Sector rotation monitor, reproduced
As of 2026-09-10, 3 of the 14 themes score 60 or above on this monitor: crowded advance for Energy (XLE); early rotation for Technology (XLK), Communication Services (XLC). Money-leaving rules are on for Consumer Discretionary (XLY), Consumer Staples (XLP), Industrials (XLI), Materials (XLB), Real Estate (XLRE). SPY itself scores 8 (money leaving, double negative, weakening), with 9% of sectors above their 20-day average. This page reproduces, rule for rule, the sector rotation monitor published by Benjamin Usagi and recomputes it nightly. The scores describe recent prices; the cut-offs are the author's judgement calls and were never back-tested. Quant Data tested the rules and found no forward edge. Nothing here is a recommendation.
What this monitor does, in plain words
Take each sector ETF and divide its price by SPY's price. That ratio goes up when the sector beats the market and down when it lags. The monitor looks at that ratio through five simple questions and gives points for each "yes". Add them up, cap at 100, and you get the score. Then a few cut-offs turn the score into a label.
| Evidence | Question asked of the sector-to-SPY ratio | Max points |
|---|---|---|
| Trend | Is the ratio above its 20-day average (+10)? Above its 50-day average (+10)? Is the 20-day average higher than five days ago (+10)? | 30 |
| Flow | Up vs SPY over 5 days (+10)? Over 20 days (+8)? Is the 5-day number better than it was five days ago (+8)? Is the 5-day pace faster than the 20-day pace, i.e. accelerating (+8)? | 34 |
| Volume | On a day the sector beat SPY, was dollar volume at least 1.3× its 20-day average (+20)? At least average (+10)? | 20 |
| Breadth | Placeholder for ETFs: 70 if the ETF is above its own 20-day average, else 30. 70 earns +15, 50 earns +8. | 15 |
| Not crowded | +8 unless the sector has run too far: up more than 12% vs SPY in 60 days, or more than 6% above its 50-day ratio average. | 8 |
Labels: 75 and above = "confirmed entry"; 60 to 74 = "early rotation"; below 45 with both 5- and 20-day relative returns negative = "money leaving". Two patterns override the score: "distribution" (ratio below its 20-day average, down vs SPY over 5 days, on heavy volume, down today) and "crowded advance" (score 60+ but already over-extended). These names are the author's; on this page they are labels for rules, not signals.
Today's table, 2026-09-10
| Theme | Score | State | Why (author's label) | 5d vs SPY | 20d | 60d | Accel. | Trend | Vol × | Breadth | Last 60 days |
|---|---|---|---|---|---|---|---|---|---|---|---|
| SPY benchmark, absolute score | 8 | Money leaving | Double negative, weakening | -0.96%abs | -1.90% | +0.40% | — | weak | 1.11 | 9% | |
| Technology XLK | 69 (+5 boost) | Early rotation | Early rotation | +1.86% | -0.03% | -3.81% | +1.87 pts | strong | 0.91 | 30placeholder | |
| Financials XLF | 39 (+5 boost) | Neutral watch | Neutral | -0.42% | +0.09% | +5.76% | -0.44 pts | weak | 0.84 | 30placeholder | |
| Health Care XLV | 26 | Neutral watch | Neutral | -3.29% | +0.25% | +7.92% | -3.35 pts | weak | 0.96 | 30placeholder | |
| Consumer Discretionary XLY | 31 (+5 boost) | Money leaving | Double negative, weakening | -1.58% | -3.19% | -5.95% | -0.78 pts | weak | 1.03 | 30placeholder | |
| Consumer Staples XLP | 18 | Money leaving | Double negative, weakening | -1.91% | -0.45% | -3.18% | -1.80 pts | weak | 1.05 | 30placeholder | |
| Energy XLE | 83 | Crowded advance | Crowded advance | +0.70% | +8.45% | +16.42% | -1.41 pts | strong | 1.43 | 70placeholder | |
| Industrials XLI | 24 | Money leaving | Double negative, weakening | -0.34% | -6.47% | -4.93% | +1.28 pts | weak | 1.10 | 30placeholder | |
| Materials XLB | 8 | Money leaving | Double negative, weakening | -3.21% | -1.59% | -3.70% | -2.81 pts | weak | 0.85 | 30placeholder | |
| Utilities XLU | 34 | Neutral watch | Holding up in a down tape | +0.61% | -1.13% | -5.34% | +0.90 pts | weak | 1.02 | 30placeholder | |
| Real Estate XLRE | 16 | Money leaving | Double negative, weakening | -0.60% | -1.36% | -4.69% | -0.26 pts | weak | 0.89 | 30placeholder | |
| Communication Services XLC | 74 | Early rotation | Early rotation | +0.14% | +3.07% | -1.01% | -0.63 pts | strong | 1.21 | 30placeholder | |
| Small caps (Russell 2000) IWM | 16 | Distribution | Distribution | -1.20% | -3.12% | -2.74% | -0.42 pts | weak | 1.47 | 30placeholder | |
| S&P 500 equal weight RSP | 8 | Distribution | Distribution | -1.54% | -1.71% | -0.26% | -1.11 pts | weak | 1.30 | 30placeholder | |
| Nasdaq-100 QQQ | 54 | Neutral watch | Holding up in a down tape | +0.89% | -0.18% | -5.12% | +0.93 pts | lean strong | 0.98 | 30placeholder |
"Accel." is the author's acceleration: the 5-day relative return minus a quarter of the 20-day one, in percentage points; positive means the last five days ran faster than the 20-day pace. "Vol ×" is today's dollar volume over its 20-day average. The score bar shows the five parts in order: trend, flow, volume, breadth, not crowded. Dotted lines on the sparklines mark 45, 60 and 75. Some rows carry a small "boost" from the relationship scores below, exactly as in the author's tool.
Eight cross-sector relationships
Beyond scoring each sector alone, the monitor asks eight questions about how sectors move against each other, with bonds (TLT), high-yield credit (HYG), gold (GLD) and regional banks (KRE) as supporting evidence. Each gets its own 0-100 score from a fixed recipe; the author treats 60 as "worth noticing". Every recipe also names what would switch it off.
| Relationship | Question | Score | Today's inputs | Switches off when | Last 60 days |
|---|---|---|---|---|---|
| Offense vs defense | Is money moving between the aggressive group (XLK, XLY, XLF) and the defensive group (XLP, XLU, XLV)? | 50 | direction: toward offense · defense minus offense acceleration -1.63 pts | The receiving group stops accelerating, or its 5-day relative return turns negative. | |
| Small-cap handoff | Are small caps (IWM) accelerating against the index while credit (HYG) holds up? | 15 | IWM acceleration vs SPY -0.42 pts | IWM acceleration turns negative, or high-yield credit weakens. | |
| Equal-weight catch-up | Is the equal-weight index (RSP) catching up, i.e. are gains spreading beyond a few leaders? | 0 | 4 of 11 sectors up vs SPY over 5 days | RSP acceleration turns negative, or fewer than 5 sectors are up vs SPY. | |
| Rate-driven | Are rate-sensitive sectors (XLU, XLRE) and financials (XLF) moving in the way a rates move would explain? | 0 | rates and sectors not aligned · TLT 5-day -1.43% | The TLT move and the sector gap stop pointing the same way. | |
| Late-cycle inflation | Are energy and materials accelerating against tech, with gold up and bonds down as supporting evidence? | 30 | energy+materials vs tech acceleration -3.91 pts | The energy+materials vs tech acceleration turns negative. | |
| Real haven vs fake defense | If defensives are leading, are bonds, gold and credit also acting like a retreat (real haven), or is it just sector rotation? | 0 | defense not leading, so this question does not apply today | Defensives stop leading, or bonds/gold/credit stop confirming. | |
| Market-wide de-risking | How many of the 11 sectors are down vs SPY and down outright over 5 days? | 47 | 7 of 11 down vs SPY · 10 of 11 down outright | Sectors down vs SPY fall back below 6. | |
| Month / quarter-end rebalance | Are we in a calendar window where mechanical rebalancing usually happens, and do winners decelerate while laggards recover? | 45 | outside the windows · 1 winners decelerating · 4 laggards recovering | The calendar window passes; the pattern outside the window needs another explanation. |
What we found when we tested the rules
The author states plainly that the weights and cut-offs are experience-based and were never back-tested. Quant
Data did the test, pre-registered before any data was read: four of the monitor's states were defined on the
11 SPDR sectors from 2010 to 2026-09-03, and the relative return against SPY over the next 10, 20 and 60
trading days was compared with ordinary days. 2010–2021 was the observation window; 2022 onward was held out.
Result: 24 of 24 judged cells were NOT CONFIRMED. The held-out numbers for
the sector pool are below; the full report, including the industry-ETF pool, is in the repository as
research/RESULTS-sector-rotation.md.
| State | Horizon | Events, 2022+ | Hit rate vs random days [95% CI] | Verdict |
|---|---|---|---|---|
| RRG enters Leading | 10 days | 377 | -0.38% [-5.84%, +4.68%] | NOT CONFIRMED |
| RRG enters Leading | 20 days | 370 | -1.43% [-5.94%, +3.34%] | NOT CONFIRMED |
| RRG enters Leading | 60 days | 359 | -0.02% [-4.36%, +4.40%] | NOT CONFIRMED |
| Crowded flag turns on | 10 days | 94 | 20-day max drawdown, median vs random: -0.35% [-1.39%, +0.52%] | NOT CONFIRMED |
| Crowded flag turns on | 20 days | 94 | 20-day max drawdown, median vs random: -0.35% [-1.39%, +0.53%] | NOT CONFIRMED |
| Crowded flag turns on | 60 days | 94 | 20-day max drawdown, median vs random: -0.35% [-1.41%, +0.54%] | NOT CONFIRMED |
| "Confirmed entry" rule turns on | 10 days | 408 | -3.28% [-7.95%, +1.48%] | NOT CONFIRMED |
| "Confirmed entry" rule turns on | 20 days | 404 | +1.72% [-3.14%, +6.97%] | NOT CONFIRMED |
| "Confirmed entry" rule turns on | 60 days | 385 | -1.32% [-6.18%, +3.62%] | NOT CONFIRMED |
| "Money leaving" rule turns on | 10 days | 528 | -0.21% [-4.05%, +3.75%] | NOT CONFIRMED |
| "Money leaving" rule turns on | 20 days | 524 | -1.07% [-5.27%, +2.95%] | NOT CONFIRMED |
| "Money leaving" rule turns on | 60 days | 502 | +0.47% [-3.65%, +4.64%] | NOT CONFIRMED |
Two things worth knowing from the same test. The crowded flag was followed by deeper 20-day drawdowns in 2010–2021 (about 2 percentage points deeper than random days, interval excluding zero) but the effect shrank to 0.4 points and lost significance after 2022. And on days these states switch on, dollar volume is only 0.9 to 1.2 times normal, so the "volume confirmation" part of the score rarely has anything to confirm.
Common questions
What is a sector rotation monitor?
A table that scores each part of the stock market (technology, energy, utilities and so on) on whether it has been doing better or worse than the whole market lately, and labels the pattern. "Rotation" is the idea that money moves between sectors in waves. This page reproduces one specific monitor, built by the publisher Benjamin Usagi, so you can see exactly how such a tool works.
What does a score of 75 mean?
Nothing more than "this sector currently satisfies most of the five rules": its price ratio against SPY is above its 20- and 50-day averages, it has gone up against SPY over 5 and 20 days, it did so on above-average volume, it is above its own 20-day average, and it has not run too far too fast. The author labels 75 and above "confirmed entry" and 60 to 74 "early rotation". Those cut-offs are his judgement; they were never calibrated on history.
Does a high score mean the sector will keep outperforming?
No. Quant Data tested the rules behind these states on 11 sectors from 2010 to 2026, with 2022 onward held out: in all 24 judged cells the following 10-, 20- and 60-day relative returns were indistinguishable from random days. The monitor describes what has already happened; it has not shown any ability to say what happens next.
Why does the breadth column say 70 or 30 for every sector?
Because the monitor has no data on the individual stocks inside each ETF. It uses a placeholder: 70 if the ETF is above its own 20-day average, 30 otherwise. The only real breadth number on the page is the SPY row, which counts how many of the 11 sector ETFs are above their 20-day averages.
Is this investment advice?
No. Every number is a description of past prices computed by fixed rules, attributed to the author who designed them. Quant Data takes no view on any sector and makes no recommendation.
How to cite this
Free to read, no key and no account — link straight to it.
Plain text
Quant Data. "Sector rotation monitor, reproduced (nightly scores)." quantdata.uk, 2026-09-10. https://quantdata.uk/research/sector-rotation-monitorBibTeX
@misc{undefined,
author = {Quant Data},
title = {Sector rotation monitor, reproduced (nightly scores)},
year = {2026},
howpublished = {\url{https://quantdata.uk/research/sector-rotation-monitor}},
}Permanent link
https://quantdata.uk/research/sector-rotation-monitorQuoting a figure from this page is fine without asking. If you want the underlying per-observation table for something we have not published, or you spot a number you think is wrong, mail quantdata@quantdata.uk — a corrected number is worth more to us than a cited one.
Method and parameters reproduced from Benjamin Usagi's "全市场板块资金轮动观测机 v1.3m" (GeoScope, 2026-08-09), a publisher tracked in this site's research ledger. Prices: yfinance daily close, split-adjusted, dividends not reinvested (price series, like TradingView defaults). Recomputed nightly and stamped to the data date, not the build time. Educational only: not investment advice, not a recommendation, not a forecast.