How to read the desk

Eight dense tiles sit across the top. Five are ETF index proxies: SPY for the S&P 500, QQQ for the Nasdaq 100, DIA for the Dow, IWM for the Russell 2000, and VIXY for volatility. Each shows the ETF level plus its daily percent change, coloured green or red by the sign of that change. VIXY is the deliberate exception: its colour is inverted, so volatility falling (VIXY down) shows green to signal de-risking conditions, and volatility rising shows red. The remaining three tiles are computed reads. Regime classifies RISK-ON versus RISK-OFF from the ratio of cyclical to defensive sectors measured against its own moving average. Breadth reports the percentage of sectors trading above their 50-day average. Leader names the single top-ranked sector by the Outlier RS composite, pulled from the same ranking that feeds the leaderboard so the strip and the leaderboard can never disagree. None of these are dollar fund flows; they are ETF proxies and relative reads.

• A green index tile means that proxy is up on the day, red means down — read SPY, QQQ, DIA and IWM together for whether large-cap, tech, blue-chip and small-cap are moving in agreement or splitting.
• On the VIXY tile the colour is inverted on purpose: green means volatility is falling (conditions easing), red means volatility is rising (conditions tightening). Do not read green VIXY as 'volatility up.'
• Regime reading RISK-ON means the cyclical/defensive ratio sits above its moving average; RISK-OFF means it has crossed below. It is a state label, not a forecast or a signal to act.
• Breadth as a high percentage means most sectors are above their 50-day line (broad participation); a low percentage means strength is narrow even if the headline indices are green.
• Leader shows the current #1 sector by the Outlier RS composite. Because it is drawn from the same ranking as the leaderboard, treat any apparent mismatch as a display issue, not two competing signals.
• Cross-check the tiles against each other: indices green but Breadth low, or Regime RISK-ON while VIXY is red, is internal disagreement worth noting before you lean on any one tile.

Once a month (roughly every 21 trading days) the rule ranks the 11 SPDR sector ETFs by a blended excess relative-strength score: it takes each sector's return over a 21-day and a 63-day lookback, subtracts SPY's return over the same windows, and blends the two excess figures. It holds the top 3 sectors equal-weight (one-third each) until the next monthly rebalance, charging 10 bps per side on whatever turnover the new basket implies. The resulting equity curve is rebased to 1.0 and drawn over a SPY buy-and-hold curve rebased the same way. Reported stats: Ann (annualised return), vs SPY (the rotation's excess over the benchmark), Sharpe (risk-adjusted return, marked with a star when the sample is too short to trust), MaxDD (worst peak-to-trough drawdown), Hit (share of rebalances that beat the benchmark, also starred on short samples), Turnover, and Rebal (number of rebalances). It runs on a limited end-of-day history and is labelled "illustrative, not a track record."

• Read the two curves together: where the rotation line sits above the rebased SPY line, the rule was ahead in this sample; where it sits below, holding the index would have done better. The gap is the visual form of the 'vs SPY' stat.
• Treat 'vs SPY' as the headline, not 'Ann'. A high annualised return that is at or below the SPY excess line means the rule mostly captured beta, not rotation skill.
• A star on Sharpe or Hit means the sample is too short to lean on that number — read it as directional, not a measured edge.
• Cross-check Turnover and Rebal against the return: if 'vs SPY' is thin while turnover is high, the 10 bps-per-side cost is eating most of what the ranking generated.
• Read MaxDD as the cost of being concentrated in only 3 sectors — a long-only, top-3 basket can draw down hard when leadership reverses between monthly rebalances.
• Hit tells you consistency, not magnitude: a Hit near or below 50% means any outperformance came from a few rebalances, not a steady edge.

The score is a weighted composite of three inputs, all computed across the same 11-sector cross-section. First, each sector's RS-Ratio is converted to a percentile rank against the other ten sectors; that contributes 0.45 of the weight. Second, each sector's RS-Momentum is converted to a percentile rank the same way, contributing 0.35. Third, a binary trend flag contributes 0.20: it is 1 if the sector is currently above its own 50-day average, 0 if not. The three pieces are summed and scaled to land on a 0–100 range. Because the two percentile inputs are cross-sectional, every score is relative to the current peer set — the numbers describe ordering among these 11 sectors, not absolute strength. The panel renders the result as a segmented bar for the score, with pills showing the underlying RS and MO percentiles and the >MA flag, and colours the row by tier: 66 and above is strong (green), 33 to 65 is mid (blue), below 33 is weak (dim).

• A green row (score ≥66) means the sector ranks highly on the blended measure relative to the other ten right now — it is not a statement that the sector is rising in absolute terms.
• Read the RS and MO pills together: a high RS percentile with a low MO percentile is an established leader that is fading on the momentum axis; the reverse is a laggard that is improving. The single score hides this tension, so the pills are where you recover it.
• The >MA flag is binary and worth only 0.20 — a sector can flip from below to above its 50-day average and move the score by a fixed step regardless of how far above or below it actually is.
• Mid-tier (blue, 33–65) is the crowded middle. Small input changes move sectors in and out of this band, so treat ordering within it as low-confidence.
• Because percentiles are cross-sectional, the whole board can shift when one sector's raw RS changes — a sector's score can fall purely because a peer improved, with nothing changing in the sector itself.
• Compare the leaderboard ranking against the rotation scatter rather than instead of it; if a sector's tier and its scatter quadrant disagree, the disagreement is the signal to inspect, not to average away.

Each sector is measured against SPY as the benchmark. The horizontal axis, RS-Ratio, is the LEVEL of that sector's relative strength versus SPY — z-score-normalised and re-centred so the neutral point sits at 100. The vertical axis, RS-Momentum, is the RATE OF CHANGE of that same relative-strength series, also centred at 100; because it is a derivative of the level, it tends to turn before the level itself does. The 100/100 crosspoint splits the plane into four quadrants: Leading (strong and rising, RS-Ratio above 100 and RS-Momentum above 100), Weakening (strong but falling), Lagging (weak and falling), and Improving (weak but rising). Each sector is drawn as a single dot with a tail attached — the tail traces the last 5 sessions of where that sector has been on the two axes, so you see the path, not just the endpoint.

• Watch RS-Momentum (the y-axis sign) first — it is the faster, leading component. RS-Ratio (the x-axis level) LAGS the underlying price-relative by weeks (StockCharts' own worked example shows roughly 10 weeks), so a sector's quadrant position is a slow, confirming read, not an early one.
• Read the tail, not just the dot — direction of the trail shows whether a sector is accelerating, stalling, or reversing; a lone dot tells you position but not travel. The tails here are short (only a few plotted sessions), so read travel as a hint, not a trajectory.
• Sectors TEND to rotate clockwise: Improving (weak, rising, lower-left up) into Leading (upper-right), then Weakening (strong but rolling over, upper-left) and down into Lagging (lower-left). This is a tendency, not a law — do not extrapolate the arc; rotations stall, skip quadrants, and whipsaw.
• RS-Momentum (y) turning before RS-Ratio (x) is expected by construction, since momentum is the rate of change of the level — a sector crossing above 100 on the y-axis while still left of 100 on x is the Improving signature, not yet confirmed leadership.
• Position relative to the 100/100 cross tells you the quadrant; both axes are centred on 100, so 'above/right of 100' means above the SPY-relative neutral, not above any absolute return level.
• Tail length is a crowding-and-conviction cue: long tails mean fast rotation, tight clusters near the centre mean sectors are tracking SPY closely with little relative signal.

For each sector we take its trailing return and subtract SPY's return over the same span, producing an excess return for three windows: 1W (5 trading sessions), 1M (21 sessions), and 3M (63 sessions). Those three excess-return figures become the columns of the grid, one row per sector. Each cell is coloured by the sign and magnitude of its excess return — green for positive excess (outperforming SPY), red for negative (underperforming), with deeper shade for larger magnitude. A window toggle re-sorts the whole grid by the selected horizon, so you can rank sectors by 1W, 1M, or 3M leadership on demand. The inputs are sector price series and SPY over those three fixed lookbacks; there are no smoothing, weighting, or risk-adjustment steps beyond the raw excess-return calculation.

• Green cell = the sector beat SPY over that window; red = it lagged. The depth of colour scales with the size of the excess return, so a deep-green 3M cell is a larger margin of outperformance than a pale-green one.
• Read a row left to right (1W, 1M, 3M) to see how a sector's leadership has evolved: green across all three suggests persistent strength; red short-window next to green longer-window suggests recent fading of a prior lead.
• Use the window toggle to ask different questions — sort by 1W to surface the most recent movers, by 3M to surface the more established trend. The same sector can sit near the top under one horizon and mid-pack under another.
• Disagreement across columns within a row is itself the signal: a sector green on 3M but red on 1W is decelerating relative to the market; red on 3M but green on 1W is a recent reversal, not a confirmed trend.
• Cross-check against the RRG: the heatmap gives the trailing magnitudes, the RRG gives the trajectory. A sector with strong green numbers here that is rolling over on the RRG is a different read than one that is still climbing.

We compute the ratio of a cyclical-sector basket to a defensive-sector basket and plot it as a single line, overlaid with its own 20-day moving average. The line's position relative to that 20-day MA is the entire signal: above the MA means cyclicals are leading defensives over the recent window (risk-ON); below means defensives are leading (risk-OFF). This is the exact signal that drives the Regime KPI tile, so the panel and the tile never disagree. A small Dispersion tile rides alongside the line — see the dispersion section for how that is built and what it adds. No forecast, no scoring, no directional call: just the ratio, its 20-day average, and which side of it we are on.

• Line above its 20-day MA: cyclicals are leading defensives over the recent window — read this as a risk-ON tilt in appetite, the same state the Regime KPI tile will be showing.
• Line below its 20-day MA: defensives are leading — a risk-OFF tilt. Again, the KPI tile mirrors this exactly.
• A fresh cross of the line through its MA marks a change in which side of the regime you are on; treat a single-bar cross with more caution than a cross that holds for several sessions.
• The line hugging or chopping across the MA means cyclicals and defensives are roughly level — appetite is undecided, and the ON/OFF label is weak rather than informative.
• Glance at the Dispersion tile next to the line for context on how broad-based the move is — a regime read carries different weight when dispersion is high versus compressed (see the dispersion section).
• Read direction and side, not magnitude: the line being far above the MA is not a stronger 'buy' than being just above — it only tells you appetite has been tilted ON for longer or more persistently.

For each tracked sector we compute its own 50-day moving average, then check whether the sector is trading above or below that line. Breadth is the percentage of sectors above their 50-day MA, recomputed each period and plotted as a single line over time. Horizontal reference bands are drawn at 80, 50, and 20 to give the line context. These bands are display conventions for high / middle / low participation — they are not signals or triggers. The most informative use described in the spec is as a divergence flag: comparing the direction of the index against the direction of breadth and watching for the two to disagree.

• Treat this as a CONFIRMATION and health-context gauge, not a timing input. It tells you whether a move is broadly supported; it is not a 'go to cash when breadth drops below 50%' gate, and using it that way over-reads a noisy series.
• Read it with hysteresis, not as a precise cross. This is an 11-name series, so it steps roughly 9.1% per sector across any single threshold and is jumpy right around the 50 line. Read 'broad' above roughly 55, 'narrow' below roughly 45, and neutral in between, rather than acting on a single tick over or under 50.
• A line up near the 80 band means most tracked sectors sit above their 50-day MA — broad participation, the move is widely shared. A line down near the 20 band means few do — a narrow tape carried by a handful of leaders.
• Its most useful job is as a divergence FLAG: when the index is rising but the breadth line is falling (or vice versa), the headline and the internals disagree — flag it and look closer. Such divergences resolve in BOTH directions and on no fixed clock.
• When index direction and breadth move together, the panel is simply confirming that the move is broadly supported; no divergence to investigate.
• Watch the shape over time, not a single dot: a breadth line drifting down while price holds up is the pattern this panel is built to surface.

Each session we take the daily return of every sector in the universe and compute the standard deviation across that cross-section — one number per day capturing how widely the sectors disagreed. That raw series is smoothed to damp single-day noise, and the smoothed value is what the panel reports. Alongside the current value we draw a sparkline of its recent path and compute a percentile rank of where today's reading sits against the tool's own history, bucketed into narrow, normal, or wide. The percentile is self-referential — it compares this universe to its own past, not to any external benchmark or absolute level.

• Headline value is the smoothed std-dev of sector returns for the session — higher means sectors are decoupling, lower means they are moving as a bloc.
• Percentile bucket is the read to act on, not the raw number: 'wide' means dispersion is high versus its own history (more room for selection to separate), 'narrow' means low (a single macro factor dominates and rotation adds little), 'normal' is the unremarkable middle.
• Use the sparkline for direction of the regime: spread climbing toward 'wide' signals sectors are starting to trade on their own drivers; spread collapsing toward 'narrow' signals correlation rising and the universe consolidating onto one factor.
• Read 'wide' with caution, not enthusiasm: empirically, high dispersion clusters at market transitions and has tended to PRECEDE WEAKER subsequent relative-strength payoffs and elevated crash risk (Stivers & Sun) — it is not a green light for rotation. Because it widens the gap between right and wrong calls symmetrically, it argues for keeping holdings equal-weight rather than concentrating.
• Because the rank is against its own history, a 'wide' tag is relative — it does not assert dispersion is high in any absolute or cross-asset sense, only high for this universe.