Risk Management

Position Sizing: Why to Bet More When Winning (Not When Losing)

By Mario Maldonado · Read time: 9 min · Leer en Español


Most traders have their position sizing logic completely backwards. They size bigger after losses — trying to recover — and get cautious after wins — afraid to give back gains. Both impulses are psychologically natural and financially counterproductive. The mathematics and optimal psychology point in the exact opposite direction.

Fixed Fractional Position Sizing: The Foundation

Always risk the same percentage of your current account, not the same dollar amount. Fixed dollar risk ($200/trade) means your relative exposure grows as your account shrinks. Fixed fractional (2%) automatically scales down during drawdowns and up during growth — self-regulating in exactly the direction you need.

Fixed Fractional Position Size
Shares = (Account × Risk%) / (Entry Price − Stop Price)

Example: $20,000 × 1.5% / ($10.00 − $9.50) = $300 / $0.50 = 600 shares

Kelly Criterion: The Mathematical Optimal Bet

Full Kelly Criterion
f* = W − (1 − W) / R
W = win rate, R = average win / average loss ratio
Half-Kelly (Recommended)
f_half = (W − (1 − W) / R) / 2

Example: 40% win rate, 2:1 average R:R:
Full Kelly = 10%  |  Half-Kelly = 5%  |  Quarter-Kelly = 2.5%
In live trading with commissions, slippage, and imprecise edge estimates, most professionals operate at Quarter-Kelly or below. The 2% Rule sits comfortably within this range.

The Anti-Martingale System

Martingale — doubling after each loss — requires infinite capital to work mathematically. Anti-Martingale inverts the logic: increase size on winning streaks, decrease on losing streaks.

ScenarioMartingale ResultAnti-Martingale Result
5-trade winning streakFixed/smaller exposureIncreasing exposure — maximizes the streak
5-trade losing streakIncreasing exposure → ruinDecreasing exposure — limits the damage

Why Martingale Destroys Accounts

Trader with $10,000, initial $200 risk on Martingale:

  • Loss 1: $200 → Loss 2: $400 → Loss 3: $800 → Loss 4: $1,600 → Loss 5: $3,200
  • Total lost: $6,200 (62%) in 5 consecutive losses

Five consecutive losses are normal variance for any 40–60% win rate system. With Martingale, normal variance destroys 62% of the account. With Anti-Martingale, the same streak automatically reduces sizing — damage is contained and survivable.

Key Point: Winning and losing streaks in trading show more persistence than random statistics predict. Anti-Martingale captures positive momentum and protects against negative momentum — aligned with how markets actually behave.

Maximum Concentration Rules

  • Single position: Maximum 25% of account at risk under any circumstance
  • Correlated positions: No more than 40% of capital simultaneously
  • Small accounts (<$25k): Maximum 1 position at a time
  • Hot streaks: Account up 20%+ quickly — drop to 1% minimum for 5 trades to prevent irresponsible "house money" sizing
Practical Anti-Martingale on $20,000
Base sizing: 1.5% = $300 per trade
After 3 consecutive wins: increase to 2.0% = $400
After 5 consecutive wins: increase to 2.5% = $500 (cap)

First loss: return to 1.5%
After 2 consecutive losses: drop to 1.0%
After 3+ consecutive losses: drop to 0.5% and review system

This structure captures positive momentum while containing damage during negative streaks.

Related Reading

Risk/Reward Ratio: Lose 70% and Be Profitable   Mental vs Hard Stop Loss   FOMO in Trading