How to Size a Position

The simplest and most robust framework: risk a fixed percentage of account on each trade. Typical values: 0.5% for conservative, 1% for standard, 2% for aggressive. The math:

stop_distance = |entry − stop_loss|
position_usd = account × (risk_pct / 100) × (entry / stop_distance)

To lose exactly risk_pct of the account when the stop hits, hold a position scaled by the inverse of stop-distance-as-a-fraction-of-entry. Tight stop → big position; wide stop → small position. The dollar amount at risk stays fixed.

Example. $10,000 account, 1% risk per trade, $50,000 BTC entry, $48,000 stop. Stop distance = 4%. Risk dollars = $100. Position = $100 / 0.04 = $2,500. If BTC drops to your stop, you lose $100 — exactly 1% of the account.

Why this works in the long run: 100 consecutive losing trades at 1% risk per trade compound to ~63% drawdown — bad, but recoverable. Same 100 losses at 5% risk compound to a 99.4% drawdown — unrecoverable. Position sizing determines the survivability of any losing streak, not just the average performance.

The position size calculator implements this directly. Input account, risk %, entry, and stop loss; get the right size.

Risk-percent sizing keeps dollar-risk constant per trade. Kelly criterion sizes by the edge of the trade: bigger positions when edge is bigger.

f* = (b × p − q) / b
  p = probability of winning
  q = 1 − p
  b = avg_win / avg_loss

The output is the fraction of bankroll to bet on each trade. In practice, full Kelly is almost always too aggressive — it assumes you know p and b exactly, which you don't. Half-Kelly captures most of the growth at a fraction of the variance; quarter-Kelly is even safer. See /learn/kelly-criterion.

Best-practice combination: use risk-percent as the default sizing rule (1% per trade is a good starting point), with half-Kelly as a ceiling. Whichever number is smaller wins. This catches cases where edge is uncertain (Kelly would be a low number anyway) without leaving signal-aware sizing on the table when edge is clear.

The third framework operates at the portfolio level rather than per-trade. Volatility-targeted sizing adjusts position sizes inversely to recent realized volatility: bigger positions when an asset is quiet, smaller when it's swinging.

The goal: keep the portfolio's overall volatility steady regardless of market regime. In low-vol periods, the strategy puts more capital to work because the same dollar-position carries less risk. In high-vol periods, it pulls back to keep risk bounded.

Common in trend-following systems. The Keel pipeline supports inverse-ATR sizing directly — a position sized as notional = target_vol / current_atr automatically reduces during volatility spikes and expands during calm. The trend following template uses this pattern.

Trade-off: vol-targeting can mute returns in strong directional regimes (you size down right when the trend is most profitable). It's a smoothing technique — better Sharpe, lower variance, sometimes lower absolute returns. Most professional trend systems use it anyway because consistent returns are easier to hold through than lumpy ones.

For a real-world strategy, the typical stack is:

1. Set a base risk-per-trade (0.5-1%).
2. Use the Kelly criterion to set a ceiling (half-Kelly max, regardless of risk-percent suggestion).
3. Apply a portfolio-level vol-target as a final scaling factor (reduce total exposure during high-vol regimes).

The three layers serve different purposes: risk-percent for per-trade survival, Kelly for edge-awareness, vol-target for portfolio-level smoothness. None replaces the others; all three matter.