Results9 Oct 2026cvd_failed_sweeps_results.md

H2 — Failed Liquidity Sweeps — Results (exploratory, 2026 in-sample)

universe (10): BTCUSDT, ETHUSDT, SOLUSDT, ZECUSDT, NEARUSDT, XRPUSDT, OGNUSDT, HYPEUSDT, BNBUSDT, DOGEUSDT
timeframe 1h · horizon 4h · lookback 48 · cooldown 24
sweep events (design+test): 1262

Notes

Exploratory: single regime (2026), in-sample; n=10 perp symbols.
Design IC positive but flipped sign out-of-sample — selection artifact, not an edge.

Primary: event pooled rank IC(af_z, oriented return)

period	IC	n
design	0.0495	508
test	-0.0022	754

test IC 95% bootstrap CI: [-0.0807, 0.0654]

Horizon consistency (test event IC)

horizon	IC
1h	-0.0067
4h	-0.0022
24h	0.0068
3d	0.0063

Test quintiles (af_z → oriented return)

quintile	n	mean	median
0	151	-0.00027	-0.00000
1	151	0.00019	0.00148
2	150	0.00034	0.00086
3	151	-0.00069	0.00083
4	151	0.00002	0.00127

Matched control (high-af − low-af sweeps)

spread: -31.96 bps (CI [-125.36, 46.00])
n: 754

Control regression (af_z coefficient)

coef: 0.00002 (CI [-0.00070, 0.00080])

Per-symbol test event IC

symbol	IC	n
BTCUSDT	0.1518	80
ETHUSDT	-0.1259	75
SOLUSDT	0.0698	75
ZECUSDT	-0.1721	63
NEARUSDT	-0.1032	83
XRPUSDT	-0.0459	83
OGNUSDT	0.0646	65
HYPEUSDT	0.0861	85
BNBUSDT	-0.1347	74
DOGEUSDT	0.1634	71

Robustness sweep (test event IC)

timeframe	lookback	flow_window	design_ic	test_ic	n_events
15m	24	1	-0.0445	0.0175	6485
15m	24	4	-0.0235	-0.0194	6485
15m	48	1	0.0072	0.0095	4993
15m	48	4	-0.0157	0.0221	4993
15m	96	1	-0.0035	-0.0053	3605
15m	96	4	-0.0145	0.0131	3605
1h	24	1	-0.0047	0.0148	1683
1h	24	4	-0.0790	0.0198	1683
1h	48	1	0.0495	-0.0022	1262
1h	48	4	0.0098	-0.0227	1262
1h	96	1	0.0525	-0.0488	857
1h	96	4	-0.0112	-0.0566	857
4h	24	1	-0.0347	-0.0927	399
4h	24	4	-0.0051	-0.0523	399
4h	48	1	0.1024	0.0256	306
4h	48	4	0.1596	-0.0495	306
4h	96	1	0.2299	-0.0163	221
4h	96	4	0.1530	-0.0860	221

Verdict: NO EDGE (H0)

#	criterion	pass
Q1		❌
Q2		❌
Q3		❌
Q4		❌
Q5		❌
Q6		❌
Q7		❌

← All reports · source (.md)