Day 32: Position Sizing Optimization
Topic: Kelly Criterion and dynamic position sizing by confidence
Date: 2026-03-14
Status: β
COMPLETED
Research Question
Whatβs the optimal position size? Should we size based on confidence/entry price?
Key Metrics (416 trades)
| Metric | Value |
|---|---|
| Win Rate | 72.1% |
| Avg Win | $10.87 |
| Avg Loss | $1.87 |
| Reward:Risk | 5.8:1 |
Kelly Criterion Analysis
Formula: f* = (p Γ R - q) / R
| Strategy | Kelly % | Recommendation |
|---|---|---|
| Full Kelly | 67.3% | TOO AGGRESSIVE |
| Half-Kelly | 33.7% | Aggressive but viable |
| Conservative | 20% | Recommended cap |
Insight: With 5.8:1 reward:risk and 72% win rate, the edge is massive. Current $10 fixed sizing is extremely conservative.
Dynamic Sizing by Entry Price
Backtested tiered sizing: - $0.30-$0.40 entry: 2.0x size (high confidence) - $0.40-$0.50 entry: 1.5x size (medium) - $0.50+ entry: 1.0x size (low confidence)
Results
| Strategy | Total PnL | vs Fixed |
|---|---|---|
| Fixed $10 | $3,042 | baseline |
| Dynamic Sizing | $4,856 | +59.6% |
Tier Breakdown
| Entry Price | Trades | Multiplier | PnL Contribution |
|---|---|---|---|
| $0.30-$0.40 | 35 | 2.0x | $904 (30%) |
| $0.40-$0.50 | 198 | 1.5x | $1,791 (59%) |
| $0.50+ | 159 | 1.0x | $633 (21%) |
Recommendation
DEPLOYABLE β Add entry-price-based position sizing: - 2.0x for entries under $0.40 - 1.5x for entries $0.40-$0.50
- 1.0x for entries $0.50+
Expected improvement: +60% on same trade set
Risk Considerations
- Current system is in dry-run (no real money)
- Kelly suggests we could size even larger
- Consider drawdown limits before full deployment