Skip to content

ADR-007: Leverage Tiers & Alpha Stack V2

Status: Approved (2026-04-19)
Context: Paper Trading setup, cost coverage required
Decision: Asset-specific static leverage tiers + Alpha Stack V2 roadmap

Leverage Tiers (Static)

Tier Leverage Assets DD Range Rationale
1 1.8x BTC, ETH 18.3-18.5% Low volatility, deepest liquidity
2 1.5x SOL, DOGE, ADA 17.9-19.9% Medium volatility
3 1.0x AVAX 18.3% High volatility, conservative

Why static, not dynamic:
Dynamic auto-derate (reduce leverage on guard trigger) adds state complexity to the State Manager. For V1, static tiers are robust, predictable, and sufficient. Dynamic leverage is a V2+ consideration.

Why not 3x:
DD at 3x = 29-51%. KILL switch fires constantly. Mathematically untenable.

Why 1.8x not 2x for Tier 1:
BTC/ETH at 2x → DD = 20.2-20.4% → exactly at KILL threshold. 1.8x gives 18.3-18.5% DD — 1.5pp buffer to KILL. Still delivers +40% return.

Economics Impact

Metric Value
Capital deployed 910€ (100€ base × leverage)
Gross monthly 37.4€
VPS cost 10€/mo
Net monthly +27.4€
Net annual +329€

Alpha Stack V2 Roadmap

Implemented (V1)

  • ✅ Asset-specific leverage tiers
  • ✅ Risk Guard states (CL≥5 SOFT_PAUSE, DD>20% KILL)

Next (Paper Trading Validation)

  • ⬜ 2x Leverage Guard-trigger frequency measurement
  • ⬜ Actual vs backtest exit reason comparison
  • ⬜ Trailing stop realism validation (CLOSE vs real-time)

Future (V2+)

  • ⬜ Asset-Ranking: Top-N by 24h ROC when multiple signals fire
  • ⬜ ADX-based position sizing: 0.5x (ADX 20-30), 1.0x (30-40), 1.5x (>40)
  • ⬜ Limit-order engineering (Post-Only + 30s chase)
  • ⬜ News/Macro Risk Layer (ADR-005 V2): KI score 0-100 → position sizing

Explicitly Rejected

  • ❌ ATR-based stops (tested, no improvement over ADX+EMA baseline)
  • ❌ Kelly criterion (requires stable win-rate — unavailable in crypto)
  • ❌ Extra HTF filter (already covered by ADX+EMA regime filter)
  • ❌ 3x leverage (DD 29-51%, constant KILL triggers)

Strategic Review Integration

The trailing stop CLOSE vs LOW analysis (STRATEGIC_REVIEW.md) revealed that backtest results are optimistic by ~69pp for BTC at 2% trailing with LOW-based checks. Paper trading with real-time WebSocket prices will determine the true exit behavior. If paper results significantly underperform backtest: - Increase trailing from 2% → 3.5% - Re-baseline with actual execution data

Configuration

LEVERAGE_TIERS = {
    "BTCUSDT":  1.8,
    "ETHUSDT":  1.8,
    "SOLUSDT":  1.5,
    "DOGEUSDT": 1.5,
    "ADAUSDT":  1.5,
    "AVAXUSDT": 1.0,
}