Fade or momentum? The data decides, regime by regime

Every ratio in the Spread Lab universe (102 with enough history). When a ratio is stretched (Spread Lab z-score at +1 or beyond = rich, -1 or below = cheap), did it keep going (momentum) or snap back (fade) over the next 21 sessions? Each regime window is judged against its own base - all days in that window - so a window's drift is never mistaken for an edge. Windows: All (2019+) (from 2019) · Nov-21 (2021-11-01) · Apr-23 (2023-04-01) · Oct-30-23 (2023-10-30) · Election-24 (2024-11-05) · Tariffs-25 (2025-04-02) · War-26 (2026-03-01).

How to read it: negative = stretches faded more often than normal; positive = momentum. A per-ratio call needs at least 10 episodes, 5+ points over base and p < 0.1. With one episode per 21 sessions a short window (the war) cannot reach 10 for a single ratio - the family pool is the regime read.

FADEall ratios, War-26: -11.2 pts (n564)

25ratios stretched today

6regime-consistent candidates (2 with n ≥ 10)

0admissible with a positive Blend

1 · Regime map: momentum edge vs base, by family and window

Family	All (2019+)	Nov-21	Apr-23	Oct-30-23	Election-24	Tariffs-25	War-26	Your start
All ratios	-1.6n5083	-1.8n3423	-3.7 FADEn2507	-5.2 FADEn2163	-4.3 FADEn1498	-6.4 FADEn1233	-11.2 FADEn564	-
Breadth & Index	-4.4 FADEn445	-5.0n275	-8.2 FADEn214	-16.2 FADEn175	-8.4 FADEn127	-6.1n106	-7.5n57	-
Crypto & MSTR	-8.1n76	-3.6n58	-0.2n42	-3.0n40	-1.2n24	-2.7n20	-17.7n6	-
International	-0.3n252	+4.6n169	+6.4n110	+3.9n98	+1.8n73	-0.2n57	-16.3n25	-
Macro / Futures	-0.1n233	-3.9n150	-4.9n114	-5.2n102	-7.9n68	-13.7 FADEn55	-10.0n27	-
Metals & Miners	-3.2n597	-5.4 FADEn422	-6.7 FADEn320	-8.3 FADEn291	-6.4 FADEn202	-8.4 FADEn170	-22.9 FADEn75	-
Sectors	+1.9n1437	+1.5n969	-0.5n690	-0.6n588	-0.8n406	-3.6n331	-6.5n153	-
Single stocks	+0.8n195	-0.9n140	-3.9n101	-5.0n87	-10.2n63	-7.6n52	-10.7n21	-
Style / Factor	+1.0n1306	+1.0n872	-2.6n647	-3.9 FADEn557	-1.9n380	-5.6 FADEn319	-11.9 FADEn150	-
Thematic / AI	-2.9n164	+1.5n107	+2.1n72	+1.5n61	+0.5n43	+1.1n33	+3.1n18	-
Volatility & Correlation	-18.7 FADEn378	-18.7 FADEn261	-16.8 FADEn197	-18.0 FADEn164	-17.3 FADEn112	-18.7 FADEn90	-13.6n32	-

Pooled across every stretch event of every ratio in the family (both sides). Pooled episodes are NOT independent - near-duplicate ratios stretch on the same days - so read the n and the p-values as optimistic: a direction, not a sized edge.

2 · Today's candidates: stretched, the family has a direction, and the ratio's own record agrees

Ratio	Stretch	Call	Trade	Evidence window	Win vs base	Median	p	Kelly / Brier / Blend	Own record, recent windows	Family
MTUM/USMV Momentum vs Min-Vol	cheap -1.23	FADE	long MTUM / short USMV	Election-24 · n7 (below 10: thin)	71% vs 62% (+9)	+3.95%	1.0	5.1 / 3.4 / 4.2%	Oct -2 Election -9 Tariffs -19	Style / Factor FADE (War-26 -11.9, pooled n150)
AIWIN/AIRISK AI Winners vs AI At-Risk (GS de-grossing)	rich +1.87	FADE	long AIRISK / short AIWIN	Oct-30-23 · n8 (below 10: thin)	75% vs 63% (+12)	+2.89%	0.719	3.5 / 3.6 / 3.6%	Oct -12 Election -12 Tariffs +4	ALL FADE (War-26 -11.2, pooled n564)
NQ=F/CL=F Nasdaq fut vs Crude oil	cheap -1.65	FADE	long NQ=F / short CL=F	Election-24 · n11	64% vs 53% (+11)	+1.23%	0.556	0.0 / 1.8 / 0.0%	Oct -4 Election -11 Tariffs -5 War +2	Macro / Futures FADE (Tariffs-25 -13.7, pooled n55)
META/QQQ META vs QQQ	cheap -1.36	FADE	long META / short QQQ	Oct-30-23 · n10	60% vs 52% (+8)	+2.02%	0.755	0.0 / 2.1 / 0.0%	Oct -8 Election -12 Tariffs -14	ALL FADE (War-26 -11.2, pooled n564)
GDX/GDXJ GDX vs GDXJ	rich +2.19	FADE	long GDXJ / short GDX	Oct-30-23 · n8 (below 10: thin)	88% vs 55% (+32)	+1.37%	0.082	0.0 / 4.7 / 0.0%	Oct -32 Election -45 Tariffs -43	Metals & Miners FADE (War-26 -22.9, pooled n75)
SPMO/SPY SPMO vs SPY	cheap -1.23	FADE	long SPMO / short SPY	Oct-30-23 · n7 (below 10: thin)	71% vs 63% (+9)	+2.04%	1.0	0.0 / 1.7 / 0.0%	Oct -9 Election -6 Tariffs -10	Style / Factor FADE (War-26 -11.9, pooled n150)

Inverse pairs and index twins are collapsed to one line. Faded rows have fewer than 10 episodes - below the house minimum. Blend = house Kelly + adapted Brier on the evidence window (both must see an edge); small median ratio moves usually size to 0 even with high win rates. HELD markers flag desk ratio pairs held the same or the opposite way.

Ask the data

Try: SPY/XLU above 1.85 z GLD/QQQ below -1.5 z, 10 days, since 2023 SPY/RSP crosses above 1.5 z since the election

Adjust: horizon sessions · from · events

Plain English in, an exact test out: the box only translates your words (any ratio in the Spread Lab universe; z-score or ratio level; above / below / crosses; a horizon; a start) and shows what it will run - every number is computed from the data. Results open in the pop-out with the chart, spaghetti, every event and the PDF button.

3 · Explorer: pick any starting point, click any ratio to pop out its chart, spaghetti and every event

custom start show unstretched ratios too

From -: events, win vs base, median and verdict are recomputed in your browser for every ratio, on TODAY's stretch side, with the house rules (n ≥ 10, 5+ pts, p < 0.1). The regime map's last column follows the same start.

Ratio	Family	Today	Side tested	Events	Momentum win vs base	Edge (pts)	Median signed	p	Verdict
loading...

Ratio since the start, every event marked

Spaghetti: the ratio's path after each event (% from the event close)

One thin line per event in your window - green = it kept going (momentum), red = it snapped back (fade). Thick line = median of the events; dashed = any day in the same window (the base). Rich and cheap events are split because 'up' means opposite things for them.

Every event and its outcome

Event date	Stretch	zc	Ratio	+5	+10	+21	Best / worst inside 21	Outcome

Outcomes are ratio moves from the event close. Best / worst are measured in the stretch's direction (momentum). Markers on the ratio chart: triangle-down = rich, triangle-up = cheap. Close with ×, Esc or a click outside.

4 · Every stretched ratio: momentum edge by window (today's stretch side)

Ratio	Family	Stretch	All (2019+)	Nov-21	Apr-23	Oct-30-23	Election-24	Tariffs-25	War-26	Verdict	Candidate filter
XMMO/MDY	Style / Factor	cheap -2.54	+7n27	+10n15	+18n9	+15n9	+17n8	+26n7	-	NO EDGE	own recent record does not agree with Style / Factor FADE
GDX/GDXJ	Metals & Miners	rich +2.19	-28n18	-34n11	-33n8	-32n8	-45n6	-43n6	-	FADE	candidate
AIWIN/AIRISK	Thematic / AI	rich +1.87	-7n23	-10n11	-8n10	-12n8	-12n7	+4n5	-	NO EDGE	candidate
XLE/TAN	Sectors	rich +1.74	+16n22	+14n15	+17n8	-4n6	+8n5	-	-	NO EDGE	own recent record does not agree with ALL FADE
NQ=F/CL=F	Macro / Futures	cheap -1.65	-10n16	-15n13	-6n11	-4n11	-11n11	-5n10	+2n6	NO EDGE	candidate
XMMO/MTUM	Style / Factor	cheap -1.65	-2n32	-6n22	-1n18	-7n17	-4n12	-3n9	-	NO EDGE	own recent record does not agree with Style / Factor FADE
SPY/XLV	Sectors	cheap -1.61	+5n27	+6n22	+12n13	+10n13	+5n11	+2n10	-	NO EDGE	own recent record does not agree with ALL FADE
QUAL/USMV	Style / Factor	cheap -1.55	+2n22	-2n21	-6n10	+0n8	+2n5	-	-	NO EDGE	own recent record does not agree with Style / Factor FADE
META/QQQ	Single stocks	cheap -1.36	+13n25	+8n19	-4n10	-8n10	-12n10	-14n10	-	NO EDGE	candidate
SPYV/SPMO	Style / Factor	rich +1.35	+8n28	+2n17	+1n12	+2n10	+9n8	-6n8	-	NO EDGE	own recent record does not agree with Style / Factor FADE
SPMO/SPY	Style / Factor	cheap -1.23	+1n28	-4n16	+2n10	-9n7	-6n6	-10n6	-	NO EDGE	candidate
GC=F/^DJI	Metals & Miners	cheap -1.23	+27n18	+30n9	+26n8	+28n8	+29n7	+32n6	+7n6	MOMENTUM	own recent record does not agree with Metals & Miners FADE
MTUM/USMV	Style / Factor	cheap -1.23	-8n24	-6n18	-4n9	-2n9	-9n7	-19n6	-	NO EDGE	candidate
EEM/SPY	International	cheap -1.22	-3n19	+12n9	-	-	-	-	-	NO EDGE	own recent record does not agree with ALL FADE
SPY/XLI	Sectors	rich +1.21	+2n23	+6n16	+8n15	+8n12	+11n9	+2n8	-	NO EDGE	own recent record does not agree with ALL FADE
MTUM/SPYV	Style / Factor	cheap -1.20	+8n25	+2n16	-9n9	-3n8	+1n7	-12n6	-	NO EDGE	own recent record does not agree with Style / Factor FADE
AIRISK/SMH	Thematic / AI	cheap -1.19	+6n24	+14n14	+13n12	+15n11	+6n9	-7n7	-	NO EDGE	own recent record does not agree with ALL FADE
XLI/SPY	Sectors	cheap -1.18	+2n23	+6n16	+8n15	+8n12	+11n9	+2n8	-	NO EDGE	duplicate of a stronger-stretched twin
SPY/XLE	Sectors	cheap -1.17	+8n27	+9n17	+17n8	+17n8	+11n8	+20n7	+9n5	NO EDGE	own recent record does not agree with ALL FADE
SPYV/SPY	Style / Factor	rich +1.14	+13n30	+15n19	+7n10	+6n10	+18n8	+13n7	-	NO EDGE	own recent record does not agree with Style / Factor FADE
MTUM/RSP	Style / Factor	cheap -1.10	+0n23	+5n14	-3n8	-2n8	+2n7	-10n6	-	NO EDGE	own recent record does not agree with Style / Factor FADE
GC=F/^GSPC	Metals & Miners	cheap -1.09	+2n24	+2n12	-1n12	-4n11	+6n8	+4n7	-12n6	NO EDGE	own recent record does not agree with Metals & Miners FADE
KBWB/XLF	Sectors	cheap -1.07	+9n20	+13n14	+10n9	-	-	-	-	NO EDGE	own recent record does not agree with ALL FADE
IWF/IWD	Style / Factor	cheap -1.02	+2n29	+5n19	+2n11	-3n11	+4n9	+2n8	+4n5	NO EDGE	own recent record does not agree with Style / Factor FADE
USMV/SPY	Style / Factor	rich +1.02	+13n21	+10n20	+8n9	+21n7	+21n5	-	-	NO EDGE	own recent record does not agree with Style / Factor FADE

Data: Spread Lab daily closes (Yahoo, from 2018) through 2026-10-28. Only events whose 21-session window has fully elapsed are counted. Educational research, not advice.