Sentivue/Research/Position Sizing

Research

The R-Multiple Framework: Position Sizing for Systematic Traders

The R-multiple framework expresses every trade in units of risk, not dollars. It is the cleanest position-sizing primitive for systematic strategies — and the one most often misused by retail.

Sentivue Capital··7 min read

Most retail trader losses don't come from bad signals. They come from bad position sizing applied to acceptable signals.

The R-multiple framework is the cleanest position-sizing primitive in systematic trading. Originally formalized by Van Tharp, the framework expresses every trade in units of risk — "R" — rather than dollars or percentages of capital.

The basics

For each trade:

  1. Define risk per trade (R). This is the dollar amount you would lose if the stop-loss were hit. R is a trade-level parameter, not a fixed dollar amount across all trades.
  2. Position-size to R. If you're risking $1,000 per trade and the stop is 2% away, position size is $50,000 ($1,000 / 2%).
  3. Express results in R. A trade that gains 3× the risked amount is "+3R." A trade that hits the stop is "−1R."

The result: every trade, regardless of instrument or volatility, is comparable on a unified scale.

Why R-multiple beats fixed-percentage sizing

A "risk 2% of capital per trade" rule sounds clean but ignores volatility differences across instruments. Trading SPY with a 2% stop is very different from trading GME with a 2% stop. R-multiple normalizes for this — the dollar risk is constant, the position size adjusts to instrument volatility.

Combining with Kelly

R-multiple sizing is compatible with Kelly-style aggressiveness. Set R per trade as a fraction of capital that, given the strategy's documented edge, the Kelly criterion approves of. For most strategies with realistic edge estimates, this lands at 0.5–1% of capital per trade — well below half-Kelly.

Common misuses

  1. R that's too large. "Risking 5% per trade" is mathematically equivalent to running an 80% Kelly fraction for most edge profiles. The strategy's realized drawdown will be worse than the math suggests because real edge estimates carry significant standard error.

  2. R that drifts with account size. Many retail traders nominally use R-multiple sizing but quietly increase R when the account grows and don't decrease it when it shrinks. The result is asymmetric: you size up after winners (which captures positive variance) and don't size down after losers (which doesn't protect from drawdowns).

  3. Treating R as a target, not a maximum. R is the amount you would lose if the stop hit. It is not a target. The actual realized loss should average meaningfully less than R because some stops fill before the strict stop level (good) and a few fill worse (bad — slippage). The average realized loss in well-executed systematic strategies is typically 0.7–0.9R.

Compatibility with portfolio risk

R-multiple sizing handles per-trade risk. It does not handle correlated-position risk. Two trades each at 1R of risk in highly correlated instruments are functionally a 2R-correlated position. Portfolio-level risk overlay is required:

  • Cap total open R across the book.
  • Cap correlated-cluster R.
  • Reduce R when portfolio drawdown breaches a threshold ("anti-martingale" sizing).

Practical takeaways

  • R is a per-trade risk unit, not a dollar amount. It rescales for instrument volatility automatically.
  • Conservative R is the strongest risk control most retail systematic strategies are missing.
  • R-multiple sizing without a portfolio overlay only handles half the problem.

Related