Methodology · Crypto · IV.4
The tuned crypto profile
These are the versioned crypto constants (the sign-off centerpiece). They are reasoned priors, not measured optima — exactly the status the equity set carries (II.2) — and a future crypto calibration pass tunes them. Departures from the equity values, with rationale:
Family weights (equity → crypto):
| Timeframe | Trend | Momentum | Volume | Structure |
|---|---|---|---|---|
| Hourly | 0.30 → 0.30 | 0.35 → 0.40 | 0.20 → 0.10 | 0.15 → 0.20 |
| Daily | 0.35 → 0.40 | 0.30 → 0.30 | 0.15 → 0.10 | 0.20 → 0.20 |
| Weekly | 0.45 → 0.50 | 0.20 → 0.20 | 0.10 → 0.10 | 0.25 → 0.20 |
Rationale: crypto is momentum- and trend-driven; volume is down-weighted (not up) because crypto volume is unreliable (wash trading, fragmented venues — IV.7), so we lean on price-derived evidence and keep the noisiest input small. Weights still sum to 1.0.
Direction threshold: 0.22 — unchanged. No evidence to move the confluence bar;
keeping it symmetric avoids an un-earned claim.
ATR multipliers — wider, for fatter tails (equity → crypto):
stop.bufferAtr 0.45 → 0.60 (levels get pierced by wicks more often)
stop.floorAtr 0.80 → 1.00 (a wider minimum stop survives 24/7 noise)
stop.capAtr 2.20 → 3.00 (allow a wider max stop given higher vol)
target.projectionAtrMult 2.5 → 3.0 (projection matches larger swings)
target.projectionR 2 → 2 (unchanged)
target.structureMinR 1.4 → 1.4 (unchanged)
Rationale: crypto's realized volatility and wick depth are proportionally larger; wider buffers/caps cut premature stop-outs while the ATR denomination keeps everything self-scaling per coin.
Confidence: base 35, slope 75, ceiling 95, floor 5, dissentThreshold 0.15 —
unchanged. Penalties: choppy 12, dissent 8, rsiExtreme 8, thin 5 — unchanged;
the earnings penalty is inert (no earnings). Buckets highMin 70, moderateMin 45
— unchanged.
Gates: minConfidence 35, minRR 1.2, evMargin 0.25 — unchanged (risk
discipline is asset-agnostic).
The crypto profile also sets the earnings-veto window to none for every timeframe (IV.5) and the fractional sizing fields in IV.6.