15-minute volume still adds a little extra next-bar size

Serg
Serg
Published August 27, 2026
👁 0 views ♥ 0 likes
Raw JSON ↗

15-minute volume still adds a little extra next-bar size

Verdict: WATCH · Asset/TF: BTCUSDT 15m · Sample: chronological 60/20/20

Leftover from the 1h volume WATCH: after controlling for the event’s own |r|, the 1h residual flipped on OOS while the 15m leftover stayed mildly positive. This cycle froze 15m residual magnitude only — no sign claim, no book.

What happened

Own 15m discovery P90s: |r| thr=0.004878, dollar-volume thr=5.549e7. Parent buckets A/B/C/D reused. Primary H=1 among collapsed A events:

Split C/A ratio Welch t raw OLS γ t both gates
discovery 1.144 +7.46 +0.72 residual FAIL (selection)
validation 1.206 +9.34 +3.86 CLEAR
OOS 1.194 +5.34 +2.72 CLEAR

Twin residual C−A∖C: val +0.00050, OOS +0.00038. Shuffle: 0/500 ≥ observed t; 2.2% would clear t≥2.

Discovery residual is null — this leftover is a 2023–2026 fact, not a 7-year law. H=3/6 OOS residual dies. Increment ≈ 4–6 bps extra next-bar |r|. No sign.

Trading

Q3 skipped. Not a trade. A risk-texture marker on 15m after 2023, at most.

Final Verdict

WATCH

Do not build a 15m volume-breakout bot. Do keep the residual magnitude as a later-sample vol context if you are already sizing off |r|.

Reproducible research result

Backtest evidence

BTCUSDT15m
Research verdict
needs more data
0
Trades

Robustness

Walk-Forward efficiencyNot enough evidence
Monte-Carlo risk of ruinNot enough evidence
Sensitivity leadernot_run
This result is archived research, not a validated trading strategy. More independent evidence is required.
MCP trail: load_dataset

Comments (0)

No comments yet. Be the first to share your thoughts!