Dynamic Position Sizing Algorithm Development
Dynamic position sizing adapts each position size based on current market conditions and portfolio state. Instead of fixed lot trading, algorithm calculates optimal size for each specific trade.
Fixed Fractional (Kelly-inspired)
Base approach: risk fixed capital percentage per trade.
def fixed_fractional_size(capital, risk_pct, entry_price, stop_price):
risk_amount = capital * risk_pct
risk_per_unit = abs(entry_price - stop_price)
qty = risk_amount / risk_per_unit
return qty
Standard risk_pct: 1–2% per trade. On 20 consecutive full-stop losses: loss (0.98)^20 = 33% deposit. Manageable.
Volatility-Adjusted Sizing
Position size inversely proportional to volatility: higher volatility → smaller position.
def volatility_adjusted_size(capital, target_risk_pct, atr, entry_price, atr_multiplier=2.0):
risk_amount = capital * target_risk_pct
stop_distance = atr * atr_multiplier # stop at 2×ATR
position_value = risk_amount / (stop_distance / entry_price)
return position_value / entry_price # in asset units
ATR 3% → stop 6% → size X. ATR 1% → stop 2% → size 3X. Result: equal monetary risk at different volatility.
Kelly Criterion
Mathematically optimal position size for maximizing long-term capital growth:
Kelly % = W - (1-W)/R
where W = win rate, R = avg win/avg loss
W=55%, R=1.5: Kelly = 0.55 - 0.45/1.5 = 0.25 = 25% capital.
Too aggressive. Usually Half Kelly (12.5%) or Quarter Kelly. Full Kelly causes huge drawdown despite theoretical optimality.
Drawdown-Based Scaling
Reduce positions approaching max drawdown:
def drawdown_scaled_size(base_size, current_equity, peak_equity,
max_drawdown=0.20):
current_dd = (peak_equity - current_equity) / peak_equity
if current_dd > max_drawdown * 0.75:
# At 75% max drawdown — reduce to 50% size
return base_size * 0.5
elif current_dd > max_drawdown * 0.5:
# At 50% max drawdown — reduce to 75% size
return base_size * 0.75
return base_size
Correlation Adjustment
Portfolio already holds correlated positions — new adds less diversification. New position size reduces proportionally:
def correlation_adjusted_size(base_size, correlation_with_portfolio):
# Correlation 0.8 → reduce to 20% of base
diversity_factor = 1 - abs(correlation_with_portfolio)
return base_size * max(diversity_factor, 0.2) # minimum 20%
Develop dynamic position sizing system with configurable method combination: fixed fractional + volatility adjustment + drawdown scaling + correlation adjustment. Single module, integrates any trading strategy.