loading the readings…

Source: Macro Review calculationscross.sml-1

The

Each core measure over one, three, six and twelve months, recent days weighing more, each tenor with , every Friday since 2014
  • All four tenors read below zero. One month −56.3, three months −51.0, six months −29.4 and twelve months −4.9pp a year. → Exhibit 7
  • With equal weights the composite reads −35.4pp a year. Equal weights let the one-month tenor make about half the line's weekly movement. An equal say gives each tenor about a quarter. → Exhibit 1
  • Resampling the market with the same days narrows the gap's band to −62.9 to +16.0pp. With the market's move held as known, the band runs from −83.2 to +36.3pp. → Exhibit 1

Exhibit 1 · The reward for beta across four tenors, weekly

Loading the readings…

Source: Macro Review calculationscross.sml-4

Exhibit 7 · Each tenor's part of the four-tenor reading

Loading the readings…

Source: Macro Review calculationscross.sml-15

Who carried the move, and who broke from their beta

Each group's part of the slope, how alpha ranks moved between a month and a year, the assets with their beta, and the order in which funds turned
  • Health Care carried +0.3pp of the members' three-month slope of −8.9pp. Information Technology pulled the other way, −5.3pp. Without information technology the four-tenor slope reads −17.7pp a year. → Exhibit 8
  • 132 members ranked in the top half of alpha over both one month and twelve, against about 124 if the two were unrelated. The two alpha ranks correlate at +0.03. → Exhibit 9
  • 33 of 43 funds have turned in the current move. Own volatility and the time since turning rank at −0.26. Ranked by beta instead, the reading is +0.14. → Exhibit 11

Exhibit 8 · Each sector's part of the line's slope

Loading the readings…

Source: Macro Review calculationscross.sml-14

Exhibit 9 · Each member's one-month alpha rank against its twelve-month rank

Loading the readings…

Source: Macro Review calculationscross.sml-16

Exhibit 10 · Members out of line with their beta

Loading the readings…

Source: Macro Review calculationscross.sml-17

Exhibit 11 · Rank correlation of own volatility with sessions since each fund turned

On 16 Nov 2026: 33 of 43 funds have turned; volatility and time since turning ranked at −0.26, beta at +0.14.

About this exhibit
43 funds; a fund has turned once it sits a month's typical move below its three-month high; daily, the last two years. Below: the high-beta spread's fall from its three-month high.
Each day the funds are ranked twice: by their own volatility over the last three months, and by how many sessions ago they turned (0 for a fund that has not). Above zero, the most volatile turned first; below, they turned last. The dashed line ranks beta instead of volatility. On the 1121 sessions since 2010 when the high-beta spread sat 8% or more below its high, the median reading was +0.23; today's is −0.26, the 6th percentile among them.
Source: Yahoo Finance; U.S. Treasury; Macro Review calculations · data to 16 Nov 2026cross.betas-8

The other five families

spreads, the count through time, the , what investors say and hold, and co-movement, each against its own history
  • Three of the six families read risk-on on 13 Nov 2026. On the risk-on side: price of risk, co-movement and positioning. On the measure used until 25 Oct 2026, the count is three. → Exhibit 4
  • Risk-on spreads averaged the 28th percentile; the price of risk the 48th. High spreads mean risk paid; a high price of risk means it is dear to hold. → Exhibit 12

Exhibit 12 · Each risk-on spread's 63-session return, in its own standard deviations

Each spread's 63-session return in multiples of its own three-year standard deviation, to 13 Nov 2026; blue paid, orange lost
About this exhibit
Bars of each risk-on spread's 63-session return in standard deviations of its own last three years, blue paid and orange lost, with the market over cash beside them; a toggle by market pattern.
What risk-on spreads paid over the last 3M
Each spread's 63-session return in standard deviations of its own last three years; to 13 Nov 2026.

Each spread is the return of its long side less its short side (or over cash), in log terms, divided by its own trailing three-year daily standard deviation times the square root of the window, so every spread reads in standard deviations of its own recent moves. Blue bars paid, orange lost. The mean of the eight spreads' 63-session percentiles is 28, and 2 of the eight are above zero over 63 sessions.

The market over cash is printed beside the eight and kept out of their mean by design: the spreads ask what risk paid against safety, not whether the market rose.

Source: Yahoo Finance; U.S. Treasury; Federal Reserve Board (H.10); Macro Review calculationscross.sentiment-6

Exhibit 13 · How many of the six families read risk-on, weekly

Every Friday for five years; the dashed line counts the reward for beta as the page did until 25 Oct 2026.

Today 3 of 6 families read risk-on, and 3 on the measure used until 25 Oct 2026. Since 25 Dec 2015 the two counts differ in 25% of weeks, today's measure higher in 15%.

About this exhibit

Each family reads risk-on on its own side of its own line. The reward for beta now reads through the four tenors' gap to the CAPM line, its percentile at or above 50. Until 25 Oct 2026 it read through the members' three-month gap within sectors. Both counts print until the next issue.

Source: Yahoo Finance; Cboe Global Indices; University of Michigan; CFTC; Macro Review calculationscross.sentiment-17

Exhibit 14 · Readings of what risk costs to hold, against their own history

%: percentile of each reading's own history, high = dear; ● latest, ○ a month ago; week to 13 Nov 2026
ReadingTodayPercentile, high = dearPercentileFrom
Variance premium, S&P 500, one monthsince 2024 · joins at two years-3.9 vol points16 Nov 20265thS&P 500 option chains
Put skew, S&P 500, 25-delta, 25 dayssince 20157.0 vol points16 Nov 202657thSkew & term structure
Excess bond premiumsince Jan 1991-39 bpSep 202613thFinancial conditions
Corporate over Treasury, 10 yearssince Jan 20000.78 ppSep 20266thCorporate spreads
Financial stress index (OFR)since 2000-1.8812 Nov 202639thOffice of Financial Research
VIXsince 199019.313 Nov 202661stCboe Global Indices
VIX three months less one monthsince 2009 · low is dear1.3 points13 Nov 202677thCboe Global Indices
Cboe SKEW indexsince 199014713 Nov 202694thCboe Global Indices
VVIX, the VIX's own volatilitysince 200611416 Nov 202675thCboe Global Indices
Implied correlation, one month (COR1M)since 200616.416 Nov 20268thCboe Global Indices
The price of riskmean of the 9 readings with two years behind them47.8priced cheaper than usual48since 2014
About this exhibit
A table of the readings priced in advance (variance premium, put skew, excess bond premium, corporate spread, OFR stress index, VIX, VIX term spread, SKEW, VVIX, implied correlation), each as a percentile of its own history, high = dear, with the price-of-risk composite.
What risk costs to hold: each reading against its own history
Percentile of each reading's own history, high = dear; ● today, ○ a month ago; week to 13 Nov 2026.

Each reading is read on Fridays as it was known — a monthly one from its publication date — and ranked against its own history from its first week, signed so that a high rank means risk is dear to hold. The price of risk is the mean over the 9 readings with two years behind them; a younger reading is shown with its start and joins when it has two years. Below 30 risk is priced cheaply, above 70 dearly.

Not shown: Chicago Fed NFCI, risk subindex (the Chicago Fed serves the index through a token-gated API; its static file stops at 24 Apr 2026); Cboe total put/call ratio (not among the index histories Cboe publishes).

Source: Cboe Global Indices; Office of Financial Research; Federal Reserve Board; Macro Review calculationscross.sentiment-12

Exhibit 15 · Consumer sentiment and margin debt against their histories, with futures positioning

Each reading's latest value, its change, and its rank (%) since its start; ● latest
ReadingLatestChangePercentile since its startPercentileReads
Consumer sentiment (Michigan)since Jan 197848.1Sep 2026−3.6since Aug 20260thrisk-off
Margin debt, change on the year (FINRA)since Jan 199837.2%Aug 2026−1.4since Jul 202689thrisk-on
Asset managers' net E-mini S&P position (CFTC)since 5 Jan 2010 · Positioning45.310 Nov 2026−2.1since 13 Oct 202691strisk-on
About this exhibit
A table of three readings (consumer sentiment, margin debt's change on the year, asset managers' net E-mini S&P position) with the latest value, the change, the percentile since its start and the side it reads.
What investors say and hold
Each reading's latest value, its change, and its percentile since its start; ● today.

Michigan's index of consumer sentiment is shown four weeks after its survey month; FINRA's margin debit balances as a change on the year, about seven weeks after the month ends; the CFTC's asset managers' net position in E-mini S&P futures as a share of open interest, from Positioning. A high percentile reads risk-on for all three.

Not shown: AAII bulls less bears (the survey's history is for members, and its results page sits behind a bot wall the site does not work around); NAAIM exposure index (subscription only since 2026); ICI weekly equity fund flows (the publisher refuses the site's requests).

Source: University of Michigan; FINRA; CFTC; Macro Review calculationscross.sentiment-13

Exhibit 16 · The weekly co-movement index of six risk-on and risk-off inputs

In standard deviations; above zero, the week moved as risk-off; every Friday for five years.

In the week to 20 Nov 2026 the index read +1.49 standard deviations, above zero: a risk-off week.

About this exhibit

The index is the shared move of six daily inputs: US and international stocks, high yield against investment grade, the dollar, gold and the change in the S&P 500's realised volatility. Each is signed so that a risk-off day is positive, and the week sums its days (after Chari, Dilts Stedman and Lundblad, 2023). The family reads risk-on when the latest week reads below zero.

Source: Yahoo Finance; Cboe Global Indices; Macro Review calculationscross.sentiment-18

Sensitivities and past sell-offs

The default reading refitted within sectors, on log returns, without its most influential asset and on the carried beta; and the order in which assets fell in past sell-offs

Exhibit 17 · The line's slope refitted four ways

Loading the readings…

Source: Macro Review calculationscross.sml-18

Exhibit 18 · Rank correlations of beta and volatility with each sell-off's order

26 high-beta unwinds of 8% or more since 2015, 16 assets; sessions from the peak until each asset was halfway down.
PeakTroughFall, %AssetsBeta with the fallBeta with the timingVolatility with the timingThe same without cryptoBitcoin less the Nasdaq-100 fund, sessions
5 Oct 20265 Nov 2026−26.416−0.95−0.20−0.10−0.13±0
22 Jun 202629 Jul 2026−19.014−0.90−0.22−0.31−0.04−6
2 Jun 202610 Jun 2026−11.615−0.88+0.43+0.35+0.63−3
6 Jan 202630 Mar 2026−11.816−0.52−0.36−0.64−0.48−29
29 Oct 202520 Nov 2025−11.215−0.81+0.19−0.07+0.19−3
23 Jan 20254 Apr 2025−27.716−0.94−0.87−0.94−0.92−9
16 Jul 20246 Sept 2024−16.014−0.55+0.16−0.62−0.39−21
18 Jul 20231 Nov 2023−15.315−0.70−0.56−0.59−0.60–
2 Feb 202325 Apr 2023−15.316−0.36+0.53+0.60+0.45+5
15 Nov 202228 Dec 2022−13.916−0.96−0.33−0.36−0.09−25
12 Aug 202223 Sept 2022−10.816−0.92−0.67−0.71−0.58−8
16 Feb 20221 Jul 2022−23.216−0.89−0.51−0.34−0.42+52
8 Nov 20217 Mar 2022−18.316−0.77−0.78−0.63−0.52−32
1 Jun 202119 Aug 2021−15.013−0.65−0.43−0.67−0.46−64
12 Mar 202120 Apr 2021−11.012−0.50−0.11−0.37−0.50−3
14 Jan 202127 Jan 2021−9.115−0.51−0.32−0.71−0.54−11
8 Jun 20209 Jul 2020−19.413−0.30+0.21−0.25+0.06–
2 Jan 202018 Mar 2020−29.416+0.16−0.63−0.07−0.40+3
24 Jul 20198 Oct 2019−13.215−0.12+0.16−0.45−0.15−13
23 Apr 201912 Aug 2019−15.115−0.02+0.20−0.52−0.38−25
23 Jul 201821 Dec 2018−21.115−0.46−0.58−0.57−0.80+12
6 Jun 201817 Aug 2018−8.911−0.54+0.13−0.27−0.01−45
25 Jan 20177 Jun 2017−12.19−0.46−0.63+0.33+0.03+1
27 Apr 20165 Jul 2016−15.412−0.36−0.23−0.27−0.30+37
16 Sept 201511 Feb 2016−25.413−0.46+0.10−0.04−0.04–
1 May 201529 Sept 2015−18.313−0.56−0.34−0.65−0.65–
Median−0.55−0.28−0.36−0.38–

Bitcoin was halfway down at least two sessions before the Nasdaq-100 fund in 15 episodes and after it in 5; as often as that by chance alone with probability 0.021.

About this exhibit
Each episode opens when the level sits 8% below its 63-session high; its peak is that high and its trough the low before it recovers a third of the way. For each asset: its beta at the close before the peak, its volatility over the 63 sessions before, its own fall, and the sessions until it was halfway down. Below zero, high-beta or volatile assets fell first or furthest.
Source: Yahoo Finance; U.S. Treasury; Macro Review calculations · data to 13 Nov 2026cross.betas-9

Reconciliation to the issue

The retired gauge — the slope and intercept composites and the four state names — week by week to today, with the 's week, and the September figures for the line

Exhibit 19 · Earlier risk-state names and scores, each week since 2015

State name and both composite scores under both weightings, 25 Dec 2015 to 13 Nov 2026; · marks a week within 5 points of a line
Week toyear-weightedmonth-weightedScores, year-weightedScores, month-weightedAgree?
13 Nov 2026defensivedefensive40.3 · 61.828.0 · 69.0yes
6 Nov 2026defensivedefensive35.1 · 62.420.4 · 69.3yes
30 Oct 2026defensivedefensive43.3 · 62.318.0 · 80.9yes
23 Oct 2026risk-on ·defensive53.5 · 46.123.6 · 73.4no
16 Oct 2026risk-ondefensive55.5 · 43.632.9 · 61.9no
9 Oct 2026risk-onrisk-on65.4 · 30.758.1 · 35.2yes
2 Oct 2026risk-onrisk-on69.2 · 23.367.1 · 22.8yes
25 Sept 2026risk-onrisk-on65.9 · 27.359.4 · 26.4yes
18 Sept 2026risk-onde-risking58.0 · 37.341.9 · 38.1no
11 Sept 2026 · the report's weekrisk-ondefensive ·59.1 · 38.331.9 · 51.2no
4 Sept 2026risk-ondefensive62.4 · 40.634.0 · 64.0no
28 Aug 2026risk-ondefensive61.4 · 43.545.0 · 58.0no
21 Aug 2026risk-ondefensive65.3 · 41.941.9 · 66.6no
14 Aug 2026risk-onbroad rally ·69.1 · 39.450.3 · 57.2no
7 Aug 2026risk-onbroad rally ·73.8 · 33.350.7 · 57.1no
31 Jul 2026risk-ondefensive67.5 · 37.131.8 · 67.2no
24 Jul 2026risk-ondefensive68.9 · 37.033.9 · 68.4no
17 Jul 2026risk-ondefensive73.2 · 34.941.6 · 61.3no
10 Jul 2026risk-onrisk-on85.2 · 18.074.3 · 31.7yes
3 Jul 2026risk-onrisk-on84.4 · 20.264.7 · 42.3yes
26 Jun 2026risk-onrisk-on87.2 · 17.976.7 · 29.0yes
19 Jun 2026risk-onrisk-on95.7 · 6.395.5 · 6.8yes
12 Jun 2026risk-onrisk-on91.5 · 12.383.1 · 25.8yes
5 Jun 2026risk-onrisk-on88.7 · 12.972.9 · 23.3yes
29 May 2026risk-onrisk-on95.1 · 6.892.7 · 9.8yes
22 May 2026risk-onrisk-on91.4 · 11.182.7 · 19.8yes
15 May 2026risk-onrisk-on89.2 · 12.085.8 · 14.4yes
8 May 2026risk-onrisk-on95.3 · 6.495.7 · 5.3yes
1 May 2026risk-onrisk-on91.4 · 13.991.3 · 14.7yes
24 Apr 2026risk-onrisk-on91.5 · 12.890.4 · 16.3yes
17 Apr 2026risk-onrisk-on86.8 · 22.485.8 · 25.2yes
10 Apr 2026risk-onrisk-on79.7 · 31.872.2 · 33.7yes
3 Apr 2026risk-onrisk-on67.7 · 39.460.2 · 30.4yes
27 Mar 2026risk-onde-risking ·58.5 · 43.349.3 · 29.9no
20 Mar 2026risk-onrisk-on68.2 · 34.260.8 · 23.0yes
13 Mar 2026broad rally ·defensive ·52.7 · 56.529.3 · 51.8no
6 Mar 2026broad rally ·defensive53.3 · 58.433.6 · 64.2no
27 Feb 2026broad rallydefensive55.7 · 64.127.6 · 85.6no
20 Feb 2026broad rallydefensive ·63.9 · 58.247.6 · 74.6no
13 Feb 2026broad rallydefensive55.6 · 62.537.7 · 80.2no
6 Feb 2026broad rallydefensive59.7 · 58.942.8 · 80.1no
30 Jan 2026risk-onbroad rally ·64.9 · 44.762.8 · 53.2no
23 Jan 2026risk-onrisk-on ·67.2 · 40.167.0 · 47.8yes
16 Jan 2026risk-onrisk-on ·70.2 · 41.770.9 · 48.5yes
9 Jan 2026risk-onbroad rally66.8 · 43.558.9 · 56.4no
2 Jan 2026risk-onrisk-on63.8 · 39.763.0 · 37.6yes
26 Dec 2025risk-onrisk-on65.4 · 38.874.3 · 29.5yes
19 Dec 2025risk-onrisk-on63.4 · 36.870.6 · 36.8yes
12 Dec 2025risk-onrisk-on66.1 · 31.563.3 · 38.3yes
5 Dec 2025risk-onrisk-on69.1 · 27.464.8 · 42.9yes
28 Nov 2025risk-ondefensive60.8 · 34.539.1 · 63.1no
21 Nov 2025risk-on ·de-risking54.8 · 34.941.9 · 40.3no
14 Nov 2025risk-onrisk-on62.2 · 32.765.2 · 33.0yes
7 Nov 2025risk-onrisk-on65.6 · 30.757.6 · 34.9yes
31 Oct 2025risk-onrisk-on72.1 · 24.174.7 · 19.5yes
24 Oct 2025risk-onrisk-on ·65.3 · 33.754.1 · 50.0yes
17 Oct 2025risk-onde-risking ·63.0 · 32.548.7 · 44.5no
10 Oct 2025risk-onrisk-on62.9 · 29.659.2 · 25.8yes
3 Oct 2025risk-onrisk-on71.5 · 29.176.5 · 28.1yes
26 Sept 2025risk-onrisk-on69.7 · 31.475.4 · 27.5yes
19 Sept 2025risk-onrisk-on74.5 · 27.483.6 · 22.4yes
12 Sept 2025risk-onrisk-on70.5 · 32.764.9 · 41.3yes
5 Sept 2025risk-onrisk-on ·70.6 · 34.369.2 · 46.2yes
29 Aug 2025risk-onbroad rally ·65.5 · 40.663.3 · 53.4no
22 Aug 2025risk-onrisk-on ·64.6 · 44.463.6 · 47.5yes
15 Aug 2025risk-onrisk-on ·66.3 · 37.758.1 · 46.7yes
8 Aug 2025risk-onrisk-on71.7 · 29.662.9 · 30.6yes
1 Aug 2025risk-onrisk-on66.8 · 29.961.9 · 29.8yes
25 Jul 2025risk-onbroad rally ·62.8 · 43.368.9 · 51.2no
18 Jul 2025risk-onrisk-on64.7 · 37.083.1 · 29.0yes
11 Jul 2025risk-onrisk-on65.3 · 42.681.8 · 30.8yes
4 Jul 2025risk-onrisk-on67.3 · 39.184.0 · 26.8yes
27 Jun 2025risk-onrisk-on66.2 · 36.582.9 · 24.6yes
20 Jun 2025risk-on ·risk-on52.3 · 44.570.8 · 26.8yes
13 Jun 2025risk-on ·broad rally ·52.5 · 47.351.1 · 51.3no
6 Jun 2025risk-onrisk-on62.2 · 38.080.6 · 25.2yes
30 May 2025risk-on ·risk-on54.9 · 39.376.0 · 25.3yes
23 May 2025risk-on ·risk-on54.1 · 34.874.2 · 23.4yes
16 May 2025risk-on ·risk-on53.4 · 40.675.3 · 29.5yes
9 May 2025de-riskingrisk-on ·43.8 · 44.754.7 · 46.7no
2 May 2025de-riskingrisk-on ·43.0 · 43.654.5 · 28.0no
25 Apr 2025de-riskingde-risking35.7 · 42.937.7 · 28.7yes
18 Apr 2025defensive ·de-risking25.0 · 50.418.1 · 38.9no
11 Apr 2025de-risking ·de-risking29.2 · 45.729.8 · 35.7yes
4 Apr 2025de-riskingde-risking23.0 · 44.016.8 · 30.1yes
28 Mar 2025defensivedefensive24.9 · 63.220.7 · 56.7yes
21 Mar 2025defensive ·de-risking ·29.7 · 53.818.1 · 49.0no
14 Mar 2025de-risking ·de-risking30.1 · 49.921.5 · 43.1yes
7 Mar 2025defensive ·de-risking ·32.6 · 51.922.8 · 49.2no
28 Feb 2025defensive ·defensive40.0 · 52.126.9 · 59.6yes
21 Feb 2025defensive ·defensive46.9 · 51.330.5 · 60.3yes
14 Feb 2025risk-on ·defensive ·55.2 · 49.247.1 · 56.4no
7 Feb 2025risk-on ·defensive56.2 · 48.741.1 · 63.3no
31 Jan 2025broad rally ·defensive52.5 · 51.843.6 · 64.3no
24 Jan 2025risk-on ·broad rally59.6 · 49.657.0 · 57.6no
17 Jan 2025risk-onrisk-on ·62.0 · 43.250.1 · 44.1yes
10 Jan 2025risk-onrisk-on60.3 · 35.956.8 · 21.6yes
3 Jan 2025risk-onrisk-on67.2 · 30.267.3 · 17.5yes
27 Dec 2024risk-onrisk-on62.9 · 35.671.0 · 15.6yes
20 Dec 2024risk-onrisk-on65.7 · 35.171.6 · 17.4yes
13 Dec 2024risk-onrisk-on69.1 · 35.467.4 · 27.5yes
6 Dec 2024risk-onrisk-on73.8 · 39.562.2 · 42.4yes
29 Nov 2024broad rally ·broad rally ·65.2 · 54.951.9 · 71.0yes
22 Nov 2024broad rally ·broad rally67.0 · 51.756.2 · 61.3yes
15 Nov 2024risk-on ·risk-on ·63.9 · 48.753.5 · 45.4yes
8 Nov 2024risk-onrisk-on ·77.0 · 43.370.7 · 45.1yes
1 Nov 2024risk-onrisk-on75.5 · 39.469.0 · 31.1yes
25 Oct 2024risk-onrisk-on ·70.9 · 40.353.3 · 46.9yes
18 Oct 2024risk-on ·broad rally ·67.4 · 48.154.5 · 53.9no
11 Oct 2024broad rally ·broad rally62.7 · 54.960.0 · 56.6yes
4 Oct 2024broad rallybroad rally64.9 · 57.363.8 · 55.7yes
27 Sept 2024risk-onrisk-on ·71.8 · 43.663.9 · 48.1yes
20 Sept 2024risk-on ·defensive ·65.1 · 47.545.8 · 67.6no
13 Sept 2024risk-on ·broad rally63.0 · 49.458.1 · 56.4no
6 Sept 2024broad rally ·defensive ·51.7 · 56.849.5 · 62.3no
30 Aug 2024risk-onbroad rally ·68.2 · 42.559.6 · 51.9no
23 Aug 2024risk-onrisk-on ·74.4 · 34.765.4 · 48.8yes
16 Aug 2024risk-ondefensive ·70.0 · 32.647.0 · 51.6no
9 Aug 2024risk-ondefensive60.3 · 37.831.6 · 63.8no
2 Aug 2024risk-ondefensive59.8 · 41.536.1 · 64.6no
26 Jul 2024risk-onrisk-on77.6 · 25.068.2 · 42.6yes
19 Jul 2024risk-onbroad rally73.7 · 31.357.4 · 55.2no
12 Jul 2024risk-onrisk-on80.5 · 19.080.2 · 24.2yes
5 Jul 2024risk-onrisk-on79.2 · 17.076.9 · 20.8yes
28 Jun 2024risk-onrisk-on75.1 · 22.469.1 · 30.7yes
21 Jun 2024risk-onrisk-on74.8 · 19.773.1 · 20.1yes
14 Jun 2024risk-onrisk-on76.2 · 21.273.1 · 19.6yes
7 Jun 2024risk-onrisk-on69.5 · 27.264.1 · 30.9yes
31 May 2024risk-onrisk-on ·68.4 · 35.662.7 · 47.6yes
24 May 2024risk-onrisk-on74.2 · 26.368.5 · 33.3yes
17 May 2024risk-onbroad rally70.9 · 37.160.5 · 62.6no
10 May 2024risk-ondefensive ·69.0 · 34.848.2 · 56.3no
3 May 2024risk-onrisk-on ·73.5 · 26.453.6 · 39.4yes
26 Apr 2024risk-ondefensive ·68.4 · 32.135.6 · 53.0no
19 Apr 2024risk-ondefensive61.6 · 36.726.9 · 63.1no
12 Apr 2024risk-onde-risking ·74.3 · 26.145.1 · 50.0no
5 Apr 2024risk-ondefensive ·76.4 · 30.245.2 · 66.2no
29 Mar 2024risk-onbroad rally75.8 · 33.657.6 · 64.5no
22 Mar 2024risk-onbroad rally78.2 · 30.960.3 · 57.4no
15 Mar 2024risk-onbroad rally ·77.7 · 28.850.6 · 65.2no
8 Mar 2024risk-onrisk-on86.6 · 19.275.7 · 41.5yes
1 Mar 2024risk-onrisk-on ·88.0 · 18.275.4 · 46.9yes
23 Feb 2024risk-onbroad rally ·82.2 · 23.552.0 · 63.5no
16 Feb 2024risk-onbroad rally ·81.3 · 22.065.2 · 51.6no
9 Feb 2024risk-onrisk-on87.8 · 12.878.1 · 27.0yes
2 Feb 2024risk-onrisk-on87.2 · 14.178.8 · 24.8yes
26 Jan 2024risk-onrisk-on88.3 · 9.868.6 · 26.5yes
19 Jan 2024risk-onrisk-on89.7 · 7.873.6 · 21.3yes
12 Jan 2024risk-onrisk-on ·84.1 · 19.566.0 · 47.0yes
5 Jan 2024risk-onrisk-on ·84.8 · 21.671.0 · 46.2yes
29 Dec 2023risk-onrisk-on92.5 · 10.687.8 · 24.2yes
22 Dec 2023risk-onrisk-on93.2 · 7.290.7 · 16.9yes
15 Dec 2023risk-onrisk-on91.7 · 7.688.9 · 19.3yes
8 Dec 2023risk-onrisk-on86.5 · 10.884.5 · 22.6yes
1 Dec 2023risk-onrisk-on82.1 · 15.382.4 · 30.1yes
24 Nov 2023risk-onrisk-on86.0 · 9.787.8 · 17.2yes
17 Nov 2023risk-onrisk-on85.8 · 8.986.4 · 14.1yes
10 Nov 2023risk-onrisk-on83.4 · 8.669.9 · 22.0yes
3 Nov 2023risk-onrisk-on75.0 · 16.059.4 · 41.3yes
27 Oct 2023risk-onde-risking ·66.9 · 11.946.4 · 20.1no
20 Oct 2023risk-onrisk-on74.8 · 10.865.9 · 15.1yes
13 Oct 2023risk-onrisk-on71.2 · 17.056.2 · 20.0yes
6 Oct 2023risk-onrisk-on78.2 · 12.266.1 · 15.4yes
29 Sept 2023risk-onrisk-on ·73.6 · 18.554.0 · 21.8yes
22 Sept 2023risk-ondefensive ·65.1 · 29.435.4 · 50.0no
15 Sept 2023risk-onrisk-on ·69.4 · 27.051.3 · 45.3yes
8 Sept 2023risk-onrisk-on76.3 · 20.460.5 · 28.8yes
1 Sept 2023risk-onrisk-on80.9 · 18.568.5 · 26.5yes
25 Aug 2023risk-onde-risking ·72.8 · 22.649.9 · 36.5no
18 Aug 2023risk-onrisk-on71.3 · 20.955.2 · 30.1yes
11 Aug 2023risk-onrisk-on ·73.9 · 23.550.4 · 49.9yes
4 Aug 2023risk-onrisk-on83.1 · 15.486.4 · 17.4yes
28 Jul 2023risk-onrisk-on85.6 · 17.882.4 · 31.6yes
21 Jul 2023risk-onrisk-on79.6 · 24.674.1 · 42.4yes
14 Jul 2023risk-onrisk-on87.1 · 15.777.9 · 29.1yes
7 Jul 2023risk-onrisk-on86.4 · 14.179.0 · 25.4yes
30 Jun 2023risk-onrisk-on87.3 · 16.086.3 · 30.4yes
23 Jun 2023risk-onrisk-on84.1 · 16.988.1 · 13.7yes
16 Jun 2023risk-onrisk-on84.3 · 16.991.3 · 11.6yes
9 Jun 2023risk-onrisk-on81.8 · 12.792.5 · 6.9yes
2 Jun 2023risk-onrisk-on81.2 · 11.592.2 · 5.5yes
26 May 2023risk-onrisk-on83.9 · 9.493.2 · 4.1yes
19 May 2023risk-onrisk-on74.2 · 15.682.9 · 9.1yes
12 May 2023risk-onrisk-on62.2 · 23.563.0 · 20.1yes
5 May 2023de-risking ·risk-on ·45.8 · 28.353.1 · 30.6no
28 Apr 2023risk-on ·defensive53.6 · 31.842.0 · 56.8no
21 Apr 2023risk-on ·defensive52.6 · 27.639.3 · 61.5no
14 Apr 2023risk-onrisk-on ·58.7 · 25.661.2 · 49.6yes
7 Apr 2023risk-on ·risk-on53.4 · 26.858.7 · 29.9yes
31 Mar 2023risk-onrisk-on59.0 · 20.881.7 · 12.1yes
24 Mar 2023risk-onrisk-on57.0 · 19.877.2 · 8.2yes
17 Mar 2023risk-on ·risk-on53.5 · 21.565.3 · 8.5yes
10 Mar 2023risk-on ·risk-on53.0 · 24.658.8 · 14.2yes
3 Mar 2023risk-on ·risk-on52.0 · 29.961.3 · 22.4yes
24 Feb 2023risk-onrisk-on56.0 · 29.474.0 · 15.8yes
17 Feb 2023risk-onrisk-on55.2 · 30.385.3 · 12.3yes
10 Feb 2023risk-onrisk-on61.1 · 27.487.1 · 12.6yes
3 Feb 2023risk-onrisk-on64.6 · 23.891.3 · 8.6yes
27 Jan 2023risk-onrisk-on63.8 · 29.988.3 · 13.9yes
20 Jan 2023risk-onrisk-on57.7 · 35.779.2 · 33.4yes
13 Jan 2023de-risking ·risk-on ·48.3 · 46.459.4 · 49.4no
6 Jan 2023defensive ·defensive32.5 · 51.431.3 · 65.9yes
30 Dec 2022de-risking ·defensive35.5 · 45.121.3 · 59.6no
23 Dec 2022de-risking ·defensive33.3 · 49.920.9 · 65.6no
16 Dec 2022de-riskingde-risking41.7 · 33.436.1 · 41.7yes
9 Dec 2022de-riskingrisk-on42.8 · 33.966.0 · 29.2no
2 Dec 2022risk-on ·risk-on ·50.2 · 43.370.8 · 49.5yes
25 Nov 2022de-riskingbroad rally ·43.1 · 42.652.1 · 63.6no
18 Nov 2022de-risking ·broad rally49.0 · 34.755.6 · 66.0no
11 Nov 2022risk-on ·broad rally54.9 · 33.857.5 · 65.5no
4 Nov 2022de-riskingdefensive43.8 · 40.313.8 · 81.4no
28 Oct 2022de-risking ·defensive46.4 · 35.828.2 · 74.4no
21 Oct 2022risk-onde-risking ·57.8 · 22.146.5 · 39.7no
14 Oct 2022risk-on ·de-risking53.4 · 20.633.4 · 24.1no
7 Oct 2022risk-onrisk-on67.6 · 13.355.9 · 14.3yes
30 Sept 2022risk-onrisk-on ·63.8 · 13.752.1 · 12.1yes
23 Sept 2022risk-on ·de-risking51.9 · 20.631.9 · 19.5no
16 Sept 2022risk-onde-risking62.3 · 28.931.6 · 40.0no
9 Sept 2022risk-ondefensive63.5 · 27.631.9 · 57.6no
2 Sept 2022risk-ondefensive61.6 · 22.126.6 · 56.1no
26 Aug 2022risk-onde-risking ·69.9 · 22.449.1 · 48.6no
19 Aug 2022risk-ondefensive ·74.2 · 30.847.0 · 70.9no
12 Aug 2022risk-onbroad rally78.6 · 28.484.2 · 57.7no
5 Aug 2022risk-onrisk-on72.9 · 20.079.7 · 26.1yes
29 Jul 2022risk-onrisk-on68.7 · 22.874.7 · 37.2yes
22 Jul 2022risk-onrisk-on66.7 · 17.572.3 · 27.3yes
15 Jul 2022de-risking ·de-risking49.0 · 24.143.4 · 40.1yes
8 Jul 2022de-riskingde-risking43.5 · 24.335.5 · 18.0yes
1 Jul 2022de-riskingde-risking26.4 · 38.913.9 · 42.6yes
24 Jun 2022de-riskingrisk-on39.9 · 31.859.2 · 20.4no
17 Jun 2022risk-on ·de-risking ·51.7 · 19.345.2 · 17.8no
10 Jun 2022risk-onrisk-on ·61.0 · 26.953.8 · 40.3yes
3 Jun 2022risk-ondefensive ·63.6 · 24.833.0 · 50.5no
27 May 2022risk-ondefensive ·61.1 · 32.337.9 · 53.7no
20 May 2022risk-onde-risking57.9 · 22.932.3 · 26.3no
13 May 2022risk-onde-risking57.6 · 25.134.7 · 28.3no
6 May 2022risk-onde-risking62.4 · 22.336.3 · 34.0no
29 Apr 2022risk-onde-risking ·60.4 · 25.724.0 · 45.6no
22 Apr 2022risk-ondefensive61.2 · 32.922.7 · 71.5no
15 Apr 2022risk-ondefensive60.7 · 31.921.1 · 74.9no
8 Apr 2022risk-ondefensive60.8 · 33.124.5 · 74.5no
1 Apr 2022risk-ondefensive72.2 · 25.244.6 · 59.3no
25 Mar 2022risk-onbroad rally ·76.3 · 24.053.0 · 62.0no
18 Mar 2022risk-ondefensive ·75.4 · 22.249.4 · 51.8no
11 Mar 2022risk-onde-risking67.5 · 19.637.6 · 34.6no
4 Mar 2022risk-onde-risking65.8 · 27.640.2 · 42.0no
25 Feb 2022risk-onrisk-on ·67.3 · 23.351.2 · 40.1yes
18 Feb 2022risk-onrisk-on70.3 · 18.057.6 · 24.3yes
11 Feb 2022risk-onde-risking ·74.7 · 12.948.8 · 23.2no
4 Feb 2022risk-onrisk-on77.4 · 12.158.7 · 20.2yes
28 Jan 2022risk-onrisk-on ·76.0 · 12.953.4 · 20.5yes
21 Jan 2022risk-onrisk-on ·71.1 · 17.850.5 · 29.3yes
14 Jan 2022risk-onrisk-on84.5 · 15.579.7 · 21.9yes
7 Jan 2022risk-ondefensive74.5 · 30.344.9 · 63.0no
31 Dec 2021risk-ondefensive67.7 · 40.231.2 · 83.6no
24 Dec 2021risk-ondefensive74.9 · 28.641.8 · 59.9no
17 Dec 2021risk-ondefensive76.9 · 24.140.5 · 56.5no
10 Dec 2021risk-onrisk-on85.0 · 15.165.7 · 31.5yes
3 Dec 2021risk-onrisk-on87.7 · 8.167.0 · 18.0yes
26 Nov 2021risk-onrisk-on88.4 · 10.075.0 · 23.6yes
19 Nov 2021risk-onrisk-on92.2 · 7.978.8 · 24.2yes
12 Nov 2021risk-onrisk-on94.1 · 7.982.2 · 30.0yes
5 Nov 2021risk-onrisk-on95.4 · 8.584.4 · 33.2yes
29 Oct 2021risk-onrisk-on93.2 · 11.978.6 · 44.3yes
22 Oct 2021risk-onrisk-on93.0 · 8.786.4 · 19.6yes
15 Oct 2021risk-onrisk-on95.1 · 6.090.7 · 10.1yes
8 Oct 2021risk-onrisk-on90.5 · 7.888.2 · 9.4yes
1 Oct 2021risk-onrisk-on86.9 · 9.783.9 · 11.5yes
24 Sept 2021risk-onrisk-on83.6 · 13.677.9 · 17.9yes
17 Sept 2021risk-onrisk-on82.7 · 17.279.7 · 19.9yes
10 Sept 2021risk-onrisk-on76.2 · 19.656.9 · 32.6yes
3 Sept 2021risk-ondefensive ·68.4 · 34.145.2 · 63.7no
27 Aug 2021risk-onbroad rally ·72.8 · 31.253.5 · 52.3no
20 Aug 2021risk-on ·defensive60.9 · 49.623.2 · 81.6no
13 Aug 2021risk-ondefensive66.9 · 42.640.6 · 69.7no
6 Aug 2021risk-ondefensive69.1 · 34.839.0 · 67.9no
30 Jul 2021risk-ondefensive70.9 · 31.431.7 · 69.3no
23 Jul 2021risk-ondefensive70.8 · 39.831.2 · 72.5no
16 Jul 2021risk-ondefensive69.1 · 34.228.7 · 69.1no
9 Jul 2021risk-onrisk-on ·82.9 · 19.456.6 · 45.4yes
2 Jul 2021risk-onrisk-on85.2 · 17.168.1 · 32.5yes
25 Jun 2021risk-onrisk-on88.2 · 16.482.8 · 21.5yes
18 Jun 2021risk-onrisk-on74.6 · 35.460.2 · 40.8yes
11 Jun 2021risk-onrisk-on ·78.2 · 31.969.6 · 45.3yes
4 Jun 2021risk-onrisk-on78.3 · 31.570.2 · 38.6yes
28 May 2021risk-onrisk-on79.2 · 29.266.0 · 40.5yes
21 May 2021risk-onrisk-on83.5 · 27.178.3 · 31.2yes
14 May 2021risk-onrisk-on93.1 · 10.382.2 · 21.5yes
7 May 2021risk-onrisk-on94.0 · 9.981.0 · 30.7yes
30 Apr 2021risk-onbroad rally ·89.8 · 18.558.0 · 55.0no
23 Apr 2021risk-ondefensive ·85.6 · 22.848.4 · 68.6no
16 Apr 2021risk-ondefensive81.2 · 30.135.9 · 78.2no
9 Apr 2021risk-ondefensive87.7 · 15.843.0 · 63.9no
2 Apr 2021risk-onbroad rally ·91.8 · 10.654.6 · 57.7no
26 Mar 2021risk-onbroad rally ·92.2 · 10.359.6 · 54.2no
19 Mar 2021risk-onrisk-on98.2 · 1.994.1 · 5.9yes
12 Mar 2021risk-onrisk-on99.3 · 0.997.9 · 1.9yes
5 Mar 2021risk-onrisk-on99.8 · 0.399.4 · 0.4yes
26 Feb 2021risk-onrisk-on98.3 · 2.398.6 · 1.4yes
19 Feb 2021risk-onrisk-on97.2 · 3.095.8 · 3.6yes
12 Feb 2021risk-onrisk-on96.2 · 4.690.2 · 13.6yes
5 Feb 2021risk-onrisk-on95.3 · 9.988.8 · 16.7yes
29 Jan 2021risk-onrisk-on95.7 · 6.394.1 · 6.1yes
22 Jan 2021risk-onrisk-on96.3 · 6.697.0 · 3.9yes
15 Jan 2021risk-onrisk-on94.3 · 8.396.4 · 3.7yes
8 Jan 2021risk-onrisk-on92.5 · 13.194.0 · 8.3yes
1 Jan 2021risk-onrisk-on87.2 · 15.091.2 · 9.4yes
25 Dec 2020risk-onrisk-on85.5 · 17.868.8 · 34.2yes
18 Dec 2020risk-onrisk-on86.8 · 15.594.3 · 6.4yes
11 Dec 2020risk-onrisk-on88.4 · 14.197.4 · 3.4yes
4 Dec 2020risk-onrisk-on94.2 · 9.898.6 · 2.6yes
27 Nov 2020risk-onrisk-on90.3 · 14.696.8 · 5.0yes
20 Nov 2020risk-onrisk-on79.6 · 25.988.5 · 15.7yes
13 Nov 2020risk-on ·risk-on57.8 · 45.571.0 · 33.6yes
6 Nov 2020defensive ·defensive41.1 · 53.714.6 · 87.1yes
30 Oct 2020risk-onrisk-on57.8 · 35.965.8 · 31.4yes
23 Oct 2020risk-onbroad rally62.2 · 44.859.7 · 68.3no
16 Oct 2020risk-on ·defensive54.6 · 49.619.1 · 88.2no
9 Oct 2020risk-on ·defensive62.0 · 47.443.1 · 79.4no
2 Oct 2020risk-on ·defensive52.6 · 47.616.6 · 80.5no
25 Sept 2020risk-on ·defensive52.0 · 44.512.9 · 87.0no
18 Sept 2020risk-on ·defensive52.5 · 46.217.0 · 84.8no
11 Sept 2020de-risking ·defensive46.9 · 48.79.5 · 90.5no
4 Sept 2020risk-on ·defensive53.1 · 47.320.0 · 85.5no
28 Aug 2020risk-ondefensive64.8 · 43.741.5 · 81.5no
21 Aug 2020risk-on ·defensive62.0 · 45.526.5 · 87.1no
14 Aug 2020risk-onrisk-on90.0 · 29.084.5 · 44.8yes
7 Aug 2020risk-onrisk-on ·70.1 · 39.773.2 · 46.6yes
31 Jul 2020risk-on ·defensive59.3 · 47.334.6 · 85.2no
24 Jul 2020risk-onbroad rally ·69.6 · 30.150.8 · 68.0no
17 Jul 2020risk-ondefensive72.2 · 23.337.7 · 66.4no
10 Jul 2020risk-ondefensive71.2 · 29.333.0 · 74.3no
3 Jul 2020risk-ondefensive74.2 · 19.942.7 · 61.0no
26 Jun 2020risk-onrisk-on77.3 · 9.762.4 · 27.1yes
19 Jun 2020risk-onrisk-on85.1 · 11.190.4 · 20.9yes
12 Jun 2020risk-onrisk-on84.2 · 6.496.8 · 2.5yes
5 Jun 2020risk-onrisk-on87.7 · 9.697.6 · 3.2yes
29 May 2020risk-ondefensive58.1 · 27.634.2 · 65.9no
22 May 2020risk-onrisk-on75.4 · 6.087.2 · 3.6yes
15 May 2020risk-onrisk-on59.6 · 5.679.9 · 2.0yes
8 May 2020risk-onrisk-on57.6 · 9.579.5 · 3.3yes
1 May 2020de-risking ·risk-on49.8 · 9.777.0 · 3.1no
24 Apr 2020de-riskingrisk-on ·40.4 · 17.766.9 · 45.5no
17 Apr 2020de-riskingrisk-on32.0 · 20.266.9 · 15.6no
10 Apr 2020de-riskingrisk-on29.4 · 23.759.5 · 17.6no
3 Apr 2020de-riskingde-risking31.6 · 5.128.7 · 2.4yes
27 Mar 2020de-riskingde-risking28.1 · 9.312.6 · 6.9yes
20 Mar 2020de-riskingde-risking33.0 · 2.617.3 · 0.9yes
13 Mar 2020de-riskingde-risking26.5 · 19.29.8 · 13.8yes
6 Mar 2020defensive ·defensive25.0 · 54.410.1 · 60.0yes
28 Feb 2020de-riskingde-risking40.5 · 24.130.4 · 15.7yes
21 Feb 2020defensivedefensive40.1 · 71.026.1 · 80.9yes
14 Feb 2020defensivedefensive40.8 · 72.823.0 · 85.4yes
7 Feb 2020defensive ·defensive46.7 · 59.523.5 · 81.6yes
31 Jan 2020defensive ·defensive47.5 · 52.926.4 · 69.1yes
24 Jan 2020broad rallydefensive59.2 · 55.038.1 · 67.9no
17 Jan 2020risk-on ·broad rally ·69.1 · 49.253.1 · 62.8no
10 Jan 2020risk-onrisk-on77.3 · 35.982.5 · 25.6yes
3 Jan 2020risk-onrisk-on87.5 · 33.992.1 · 17.1yes
27 Dec 2019risk-onrisk-on80.8 · 43.084.8 · 24.0yes
20 Dec 2019risk-onrisk-on84.6 · 34.785.7 · 24.6yes
13 Dec 2019risk-onrisk-on75.3 · 27.765.2 · 35.5yes
6 Dec 2019risk-onrisk-on ·66.3 · 34.051.9 · 49.5yes
29 Nov 2019risk-onrisk-on65.2 · 34.670.6 · 31.2yes
22 Nov 2019risk-onrisk-on70.8 · 31.979.9 · 22.1yes
15 Nov 2019risk-onrisk-on73.6 · 29.983.6 · 22.9yes
8 Nov 2019risk-onrisk-on70.4 · 28.388.2 · 16.7yes
1 Nov 2019broad rally ·risk-on58.1 · 51.974.2 · 43.3no
25 Oct 2019broad rally ·risk-on ·50.9 · 53.056.8 · 45.8no
18 Oct 2019defensivedefensive24.4 · 76.124.4 · 72.7yes
11 Oct 2019defensivedefensive30.6 · 62.628.9 · 58.5yes
4 Oct 2019defensivedefensive19.8 · 73.729.7 · 63.2yes
27 Sept 2019defensivedefensive ·23.4 · 77.949.7 · 64.1yes
20 Sept 2019defensivebroad rally ·32.4 · 62.956.3 · 50.2no
13 Sept 2019de-risking ·risk-on46.6 · 48.175.3 · 42.0no
6 Sept 2019defensivedefensive ·32.4 · 61.849.0 · 59.1yes
30 Aug 2019defensivedefensive25.7 · 68.326.2 · 71.4yes
23 Aug 2019defensivedefensive22.4 · 63.616.8 · 71.7yes
16 Aug 2019defensivedefensive21.8 · 65.718.2 · 67.0yes
9 Aug 2019defensivedefensive20.2 · 67.519.6 · 66.5yes
2 Aug 2019defensivede-risking ·25.3 · 58.730.1 · 49.3no
26 Jul 2019defensivedefensive ·33.9 · 56.346.6 · 60.1yes
19 Jul 2019defensiverisk-on31.2 · 60.955.2 · 43.3no
12 Jul 2019defensiverisk-on36.9 · 57.863.2 · 42.7no
5 Jul 2019defensivebroad rally ·35.3 · 62.959.8 · 52.8no
28 Jun 2019defensive ·broad rally ·39.9 · 54.963.6 · 50.6no
21 Jun 2019defensivedefensive25.2 · 76.838.7 · 71.4yes
14 Jun 2019defensivedefensive15.0 · 81.416.4 · 83.2yes
7 Jun 2019defensivedefensive18.0 · 78.318.2 · 81.1yes
31 May 2019defensivedefensive16.1 · 69.514.3 · 65.7yes
24 May 2019defensivedefensive15.2 · 79.612.9 · 82.1yes
17 May 2019defensivedefensive23.1 · 71.620.0 · 72.5yes
10 May 2019defensivede-risking ·40.0 · 57.144.6 · 46.5no
3 May 2019broad rally ·risk-on ·52.5 · 52.561.9 · 46.1no
26 Apr 2019broad rally ·risk-on52.2 · 55.375.6 · 38.5no
19 Apr 2019risk-on ·risk-on54.6 · 46.678.4 · 32.6yes
12 Apr 2019defensive ·risk-on46.4 · 66.575.5 · 42.4no
5 Apr 2019defensive ·broad rally ·49.7 · 64.576.5 · 50.5no
29 Mar 2019defensivebroad rally43.6 · 69.956.6 · 58.2no
22 Mar 2019defensiverisk-on ·37.9 · 68.756.4 · 46.4no
15 Mar 2019defensiverisk-on ·38.9 · 64.255.8 · 49.6no
8 Mar 2019defensivedefensive28.1 · 66.535.9 · 60.4yes
1 Mar 2019defensivebroad rally38.6 · 63.757.9 · 57.4no
22 Feb 2019defensivebroad rally41.1 · 63.169.1 · 58.6no
15 Feb 2019defensivebroad rally39.9 · 59.762.3 · 61.7no
8 Feb 2019defensivedefensive26.3 · 64.129.5 · 74.4yes
1 Feb 2019de-riskingbroad rally41.5 · 40.958.8 · 65.9no
25 Jan 2019de-riskingbroad rally ·40.5 · 36.071.2 · 51.5no
18 Jan 2019de-riskingrisk-on34.4 · 38.962.8 · 38.2no
11 Jan 2019de-riskingrisk-on35.6 · 29.661.6 · 20.9no
4 Jan 2019de-riskingde-risking28.0 · 30.123.8 · 26.3yes
28 Dec 2018de-riskingde-risking26.4 · 29.424.5 · 26.5yes
21 Dec 2018de-riskingde-risking21.8 · 29.215.9 · 28.1yes
14 Dec 2018de-riskingdefensive ·24.9 · 43.615.9 · 52.3no
7 Dec 2018de-risking ·defensive26.5 · 49.012.7 · 74.1no
30 Nov 2018de-risking ·defensive34.9 · 45.527.0 · 70.8no
23 Nov 2018de-riskingde-risking ·37.5 · 36.738.3 · 49.8yes
16 Nov 2018de-riskingdefensive38.7 · 37.019.1 · 67.6no
9 Nov 2018de-riskingdefensive38.1 · 43.018.6 · 75.3no
2 Nov 2018de-riskingdefensive39.6 · 32.917.7 · 60.6no
26 Oct 2018de-riskingdefensive35.0 · 37.312.6 · 66.0no
19 Oct 2018de-riskingdefensive41.2 · 36.316.9 · 55.5no
12 Oct 2018risk-onde-risking ·57.6 · 22.448.2 · 20.4no
5 Oct 2018risk-onrisk-on64.2 · 24.964.2 · 22.7yes
28 Sept 2018risk-onrisk-on71.0 · 25.961.3 · 29.9yes
21 Sept 2018risk-onrisk-on69.3 · 30.162.6 · 36.2yes
14 Sept 2018risk-onrisk-on ·66.3 · 35.859.6 · 46.6yes
7 Sept 2018risk-ondefensive64.7 · 34.136.0 · 60.0no
31 Aug 2018risk-onbroad rally ·71.3 · 30.953.3 · 50.9no
24 Aug 2018risk-ondefensive69.8 · 32.844.3 · 59.8no
17 Aug 2018risk-ondefensive62.3 · 41.226.7 · 76.1no
10 Aug 2018risk-onrisk-on ·72.1 · 28.553.2 · 46.0yes
3 Aug 2018risk-onrisk-on ·74.2 · 29.262.1 · 45.9yes
27 Jul 2018risk-onrisk-on79.2 · 22.171.5 · 36.2yes
20 Jul 2018risk-ondefensive71.3 · 29.640.4 · 62.0no
13 Jul 2018risk-ondefensive65.8 · 36.726.6 · 75.7no
6 Jul 2018risk-ondefensive70.0 · 34.129.5 · 76.1no
29 Jun 2018risk-ondefensive70.7 · 31.434.1 · 69.3no
22 Jun 2018risk-ondefensive74.3 · 30.639.4 · 68.2no
15 Jun 2018risk-onrisk-on ·78.5 · 22.057.1 · 49.9yes
8 Jun 2018risk-onrisk-on83.3 · 17.675.3 · 32.1yes
1 Jun 2018risk-onrisk-on84.4 · 16.781.6 · 24.0yes
25 May 2018risk-onrisk-on83.0 · 14.878.3 · 21.7yes
18 May 2018risk-onrisk-on87.6 · 10.883.0 · 16.0yes
11 May 2018risk-onrisk-on83.5 · 20.069.7 · 39.7yes
4 May 2018risk-onrisk-on77.9 · 16.362.7 · 29.0yes
27 Apr 2018risk-onrisk-on73.8 · 16.961.2 · 36.4yes
20 Apr 2018risk-onde-risking72.8 · 16.544.1 · 37.6no
13 Apr 2018risk-onde-risking ·72.1 · 16.747.7 · 33.3no
6 Apr 2018risk-ondefensive65.8 · 20.830.8 · 56.0no
30 Mar 2018risk-onde-risking ·72.6 · 17.844.9 · 47.3no
23 Mar 2018risk-onrisk-on78.0 · 11.155.4 · 23.1yes
16 Mar 2018risk-onrisk-on78.8 · 15.067.0 · 31.0yes
9 Mar 2018risk-onrisk-on83.8 · 13.781.6 · 23.0yes
2 Mar 2018risk-onrisk-on78.2 · 9.074.5 · 8.5yes
23 Feb 2018risk-onrisk-on75.9 · 13.965.1 · 14.6yes
16 Feb 2018risk-onrisk-on68.1 · 21.959.3 · 21.5yes
9 Feb 2018risk-onrisk-on65.9 · 18.762.0 · 14.3yes
2 Feb 2018risk-onrisk-on72.5 · 28.973.4 · 20.5yes
26 Jan 2018risk-onrisk-on72.9 · 43.280.7 · 33.8yes
19 Jan 2018risk-onrisk-on78.5 · 32.787.3 · 18.8yes
12 Jan 2018risk-onrisk-on79.9 · 30.288.6 · 17.2yes
5 Jan 2018risk-onrisk-on70.0 · 38.480.3 · 26.7yes
29 Dec 2017risk-on ·risk-on63.1 · 48.370.0 · 33.7yes
22 Dec 2017risk-on ·risk-on65.3 · 46.879.5 · 28.3yes
15 Dec 2017broad rally ·risk-on54.6 · 58.472.0 · 41.0no
8 Dec 2017broad rally ·risk-on57.2 · 52.675.9 · 31.7no
1 Dec 2017broad rallyrisk-on ·56.9 · 56.965.7 · 47.4no
24 Nov 2017broad rallyrisk-on ·56.6 · 55.764.5 · 45.2no
17 Nov 2017broad rally ·risk-on60.2 · 51.966.7 · 35.7no
10 Nov 2017risk-on ·risk-on62.5 · 49.665.0 · 36.9yes
3 Nov 2017risk-onrisk-on75.6 · 35.970.9 · 28.5yes
27 Oct 2017risk-onrisk-on70.1 · 42.271.1 · 33.4yes
20 Oct 2017risk-onrisk-on68.9 · 42.775.2 · 32.4yes
13 Oct 2017risk-onrisk-on67.6 · 41.477.7 · 24.9yes
6 Oct 2017risk-onrisk-on67.6 · 44.985.8 · 17.5yes
29 Sept 2017risk-onrisk-on73.7 · 32.686.7 · 15.5yes
22 Sept 2017risk-onrisk-on70.6 · 29.584.7 · 16.4yes
15 Sept 2017broad rally ·risk-on57.1 · 50.365.6 · 36.6no
8 Sept 2017defensivedefensive41.2 · 62.030.3 · 68.2yes
1 Sept 2017risk-on ·risk-on ·53.0 · 47.151.5 · 42.2yes
25 Aug 2017defensive ·defensive47.0 · 50.831.3 · 60.8yes
18 Aug 2017defensive ·defensive44.3 · 52.528.3 · 61.7yes
11 Aug 2017risk-on ·defensive50.1 · 43.730.7 · 63.8no
4 Aug 2017risk-ondefensive ·60.5 · 37.748.0 · 59.3no
28 Jul 2017risk-onrisk-on70.6 · 26.162.0 · 36.0yes
21 Jul 2017risk-onrisk-on68.2 · 28.466.3 · 33.5yes
14 Jul 2017risk-onrisk-on71.4 · 24.867.3 · 29.9yes
7 Jul 2017risk-onrisk-on72.3 · 24.369.3 · 24.9yes
30 Jun 2017risk-onrisk-on68.1 · 29.562.4 · 34.2yes
23 Jun 2017defensive ·defensive48.5 · 61.136.4 · 66.9yes
16 Jun 2017broad rally ·defensive52.0 · 58.639.6 · 71.3no
9 Jun 2017broad rally ·defensive50.7 · 59.235.1 · 68.9no
2 Jun 2017defensivedefensive43.9 · 66.026.5 · 76.4yes
26 May 2017broad rally ·defensive51.0 · 55.737.5 · 62.8no
19 May 2017risk-onrisk-on ·61.2 · 41.554.4 · 41.5yes
12 May 2017risk-onrisk-on67.0 · 33.155.6 · 41.9yes
5 May 2017risk-onbroad rally ·64.5 · 39.952.5 · 51.9no
28 Apr 2017broad rally ·defensive51.4 · 51.640.2 · 60.7no
21 Apr 2017risk-on ·defensive ·54.1 · 48.949.2 · 52.9no
14 Apr 2017risk-ondefensive60.8 · 36.032.3 · 60.5no
7 Apr 2017risk-ondefensive67.7 · 33.541.3 · 60.1no
31 Mar 2017risk-onrisk-on ·70.4 · 29.751.6 · 44.7yes
24 Mar 2017risk-ondefensive63.8 · 36.231.2 · 65.8no
17 Mar 2017risk-ondefensive65.7 · 35.732.0 · 69.6no
10 Mar 2017risk-ondefensive67.5 · 34.141.8 · 64.4no
3 Mar 2017risk-ondefensive67.2 · 37.234.6 · 74.8no
24 Feb 2017risk-ondefensive69.3 · 34.029.9 · 74.6no
17 Feb 2017risk-onbroad rally ·78.6 · 27.151.8 · 62.5no
10 Feb 2017risk-onbroad rally ·86.7 · 18.356.0 · 52.4no
3 Feb 2017risk-onrisk-on89.4 · 14.363.5 · 39.9yes
27 Jan 2017risk-onrisk-on91.1 · 11.965.7 · 36.8yes
20 Jan 2017risk-onrisk-on89.8 · 12.162.3 · 37.8yes
13 Jan 2017risk-onrisk-on89.2 · 11.957.7 · 38.9yes
6 Jan 2017risk-onbroad rally ·87.3 · 15.351.6 · 54.2no
30 Dec 2016risk-onrisk-on ·89.3 · 12.760.3 · 46.4yes
23 Dec 2016risk-onrisk-on94.9 · 7.076.3 · 28.8yes
16 Dec 2016risk-onrisk-on96.3 · 6.685.3 · 24.1yes
9 Dec 2016risk-onrisk-on97.5 · 5.194.5 · 11.6yes
2 Dec 2016risk-onrisk-on97.0 · 2.097.6 · 2.6yes
25 Nov 2016risk-onrisk-on86.8 · 6.992.4 · 7.9yes
18 Nov 2016risk-onrisk-on86.3 · 5.291.3 · 7.8yes
11 Nov 2016risk-onrisk-on81.7 · 7.187.1 · 10.1yes
4 Nov 2016risk-onde-risking ·64.7 · 21.548.6 · 31.2no
28 Oct 2016risk-onrisk-on78.9 · 6.074.1 · 14.5yes
21 Oct 2016risk-onrisk-on85.2 · 2.486.3 · 6.9yes
14 Oct 2016risk-onrisk-on77.3 · 14.482.4 · 11.6yes
7 Oct 2016risk-onrisk-on81.0 · 10.782.7 · 10.3yes
30 Sept 2016risk-onrisk-on90.8 · 5.882.2 · 13.1yes
23 Sept 2016risk-onrisk-on74.0 · 36.462.1 · 35.3yes
16 Sept 2016risk-onrisk-on73.0 · 28.773.3 · 18.8yes
9 Sept 2016risk-onrisk-on76.3 · 28.289.1 · 9.0yes
2 Sept 2016risk-on ·risk-on66.1 · 47.181.8 · 21.0yes
26 Aug 2016risk-onrisk-on75.4 · 31.887.3 · 13.5yes
19 Aug 2016risk-onrisk-on78.6 · 20.287.7 · 17.3yes
12 Aug 2016risk-onrisk-on66.4 · 41.978.0 · 26.9yes
5 Aug 2016risk-on ·risk-on62.3 · 48.582.1 · 28.6yes
29 Jul 2016defensivedefensive ·37.3 · 69.748.0 · 73.8yes
22 Jul 2016defensivedefensive26.9 · 77.826.9 · 84.1yes
15 Jul 2016defensivedefensive ·38.1 · 61.746.8 · 65.0yes
8 Jul 2016defensivedefensive30.6 · 72.731.7 · 73.8yes
1 Jul 2016defensivedefensive31.8 · 64.740.9 · 59.4yes
24 Jun 2016defensivedefensive ·30.2 · 62.841.0 · 53.9yes
17 Jun 2016defensivebroad rally ·35.7 · 57.750.0 · 56.1no
10 Jun 2016defensiverisk-on37.7 · 58.471.3 · 30.3no
3 Jun 2016defensive ·risk-on42.6 · 50.980.4 · 18.3no
27 May 2016de-risking ·risk-on44.1 · 48.364.8 · 32.8no
20 May 2016de-risking ·risk-on43.6 · 47.164.0 · 26.7no
13 May 2016defensive ·risk-on36.7 · 53.161.3 · 27.7no
6 May 2016defensive ·risk-on40.5 · 54.683.6 · 18.0no
29 Apr 2016de-risking ·risk-on45.7 · 48.388.0 · 9.8no
22 Apr 2016risk-on ·risk-on51.3 · 44.791.4 · 8.8yes
15 Apr 2016defensiverisk-on41.8 · 55.580.5 · 25.9no
8 Apr 2016defensiverisk-on33.5 · 62.772.6 · 33.7no
1 Apr 2016defensivebroad rally22.1 · 75.355.6 · 60.5no
25 Mar 2016defensiverisk-on26.4 · 67.269.1 · 32.1no
18 Mar 2016defensiverisk-on28.6 · 66.070.9 · 30.7no
11 Mar 2016defensiverisk-on23.7 · 72.365.9 · 39.3no
4 Mar 2016defensivebroad rally19.3 · 75.255.7 · 60.7no
26 Feb 2016defensivedefensive9.3 · 82.215.9 · 90.6yes
19 Feb 2016defensivedefensive9.8 · 80.623.2 · 83.8yes
12 Feb 2016defensivedefensive10.5 · 70.417.7 · 77.6yes
5 Feb 2016defensivedefensive11.8 · 65.612.3 · 80.2yes
29 Jan 2016de-riskingdefensive26.3 · 44.717.2 · 66.7no
22 Jan 2016de-riskingdefensive22.0 · 43.314.2 · 73.0no
15 Jan 2016de-risking ·defensive21.2 · 48.64.9 · 86.0no
8 Jan 2016defensive ·defensive19.5 · 52.27.3 · 82.4yes
1 Jan 2016defensivedefensive27.1 · 60.813.8 · 84.6yes
25 Dec 2015defensivedefensive31.2 · 57.315.8 · 84.2yes
About this exhibit
A weekly table of the former state name (risk-on, broad rally, defensive, de-risking), the two composite scores under year- and month-weighted weights, and whether the two agree.
The retired gauge, week by week
The two composite scores and the state name under both sets of weights, from 25 Dec 2015; · marks a week within 5 points of a line.

Until 11 Oct 2026 this page ranked the funds' slope and intercept at four horizons, weighted the ranks into two composite scores and named a state from them: risk-on, broad rally, defensive, de-risking. The scores rested on the ordinary least-squares line, whose error the funds' shared days make about two and a half times too narrow, and named the state from one family alone; both retire, and the table reproduces them by the same rules.

The September report's week, 11 Sept 2026: year-weighted risk-on (59.1 and 38.3; printed risk-on, 59.1 and 38.3); month-weighted defensive (31.9 and 51.2; printed defensive, 31.9 and 51.2).

Source: Yahoo Finance; U.S. Treasury; Federal Reserve Board (H.10); Macro Review calculationscross.sentiment-9

Exhibit 20 · The September report's beta line, as printed and on closing prices

Intercept, slope and median slope (pp) and VTI's year-to-date return, printed and on closing prices; betas to 18 Sep 2026, returns to 21 Sep 2026
MeasureAs printedOn closing pricesDifference
Interceptpp2.3832.591+0.208
Slopepp per unit of beta13.73913.898+0.159
Median (50th-percentile) slopepp per unit of beta9.0727.885−1.187
VTI, year to date%14.1814.35+0.17
Betas to · returns to18 Sept 2026 · 21 Sept 202618 Sept 2026 · 21 Sept 2026–

The figures printed in September for the factors page's year-to-date line — intercept 2.383pp, slope 13.739pp per unit of beta, median slope 9.072pp, VTI 14.18% since the end of 2025 — set 156-week weekly betas to 18 Sep 2026 against returns to 21 Sep 2026, and their return leg ends on a 21 Sep 2026 price taken before that session closed. No line drawn on closing prices can reproduce them, so they are shown as printed, with their dates, beside the same construction on the closing prices of the same two dates. On closing prices: intercept 2.591pp, slope 13.898pp, median slope 7.885pp, VTI 14.35%.

Inputs, as the September figures were made: the 18 sector and industry ETFs of this page; 156-week weekly log-return betas against VTI to the last complete week; year-to-date total returns on adjusted closes from the last close of 2025; the 2026 average 3-month Treasury par yield accrued to date as the risk-free rate; OLS and statsmodels QuantReg at the 10th, 50th and 90th percentiles.

Fund, on closing pricesBetaYear to date, %
TechnologyXLK1.451+35.83
Communication servicesXLC0.883−1.63
Consumer discretionaryXLY1.148−5.43
Consumer staplesXLP0.249+7.49
Health careXLV0.450+10.53
FinancialsXLF0.861+3.31
IndustrialsXLI0.907+10.45
EnergyXLE0.335+42.47
MaterialsXLB0.758+11.02
UtilitiesXLU0.286−2.77
Real estateXLRE0.607+8.09
SemiconductorsSMH1.859+65.50
Regional banksKRE1.058+13.02
HomebuildersXHB1.046−4.88
Oil & gas E&PXOP0.458+48.22
TransportsIYT1.022+8.00
REITsVNQ0.627+8.09
RetailXRT0.974−2.52
About this exhibit
A table of the September report's line figures (intercept, slope, median slope, VTI year to date) as printed beside the same construction on closing prices and the difference, with the funds' betas and year-to-date returns.
The September figures beside the same line on closing prices
The year-to-date line through the 18 funds: betas to 18 Sept 2026, returns to 21 Sept 2026. The figures as printed on 22 Sept 2026, beside the same construction on closing prices to the same two dates, and the difference.
Source: Macro Review, September 2026 (the figures as printed); Yahoo Finance; U.S. Treasury; Macro Review calculationscross.sml-13

Appendix

How each family is measured, the percentile bases, the count rule, the band, and every source with its date208 checks, all pass
About this page

Risk sentiment shows in several places at once, so no single number names it. The page starts from pricing, what a unit of beta paid across assets, reads it at four tenors and together, asks who carried the move, and checks it against five other families. Each block reads against its own history; the page counts the families and names no state.

The line, the CAPM line and the gap

Ri−rf=a+γ βi+ei,the CAPM line: rf+β (Rm−rf)R_i - r_f = a + \gamma\,\beta_i + e_i, \qquad \text{the CAPM line: } r_f + \beta\,(R_m - r_f)
αi=Ri−rf−βi (Rm−rf),gap=γ−(Rm−rf)=∑i(βi−βˉ) αi∑i(βi−βˉ)2\alpha_i = R_i - r_f - \beta_i\,(R_m - r_f), \qquad \text{gap} = \gamma - (R_m - r_f) = \frac{\sum_i (\beta_i - \bar\beta)\,\alpha_i}{\sum_i (\beta_i - \bar\beta)^2}
SymbolMeaningFromAssumed or estimated
RiR_iasset i's simple return over the tenorYahoo Finance; State Street's SPY holdings for the members–
βi\beta_iits 252-session beta to the US market, known when the tenor openedthe line's own estimate–
Rm−rfR_m - r_fthe market's return over cash over the tenor: the CAPM line's slopeVTI; the 3-month Treasury par yield–
αi\alpha_iwhat the asset earned beyond the CAPM line at its betathe line's own estimate–

The gap is the slope of alpha on beta, an identity checked every Friday. Positive: high-beta assets beat what the market's own move implies.

The reward for beta, the four tenors together

vi(h)=∑t=1hwt(h) xit,wt(h)∝2−(h−t)/(h/3),∑twt(h)=hv_i^{(h)} = \sum_{t=1}^{h} w_t^{(h)}\, x_{it}, \qquad w_t^{(h)} \propto 2^{-(h-t)/(h/3)}, \quad \sum_t w_t^{(h)} = h
vi(h)=a(h)+γ(h) βi+ei,gap(h)=γ(h)−∑twt(h)xmtv_i^{(h)} = a^{(h)} + \gamma^{(h)}\,\beta_i + e_i, \qquad \text{gap}^{(h)} = \gamma^{(h)} - \sum_t w_t^{(h)} x_{mt}
γˉ=∑hωh 252h γ(h),ωh∝h\bar\gamma = \sum_h \omega_h\, \tfrac{252}{h}\, \gamma^{(h)}, \qquad \omega_h \propto \sqrt{h}
SymbolMeaningFromAssumed or estimated
xitx_{it}asset i's return over cash on session t of the tenorYahoo Finance; the 3-month Treasury par yield–
wt(h)w_t^{(h)}the session's weight: halving every third of the tenor, summing to its lengthfixed by the tenor–
βi\beta_ithe asset's 252-session beta, known when the tenor openedthe line's own estimate–
ωh\omega_hthe tenor's weight: 12, 20, 28 and 40% for 21, 63, 126 and 252 sessionsfixed by the tenors' lengths–

the author's decision of 25 Oct 2026 (D-2026-10-25-RS5). The rank correlation of beta with alpha is averaged over the tenors with equal weights; the band resamples the year's days once for all four tenors.

The four tenors' band: one resampling of the year's days

γˉ(b)=∑hωh 252h γ(h,b),γ(h,b) fitted on ∑tcbt wt(h)xit\bar\gamma^{(b)} = \sum_h \omega_h\, \tfrac{252}{h}\, \gamma^{(h,b)}, \qquad \gamma^{(h,b)} \text{ fitted on } \sum_t c_{bt}\, w_t^{(h)} x_{it}
band=γˉ±1.645 sd⁡b(γˉ(b))\text{band} = \bar\gamma \pm 1.645\, \operatorname{sd}_b\big(\bar\gamma^{(b)}\big)
SymbolMeaningFromAssumed or estimated
cbtc_{bt}how often draw b takes session t of the year: blocks of 5 sessions on average, drawn tenor by tenor so that every draw holds each tenor's length, 200 drawsseeded by the Friday–
γ(h,b)\gamma^{(h,b)}tenor h's slope on draw b, every asset moving with the same drawn day and the betas held fixedthe line's own estimate–

the author's decision of 25 Oct 2026 (D-2026-10-25-RS4). The gap's band holds the market's move as known, as the line's band does, or resamples it on the same days.

The rank: did alpha line up in beta order?

ρ(h)=corr⁡(rank⁡βi, rank⁡αi(h)),ρˉ=14∑hρ(h)\rho^{(h)} = \operatorname{corr}\big(\operatorname{rank}\beta_i,\ \operatorname{rank}\alpha_i^{(h)}\big), \qquad \bar\rho = \tfrac{1}{4}\sum_h \rho^{(h)}
band=ρˉ±1.645 sd⁡b(ρˉ(b)), cut at ±1\text{band} = \bar\rho \pm 1.645\, \operatorname{sd}_b\big(\bar\rho^{(b)}\big), \text{ cut at } \pm 1
SymbolMeaningFromAssumed or estimated
rank⁡\operatorname{rank}the asset's place among the selection's assets, ties averaged—–
αi(h)\alpha_i^{(h)}the asset's alpha over tenor hthe line's own estimate–

Spearman's rank correlation: one extreme return moves a rank by one place, not the line by its whole distance.

Out of line with their beta

qβ(i), qα(i)∈{bottom quarter, middle half, top quarter}q_\beta(i),\ q_\alpha(i) \in \{\text{bottom quarter},\ \text{middle half},\ \text{top quarter}\}
Rm>rf: listed if (qβ,qα)=(top,bottom) or (bottom,top);Rm≤rf: (top,top) or (bottom,bottom)R_m > r_f:\ \text{listed if } (q_\beta, q_\alpha) = (\text{top}, \text{bottom}) \text{ or } (\text{bottom}, \text{top}); \quad R_m \le r_f:\ (\text{top}, \text{top}) \text{ or } (\text{bottom}, \text{bottom})
SymbolMeaningFromAssumed or estimated
qβ(i)q_\beta(i)the quarter of the selection the asset's beta falls inthe line's own estimate–
qα(i)q_\alpha(i)the quarter its alpha over the tenor falls inthe line's own estimate–

the author's decision of 25 Oct 2026 (D-2026-10-25-RS6). Weeks in a corner count the Fridays in a row the asset has sat there. A description of the tenor, not a signal.

The order gauge

turnedi  ⟺  Pitmax⁡63Pi−1≤−σi21,si=sessions since its high if it has turned, else 0\text{turned}_i \iff \frac{P_{it}}{\max_{63} P_i} - 1 \le -\sigma_i\sqrt{21}, \qquad s_i = \text{sessions since its high if it has turned, else } 0
ρt=corr⁡(rank⁡vi, rank⁡si)\rho_t = \operatorname{corr}\big(\operatorname{rank} v_i,\ \operatorname{rank} s_i\big)
SymbolMeaningFromAssumed or estimated
σi\sigma_ithe fund's daily standard deviation over three monthsYahoo Finance–
max⁡63Pi\max_{63} P_iits highest close over the last three monthsYahoo Finance–
viv_iits own volatility over three months (the dashed line ranks its beta instead)Yahoo Finance–

the author's decision of 24 Oct 2026 (D-2026-10-24-RK6). Above zero the most volatile funds turned first; below, last. A description of the current move's order, not a signal.

The reward for beta within sectors, the measure the family read until 25 Oct 2026

Ri=γ0,s(i)+γ1 βi+eiR_i = \gamma_{0,s(i)} + \gamma_1\,\beta_i + e_i
gap to the CAPM line=γ1−(Rm−rf)\text{gap to the CAPM line} = \gamma_1 - (R_m - r_f)
SymbolMeaningFromAssumed or estimated
RiR_imember i's return over the window (21, 63, 126 or 252 sessions)Yahoo Finance; State Street's SPY holdings–
βi\beta_iits 252-session beta against VTI, known when the window openedthe line's own estimate–
γ0,s\gamma_{0,s}one intercept for each sector of the indexthe regression–
γ1\gamma_1what a unit of beta paid within sectorsthe regression–
Rm−rfR_m - r_fthe market's return over cash — the CAPM line's slopeVTI; the 3-month Treasury par yield–

Each tenor's band: resampling the tenor's days

γ^1(b)=∑iβ~i Ri(b)∑iβ~i2,Ri(b)=exp⁡(∑tcbt rit)−1\hat\gamma_1^{(b)} = \frac{\sum_i \tilde\beta_i\, R_i^{(b)}}{\sum_i \tilde\beta_i^{2}}, \qquad R_i^{(b)} = \exp\Big(\sum_t c_{bt}\, r_{it}\Big) - 1
band=γ^1±1.645 sd⁡b(γ^1(b))\text{band} = \hat\gamma_1 \pm 1.645\, \operatorname{sd}_b\big(\hat\gamma_1^{(b)}\big)
SymbolMeaningFromAssumed or estimated
cbtc_{bt}how often draw b takes session t: blocks of 5 sessions on average, wrapping round, 200 drawsseeded by date and window–
ritr_{it}member i's daily log return on session t of the windowYahoo Finance–
β~i\tilde\beta_ithe beta less the mean beta (its sector's mean, within sectors), held fixed in every drawthe line's own estimate–

Decision D-2026-10-11-RF6. The ordinary least-squares error treats the members as independent and is too narrow; the band keeps the days they share.

The price of risk and the family count

At=1∣Jt∣∑j∈Jtpj,tA_t = \frac{1}{|J_t|} \sum_{j \in J_t} p_{j,t}
Nt=∑f=161[ xf,t on the risk-on side of its line ]N_t = \sum_{f=1}^{6} \mathbf{1}\big[\,x_{f,t} \text{ on the risk-on side of its line}\,\big]
SymbolMeaningFromAssumed or estimated
pj,tp_{j,t}reading j's percentile among its own weeks to t, signed so that high is dearCboe, OFR and the options and credit pages–
JtJ_tthe readings with two years of history behind them at t (decision D-2026-10-11-RF4)—–
xf,tx_{f,t}family f's reading: the price of risk, the reward for beta, the spreads, the co-movement index, positioning, surveysthis page and Correlations & regime–

Decision D-2026-10-11-RF7: the count describes how many families agree; it names no state.

Exhibit 21 · Definitions

  • The monitor
    A description, not a signal.

    Model-conditional only where an estimate is printed: the reward for beta is a regression, with its band. Everything else on this page describes what has already been paid or priced, ranked against its own history; none of it is a forecast, a probability or a recommendation. The page reports each family the industry uses to read risk appetite side by side and counts how many read risk-on (decision D-2026-10-11-RF7); no single family names the state.

  • Reward for beta
    The line, its band, two universes.

    The reward for beta (the author's decision of 25 Oct 2026 (D-2026-10-25-RS5)): each asset's daily returns over cash in a tenor of 21, 63, 126 or 252 sessions, weighted to the recent so that the weight halves every third of the tenor and the weights sum to its length, regressed across the assets on the 252-session beta known when the tenor opened, with no sector intercepts. The slope is what a unit of beta paid; the gap is the slope less the market's value made the same way, which is also the slope of alpha on beta. Each is annualised and the four tenors are averaged with weights in proportion to the square root of the tenor's length (12, 20, 28 and 40%), so each tenor has an equal say in the line's weekly movement; the rank correlation of beta with alpha takes equal weights. The band is one stationary bootstrap of the year's days shared by the four tenors, drawn within the nested stretches 1–21, 22–63, 64–126 and 127–252 sessions back, 200 draws, every tenor refitted on each draw: the composite ± 1.645 standard deviations of the draws. The S&P 500's members are the default; the 18 sector funds and the funds across asset classes, and each universe's presets, are the same fit on their own assets. Every number is the cross-sectional line's and Betas across assets' own, copied as they publish it.

  • Rank form
    Did beta order the alphas? Printed, not counted.

    Beneath the reward for beta the page prints the members' rank correlation of beta with alpha, the four tenors together, with its band: each member ranked by the beta it had when the tenor opened and by its alpha over the tenor, the ranks correlated, and the four tenors averaged with equal weights. It is not one of the six families and does not move the count; a rank correlation of +1 would mean every alpha ranked as its beta did.

  • Spreads
    In standard deviations of their own moves.

    Risk-on spreads (pillar B): each spread's 63-session log return — a side is the mean of its legs' daily log returns; over cash subtracts the 3-month bill accrued daily — is divided by its own trailing 756-session daily standard deviation times √63, so every spread reads in standard deviations of its own recent moves. The equity, credit and commodity legs are exchange sessions; bitcoin is read on the same sessions; the Australian dollar − yen pair is the Board's two daily rates, on its business days. Each spread is ranked against every Friday since its first, blank for the first 104 weeks, and the composite is the mean of the eight ranks; the market over cash is printed beside it and kept out by design, so the composite asks what risk paid relative to safety rather than whether the market rose.

  • Price of risk
    Priced in advance, signed so high is dear.

    The price of risk (pillar A): ten readings priced in advance — the S&P 500's one-month variance premium and 25-delta put skew from the options pages, the excess bond premium and the corporate-over-Treasury 10-year spread from the credit pages, the OFR Financial Stress Index, and from Cboe the VIX, its three-month less one-month term spread (signed so that inversion is dear), SKEW, VVIX and one-month implied correlation. Each is read on Fridays as it was known — a monthly reading from its publication date on — ranked against its own history from its first week, and signed so that a high rank is dear. The composite is the mean over the readings with two years behind them (decision D-2026-10-11-RF4); below 30 risk is priced cheaply, above 70 dearly, as the Valuation page's strip reads its own.

  • Say and hold
    Surveys, leverage, positioning.

    What investors say and hold (pillars E and D): Michigan's index of consumer sentiment, each month shown from 28 days after the survey month's last Friday; FINRA's margin debit balances as a change on the year, each month shown about seven weeks after it ends; and the CFTC's asset managers' net position in E-mini S&P futures as a share of open interest, from Positioning. Each is ranked against its own history; blocked sources are named rather than replaced.

  • Families
    Each family's side of its own line, counted.

    The family count: the price of risk reads risk-on below 50; the reward for beta (the members' gap composite, its point-in-time percentile among the weeks since 2014, decision D-2026-10-25-RS5), the spreads composite, equity positioning and consumer sentiment at or above 50; the co-movement index of Correlations & regime below zero. Each side is counted on every Friday from the first on which all six have a value.

  • Percentiles
    Point in time, ties at or below.

    Every percentile is expanding and point in time: the value at week t is the share of the values from the series' first week to t that are at or below it, ties included, so the line drawn through history is the line a reader would have seen at the time; today's value uses the whole history and equals the full-sample share (a check).

  • Retired gauge
    The composites and the four names.

    The retired gauge (decision D-2026-10-11-RF7): until 11 Oct 2026 this page ranked the funds' slope and intercept at four horizons, weighted the ranks into two composite scores and named four states from them — risk-on, broad rally, defensive, de-risking. The scores rested on the ordinary least-squares line, whose error the funds' shared shocks make about 2.5 times too narrow, and named a state from one family alone. They are reproduced week by week in the reconciliation fold, by the same rules, with the September report's week marked.

  • Caveats
    What this cannot tell you.

    What this cannot tell you. A percentile has no units: a 90th-percentile VIX in a calm decade and in a violent one are not the same number of points. The members' line describes today's members through their history. The count treats the six families as equals, which they are not, and two of them (spreads and the reward for beta) share the market's days.

  • Licence
    Whose numbers these are.

    Prices are Yahoo Finance data; index histories are Cboe's; the stress index is the Office of Financial Research's; the surveys are the University of Michigan's and FINRA's; the copied readings belong to their pages and keep those pages' attributions.

What each family shares with the market. Correlation of each measure's weekly values with the S&P 500's 13-week return and with the VIX: weekly values since 2014 against the S&P 500's (SPY) 13-week log return and the VIX close.

MeasureWith the marketWith the VIX
Reward for beta beyond CAPM, members, 63 sessions+0.03+0.17
gap_composite_members−0.06+0.27
q50_63_members+0.70−0.23
spread_63_members+0.25−0.06
Risk-on spreads, mean percentile+0.57−0.25
The price of risk, mean percentile−0.51+0.79
Consumer sentiment (Michigan)−0.07−0.22
Margin debt, change on the year+0.10−0.09
Asset managers' E-mini S&P position+0.36−0.20
Families reading risk-on+0.46−0.53
Source: Yahoo Finance; Cboe Global Indices; Office of Financial Research; University of Michigan; FINRA; CFTC; Macro Review calculationscross.sentiment-10

Exhibit 22 · Definitions

  • Caveats
    What this cannot tell you.

    Three limits, first. The cross-section is 18 sector and industry ETFs whose betas are packed together: over the latest 63-session window half of them lie inside an interquartile range of 0.66 (full spread −0.25 to 2.33), so the slope is identified off a narrow spread of the regressor and its standard error is correspondingly wide (1.8pp at 63 sessions, 10.5pp at 252). Alphas at 21 sessions carry Newey–West standard errors of the order of 49% a year (median across the assets), so almost nothing at that horizon is distinguishable from zero. And the 21- and 63-session alphas of the same asset overlap in data and correlate 0.57 on average since 2014, so the three-horizon composite is not three independent readings. ETF returns are computed on adjusted closes (dividends reinvested), so a dividend is reinvested on its ex-date rather than dropped: the return is what a holder earned, not what the quoted price did.

  • Universe
    18 funds, one benchmark.

    The universe is the 18 US sector and industry ETFs of Factor analysis's weekly security market line (XLK, XLC, XLY, XLP, XLV, XLF, XLI, XLE, XLB, XLU, XLRE, SMH, KRE, XHB, XOP, IYT, VNQ, XRT), the benchmark is the total-market ETF VTI. Two of them list later than the sample start — Communication services from 19 Jun 2018 and Real estate from 8 Oct 2015 — so a cross-section before those dates has 16 or 17 assets; the number in every window is stored and shown. Styles, indexes and a global universe are not on this page.

  • Members
    The S&P 500's members.

    The constituents' line (decision D-2026-10-11-RF9; the page's default): the same window and the same betas — each member's 252-session beta of its excess return on VTI's, known when the window opened — on the S&P 500's 503 members, with one intercept per sector (11 sectors from State Street's sector funds), so the slope is what beta paid within sectors rather than which sectors did well. Its band is the same day bootstrap on the same draws as the funds' line. The members are the index's on 13 Nov 2026, the first holdings file this site stored, and every earlier week uses them: a company that left the index before then is missing from the history and one that joined later is counted before it joined. The history describes today's members, not the index as it stood. Since 1 Jan 2024 the two lines' weekly 63-session slopes correlate 0.83 over 150 Fridays. The 18 funds remain the alternative view.

  • Risk-free
    Accrued daily.

    The risk-free rate is the 3-month Treasury par yield accrued daily: rf_d = (1 + y/100)^(1/252) − 1 on each NYSE session, with the yield forward-filled from Treasury business days. Over a window of h sessions the rate compounds: rf_h = Π(1 + rf_d) − 1 (1.02% over the latest 63 sessions, 3.84% over 252). On 13 Nov 2026 the published yield was 4.32%; a check compounds the daily rate back to it.

  • Cross-section
    The security market line.

    Cross-sectional line, every Friday since 3 Jan 2014: the realised h-session total return of each asset regressed on its trailing beta, R_i = γ0 + γ1 β_i + e_i, by OLS (classical standard errors) and by quantile regression at the 10th, 50th and 90th percentiles (statsmodels QuantReg, as in the September review's SML). The beta is the 252-session estimate ending at the session that opens the window, so it was known when the window opened; an asset enters a window only when it has both that beta and the full h-session return. Beside each window the page prints the CAPM comparison on the same sessions: rf_h and R_m,h − rf_h. γ0 − rf_h is the zero-beta excess (positive means the low-beta end earned more than cash — the Black / betting-against-beta sign) and γ1 − (R_m,h − rf_h) the shortfall of the reward per unit of beta against the CAPM line.

  • Four tenors
    Recent days weigh more; an equal say.

    The four tenors together. Each Friday the line is fitted over 21, 63, 126 and 252 sessions on each asset's excess return, its days decayed so that a day's weight halves every 33% of the tenor (a half-life of 7, 21, 42, 84 sessions); the weights sum to the tenor, so a steady return keeps its sum. Each tenor's slope, intercept, gap and quantile spread are annualised by 252 over the tenor and averaged with weights in proportion to the square root of the tenor, which gives each tenor about an equal share of the line's weekly movement (decision D-2026-10-25-RS3); equal weights are the other view. The rank correlation is not annualised and takes equal weights. The cross-section the scatter draws is the plain line on simple returns over the tenor, each asset's beta the 252-session beta known when the tenor opened, with no sector intercepts (D-2026-10-25-RS2).

  • Band
    One resampling shared by the tenors.

    The band. The four tenors share one resampling of the year's sessions: 200 draws of a stratified stationary bootstrap with blocks of 5 sessions on average, drawn shell by shell (0–21; 21–63; 63–126; 126–252 sessions back) so that every draw holds each tenor's length, with every asset moving together and the betas held fixed. The band is the estimate ± 1.645 standard deviations of the draws (90%); the rank correlation's band is cut at ±1. The gap's band holds the market's move as known, as the line's band does, or resamples it on the same days (D-2026-10-25-RS4). The draws are seeded by the Friday, so a band reproduces.

  • Line's band
    Resampling the tenor's days.

    The band (decision D-2026-10-11-RF6): each Friday's slope is refitted on 200 draws of the window's sessions, resampled jointly for every fund — a circular stationary bootstrap with blocks of 5 sessions on average — with the betas held fixed, so the funds' common shocks stay in every draw; the band is the estimate ± 1.645 × the draws' standard deviation (90%). The ordinary least-squares error treats the funds as independent and is too narrow by a factor of 2.0 at 21, 2.5 at 63, 2.5 at 126, 2.6 at 252 sessions (the median ratio of the two errors across Fridays since 2014); with it the CAPM line sat outside the band in about half the weeks, with the bootstrap in 14%, 7%, 5%, 5% of them. The draws are seeded by date and horizon, so every band reproduces. Recompounding the drawn sessions lifts the draws' mean above the estimate, more for volatile, high-beta funds and longer windows (by 0.07, 0.17, 0.26, 0.36 ordinary errors on average): compounding is convex. On log returns, where nothing compounds, the draws centre on the estimate within Monte Carlo error, and that is the check. Across weeks the mean slope carries a Fama–MacBeth error: the standard deviation of the weekly slopes with Newey–West weights over ⌈h ÷ 5⌉ weeks, which prices the overlap of neighbouring windows.

  • Rank
    Did beta order the alphas?

    The rank gauge (the author's decision of 24 Oct 2026 (D-2026-10-24-RK1)): each Friday and window, Spearman's rank correlation across the universe of each asset's beta, known when the window opened, with its CAPM alpha over the window, α_i = R_i − rf_h − β_i (R_m,h − rf_h) — the Pearson correlation of the two columns' average ranks. It is the rank form of the gap to the CAPM line: regressing those alphas on beta gives that gap as the slope, and ranking them asks whether the order of the betas was the order of the alphas, so one member's extreme return moves its rank by one place rather than the slope by its whole distance. Beta against the window's return is printed beside it as a sensitivity. The members are ranked within sectors (the author's decision of 24 Oct 2026 (D-2026-10-24-RK2)): each member's beta, alpha and return less its sector's mean, as the scatter draws them; the plain ranks across all the members are −0.20 at 21, −0.29 at 63, −0.20 at 126, −0.05 at 252 sessions today, and the two versions' weekly 63-session readings correlate 0.92 over 672 Fridays. Its band is the line's own: the same 200 draws of the window's sessions, each recompounding every asset's return and the market's, the betas and the risk-free accrual held fixed; the band is the estimate ± 1.645 × the draws' standard deviation, cut at ±1. If the assets were independent the error near zero would be 1/√(n − 1) — 0.045 for the 499 members, 0.243 for the 18 funds — but they share the window's days, which the draws keep, so the band is wider than that. Over 63 sessions the members' gauge and their gap to the CAPM line correlate 0.88 across 672 Fridays, the funds' 0.79.

  • Identity
    Why the two agree.

    The two readings are the same object. Define each asset's alpha over the window as R_i − rf_h − β_i (R_m,h − rf_h); regressing those alphas on beta gives g0 = γ0 − rf_h and g1 = γ1 − (R_m,h − rf_h) exactly, because the transformation is linear in the regressor. A check reproduces it on the latest window at all four horizons to 1e-09.

  • Quantiles
    The 10th, 50th and 90th lines.

    Quantile regression is a linear programme whose optimal basic solution has two zero residuals, so with at most 18 points the exact fit is the best of the at most 153 lines through two of them. This page solves it that way at every window (8,064 fits: 4 horizons × 672 weeks × 3 quantiles), which removes the stopping error of an iterative solver. A check confirms that statsmodels' QuantReg — the routine the September review's SML used, and the one this page uses to reproduce it — reaches no lower value of the check function on the displayed windows; on those 12 fits its slope differs from the exact one by at most 0.00pp, the difference being windows where the iteration stops at a neighbouring vertex.

  • Range
    Where the scatter stops.

    The scatter's range. With 100 assets or more, the range stops 3 interquartile ranges beyond the quartiles of the returns (or the alphas); an asset beyond is drawn as a marker at the edge and named beneath, and every asset is in the fit (D-2026-10-25-RS8). The lines are drawn from the lowest beta through the cloud's centre to the highest; the slope's band pivots on the centre, where every line through the cloud passes.

  • Watch list
    Out of line with beta.

    The watch list and the rank map. Beta (252 sessions, known when the tenor opened) and alpha over the tenor are each split into quarters. With the market up over cash, high beta in the bottom alpha quarter and low beta in the top are listed; with it down, the reverse; the middle half of beta is never listed (D-2026-10-25-RS6). Weeks in a corner count the Fridays in a row the asset sat there. The rank map ranks alpha over one month and over twelve, in percentiles of the assets with both. Both describe the tenor and are not signals.

  • Reconciliation
    The September figures, as printed.

    Reconciliation to the issue — the author's decision of 3 Oct 2026 (D-2026-10-03-01). The September pack's figures for that line (OLS 2.383 / 13.739, median slope 9.072, VTI 14.18%) are the factors page's line as published on 22 Sep 2026: the 18 sector and industry ETFs of this page; 156-week weekly log-return betas against VTI to the last complete week; year-to-date total returns on adjusted closes from the last close of 2025; the 2026 average 3-month Treasury par yield accrued to date as the risk-free rate; OLS and statsmodels QuantReg at the 10th, 50th and 90th percentiles. Their return leg ends on the 21 Sep 2026 bar taken before that session closed, so no line drawn on completed sessions can reproduce them, and the check that compared them with one is retired by that decision. They are printed like-dated in the Detail fold beside the same construction computed here on completed sessions to the same two dates (18 Sep 2026 and 21 Sep 2026), whose fit a check reproduces with statsmodels.

  • Storage
    What the page stores.

    What is stored: the 63- and 252-session alpha and beta of every asset as weekly (Friday) values at two decimals from 1 Jan 2016; the alpha, beta and both standard errors at each of the three half-lives on the same weeks; the three composites and the z-scores behind them; the smoothed Kalman alpha and its standard error weekly; the latest values of all four windows for every asset (the term structure); and the γ series of all four horizons weekly from 3 Jan 2014. That is 18 × 568 × 12 half-life numbers alone, so the page's data is published in two files: this page loads the last 5 years and opens the full-history file when a chart asks for more, exactly as the consumer-price page does. The 21- and 126-session windows keep only their latest values and their summary statistics. Rolling estimates at adjacent dates share almost all of their data.

  • Licence
    No level shown.

    ETF prices are Yahoo Finance data, shown with their attribution; the 3-month Treasury par yield is free to use. Every regression on this page — alphas, betas, slopes and their standard errors — is computed here from adjusted closes.

Source: Yahoo Finance chart API; U.S. Treasury daily par yield curve; State Street Global Advisors' holdings.cross.sml-2

Exhibit 23 · Betas across assets: definitions

  • The line

    The same construction as the funds' line, on 44 funds against VTI: simple returns of adjusted closes on the NYSE calendar; the 3-month Treasury par yield accrued daily as the risk-free rate; each fund's 252-session least-squares beta of its excess return on VTI's, known when the window opens; the window's simple return regressed on it by ordinary least squares, every fund weighted equally. One-week windows (Friday to Friday) since 2014; one-day windows for the last 63 sessions. Bitcoin is in the line (the author's choice of 15 Oct 2026); its weekend moves fall into Monday's return.

  • A fund's part

    A fund's part of the slope is (its beta − the average beta) × (its return − the average return) ÷ the sum of the squared beta gaps. The parts add up to the slope exactly. A fund with a beta below the average and a return above it pulls the slope down; so does a high-beta fund that lagged.

  • The band

    The week's band is the funds' line's day bootstrap, the author's decision of 11 Oct 2026 (D-2026-10-11-RF6): the window's five sessions resampled jointly for every fund in blocks of mean length 5, 200 draws, each fund's return recompounded and the slope refitted with the betas held fixed. A one-day window has no band: one session cannot be resampled.

  • Dynamic betas

    GARCH(1,1) with Student-t errors on each fund's daily log returns since 2006 (or its first session), then one DCC(1,1) correlation process for every pair shown, its a = 0.0314 and b = 0.9585 fitted by composite likelihood over 226 pairs (the sum of the exact two-asset likelihoods, each on its own overlap; Engle, Shephard and Sheppard 2008). A fund's daily beta is its conditional correlation × its volatility ÷ the market's, as of each close. Beside it the 252-session least-squares beta on daily log returns.

  • The pull to the long run

    The correlation is (1 − w) × the pair's 2006–26 average + w × a 16-session running average of recent co-movement, w = a/(1 − b) = 0.76: a lasting change shows at about 76% of its size. The parameters and the long-run averages are re-estimated on the whole sample at every update, so the history moves a little with each new day (look-ahead in the parameters, not in the daily path).

  • Each pair's own persistence

    Each fund's pair with the market refitted with its own a and b on a grid: half-lives from 11 to 18 sessions for the middle half (median 13); the table's "own" column prints that fund's beta under its own persistence.

  • Country funds

    US-listed country funds trade in New York hours, so their daily betas carry no timing gap with the US market; the table prints the three-year weekly beta beside the daily one.

  • Flags

    A fund whose volatility fit puts nearly all its weight on the last day's move (α above 0.5 or ν above 100), or sits on the boundary where its variance never moves (α below 0.001 and β above 0.999), is marked † and its daily beta is not read.

  • Coverage

    Funds without prices on the data day, and so not in any window: none.

  • The 18-fund line

    The site's 18-fund line on the same construction: its one-week slope correlates 0.58 with this line since 2014 and 0.76 over the last 104 weeks; its 21-session slope is checked against the funds' line published on this page.

  • The rank gauge

    The rank gauge (the author's decision of 24 Oct 2026 (D-2026-10-24-RK1)): every Friday and window, Spearman's rank correlation across the funds of each fund's 252-session beta, known when the window opened, with its CAPM alpha over the window, α = R − rf − β (R_m − rf): the Pearson correlation of the two columns' average ranks. Beside it, the same with the window's return. Its band is the line's day bootstrap on the same statistic, on the same draws as the S&P 500 members' line for that Friday and window: 200 resamplings of the window's sessions, every fund's and the market's return recompounded, the betas held fixed; the band is the estimate ± 1.645 × the draws' standard deviation, cut at ±1. Were the funds independent, its error near zero would be 1/√(n − 1), 0.15 for the 44 funds over the latest 63 sessions against 0.21 from the draws.

  • Betas as borne

    Betas as borne (the author's decision of 24 Oct 2026 (D-2026-10-24-RK5)): each fund's beta on the daily model known at the previous close, applied session by session — its conditional alpha is the sum over the window of r − rf − β (r_m − rf), ranked against the window's mean of that beta. A check holds each fund's excess return to its conditional alpha + the mean beta × the market's excess return + the timing term, every window. Its weekly readings over 63 sessions correlate 0.79 with the gauge's over 672 Fridays. On the 504-session beta, the attribution page's exposure window, they correlate 0.94; today that reading is −0.28.

  • The rank map

    The rank map (the author's decision of 24 Oct 2026 (D-2026-10-24-RK3)): on the latest Friday, each fund's beta, alpha and return, the least-squares line of alpha on beta across the funds and each fund's residual to it, and each column's rank (1 the lowest). A rank says where a fund sits; it is not a signal, a lag or a target.

  • Bitcoin's beta

    Bitcoin's beta has no settled value (the author's decision of 24 Oct 2026 (D-2026-10-24-RK7)): on 13 Nov 2026 it is +1.48 over 252 sessions of daily returns, +1.09 over 504, +0.73 on 156 weeks of weekly returns and +1.21 on the daily model — a spread of 0.74. Weekend moves fall into Monday's daily return; a weekly return holds them whole.

  • The order gauge

    The order gauge (the author's decision of 24 Oct 2026 (D-2026-10-24-RK6)): each session, across the funds, a fund has turned once it sits at least one month's sigma — its 63-session standard deviation of daily returns × √21 — below its 63-session high, and the sessions since that high count from then (0 for a fund that has not turned). The gauge is Spearman's rank correlation of each fund's own 63-session volatility with those sessions; the second line ranks its 252-session beta known at the previous close instead. Above zero the most volatile turned first; below, last. It uses closes to the day only: a check reruns it on data to the day before and reproduces that day. Beneath it, the high-beta spread (SPHB less SPLV, the legs of the spreads above) from its own 63-session high; the reference line is the gauge's median on the sessions it sat 8% or more below it.

  • Past sell-offs

    The sell-offs behind it, descriptive and dated: since 2015, 26 high-beta unwinds of 8% or more: the median rank correlation of beta with the fall −0.55, of volatility with the timing −0.36 and of beta with the timing −0.28; 16 US market falls of 7% or more: the median rank correlation of beta with the fall −0.56, of volatility with the timing −0.37 and of beta with the timing −0.31. An episode opens when the level sits its threshold below its 63-session high; its peak is that high and its trough the low before it recovers a third of the way; episodes are merged on their trough. Per asset: its beta at the close before the peak, its volatility over the 63 sessions before, its own fall, and the sessions from the peak until it was halfway down.

Source: Yahoo Finance; U.S. Treasury; Macro Review calculations · data to 13 Nov 2026cross.betas-7

Before you read

Macro Review is a personal research site, still being built.

Nothing on it is investment advice, a recommendation or a solicitation.

Check every figure for yourself. Each names its publisher — confirm it there before you rely on it.

Figures can be revised, delayed or wrong.

Cones, fans, scenarios, fair values and “what followed” tables are conditional on the model each page names: they describe, they do not predict.

No warranty is given; use is at your own risk.