Relative RotationBeta

Each point is one security against the S&P 500: horizontal is relative strength, vertical is the momentum of that relative strength, both centred on 100. The trail is the last few periods, so direction matters as much as position — the usual rotation runs clockwise, Leading → Weakening → Lagging → Improving → Leading.

This is an approximation, not the licensed JdK formula. RS-Ratio and RS-Momentum are proprietary to RRG Research; this uses the published approximation — smoothed relative strength, z-scored onto a 100-centred scale. Quadrants and rotation behave correctly, but the numbers will not tie out against a Bloomberg or Optuma RRG.

Sector points are the eleven Select Sector SPDRs — real funds, not baskets. The SPDRs are exactly the eleven GICS sectors, liquid, properly index-weighted and tradeable, so the sector view plots the funds themselves rather than an average of our own making. The Indian board has to build equal-weighted baskets only because no such fund exists there. ETF groups are the one place a basket is still needed: no single fund covers “fixed income” or “country & region”, so those points are equal-weighted aggregates of their members.

ETFs sit on the same scale as stocks, and that is sound rather than convenient. Each symbol is z-scored over its own history, so a bubble’s position never depends on what else is plotted beside it — putting a T-bill fund on the same chart as semiconductors distorts neither. The groups exist because 503 stocks and 112 funds on one canvas is unreadable, and because a group is the unit you actually think in: bonds against equities, defensives against cyclicals.

US splits need no repair; spin-offs need a judgement. Measured across eight known splits — NVDA 10:1, AAPL 4:1, AMZN 20:1, GOOGL 20:1, TSLA 3:1, CMG 50:1, WMT 3:1, MSTR 10:1 — the close-to-close move across every ex-date is an ordinary day, so Yahoo’s US series is already adjusted and correctly dated. (Its Indian series is neither, which is why that board carries a whole repair layer this one does without.) A spin-off is the real question, and it is one of degree: Yahoo prices it in as a fractional adjustment whose ratio is the size of the separation. Above 20% of the parent removed, the earlier bars belong to a materially larger company, so usable history starts at the event and the name reads NA until it has enough of its own — clearing itself as the window rolls. Below it, the surviving business still dominates its own path and the back-adjusted series is kept. DuPont (58% to Qnity), Dell (49% to VMware), Exelon, AT&T, Western Digital, Fortive, Flex, GE and BDX fall above the line; 3M, Jacobs, Danaher, Honeywell, Comcast and IBM fall below it.

A large move is not evidence of a corporate action, so none is ever inferred from price. Moderna rose 177% in one session on 19 August 2026 — by size indistinguishable from a reverse split, and entirely real: the first positive Phase 3 for a personalised mRNA cancer vaccine. Smoothing it would have deleted the largest true signal on the board. Unexplained gaps are reported in the build log and left untouched. The one exception is a separation that has happened but which Yahoo has not yet published — Corteva’s seed spin-off as Vylor on 1 October 2026 — and that is entered by hand only against the company’s own filing, never against a number.

Large moves that are not on the reviewed corporate-action list are left strictly alone. F&O stocks have no daily price band, so a 25–30% day is legal and often real — Adani Enterprises in Feb 2023, IndusInd in Mar 2025, IEX in Jul 2025 — and smoothing those away would erase exactly the risk this board exists to show.

Every symbol is z-scored over an identical window. That matters more than it sounds: the same stock on the same day lands in a different spot if normalised over a different amount of history, so anything without enough history is excluded rather than rescaled.

The 3D view puts a third measure on the depth axis. RRG has no natural third axis — RS-Momentum is already the derivative of RS-Ratio, so the two are a matched pair and anything on Z is a different measure bolted on. Two are offered, each with one reading per tail point so the trail moves in depth too rather than implying the measure held still.

Volatility — trailing realised volatility, annualised. It answers what the flat plot cannot: how much risk sits behind a given relative strength. Its reference plane is the back of the box, since volatility is a magnitude with no natural zero.

Absolute return — the plain return over the stretch the tail covers, and the more pointed of the two. RRG is purely relative: both its axes are measured against the benchmark, so a security can sit deep in Leading while losing money, simply falling more slowly than the index. This axis puts that back on the chart. Its plane sits at zero, so the plane is break-even, a sphere in front of it made money, one behind it lost money, and the stem's direction is the sign. Size is the size of the move, depth is its direction. Sphere size is driven by volatility directly, not by perspective alone, so bigger = more volatile still holds after you rotate the chart; larger spheres are drawn more translucent so the volatile names in front don't hide the quiet ones behind them. Depth is still harder to read than a flat position, which is why this is a toggle beside the 2D board and not a replacement for it — the quadrant call should be made on the 2D plot.

RRG is a relative-strength tool, not a standalone buy/sell signal.