Position sizing is the decision of how much capital to commit to a trade — how many shares or contracts to hold — given an account size, a per-trade risk tolerance, and a stop distance. It is the bridge between an idea (this setup has an edge) and a risk: the entry signal tells you whether and where to trade, while position sizing alone determines how much you win or lose when you are right or wrong. Its core tension runs through every method in this section — bigger size compounds capital faster but increases the depth of drawdowns and the probability of being wiped out, while smaller size guarantees survival at the cost of growth. Every sizing rule is a different point on that growth-versus-survival trade-off, and choosing one is choosing how aggressively to push that dial.

What this section covers

This is the how-much half of risk management, distinct from the where (stops, covered in the Technical-Analysis and Risk-Management branches) and the whether (the entry edge itself). The two are mechanically linked: the canonical position-sizing identity is

Shares = Dollar Risk / Per-Share Risk = (Equity × Risk%) / (Entry − Stop)

so size falls out of three inputs — account equity, the fraction of equity you accept losing, and the stop distance. A tight stop permits a large position at the same risk; a wide stop forces a small one. The most common beginner error in the whole domain is confusing percent risked (a fraction of equity measured on the stop) with percent allocated (a fraction of capital deployed) — they are unrelated numbers (TradeZella; Chart Guys). Everything in this section is a variation on how to choose the inputs to that identity, or on the deeper question of what fraction is optimal rather than merely safe.

When it matters — and when it doesn't

Position sizing matters most precisely where edges are thin and uncertain, which is most of real trading. A celebrated (and somewhat folkloric) claim from Van Tharp is that position sizing, not the entry system, is what determines whether a trader meets their objectives — with figures of "60%" or even "90%+ of performance variability" attributed to it (Van Tharp Institute; turtletrader.com). Those specific percentages should be treated as emphasis, not measurement: the original high figures derive from re-readings of asset-allocation studies (Brinson-style) rather than a clean experiment on discretionary traders, and the exact number is disputed. The qualitative claim, however — that two traders with the same signals and different sizing get wildly different equity curves — is mathematically certain and uncontroversial. Sizing matters less when positions are tiny relative to capital (the survival math is slack either way) and when an edge is genuinely large and stable (rare). It matters most under leverage, with correlated simultaneous positions, and in fat-tailed / gapping instruments, where the realized loss can exceed the planned per-trade risk and naive sizing breaks.

Map of the sub-topics

This section organizes sizing methods from the simplest and most robust to the theoretically optimal but fragile:

• Fixed-Fractional Sizing — the canonical retail "1% / 2% rule": risk a constant fraction of current equity on every trade. Edge-blind by design, but trivial to compute, auto-scaling (size grows with the account, shrinks in drawdowns), and the most-recommended default precisely because it is robust to not knowing your edge. Covers the fixed-ratio (Ryan Jones) variant. This is the practical baseline the rest of the section is measured against.

• Volatility-Based Sizing — sets share count inversely to an instrument's volatility (almost always via ATR) so every position carries roughly equal dollar risk — a risk-parity idea at the single-trade level. The dominant use is normalizing a watchlist of different-volatility names. Scales up to inverse-vol weighting, Equal Risk Contribution, and volatility targeting at the portfolio level. Like fixed-fractional, it sizes for risk parity, not edge.

• Kelly Criterion (with children) — the edge-weighted alternative: the fraction that maximizes long-run geometric growth, f* = W − (1−W)/R. This is where sizing stops being safety-first and tries to be growth-optimal. Its sub-nodes are the honest caveats that make it usable:

- Full vs Fractional Kelly — why almost no professional bets full Kelly (violent drawdowns, estimation fragility) and why ¼–½ Kelly is the standard haircut (half Kelly ≈ 75% of the growth at roughly half the volatility). - Estimating Edge & Odds — the binding constraint: Kelly assumes W and R are known; in trading they are noisy, overfit-prone sample estimates, and the plug-in estimator is biased toward overbetting.

• Risk of Ruin — the survival counterweight to Kelly. From the classical gambler's-ruin problem: a positive-edge system can still bankrupt you if each bet is too large a fraction of capital. RoR falls exponentially as per-trade risk shrinks, which is the quantitative backbone of the 0.5–2% rule. Kelly sets the growth ceiling; RoR sets the conservative floor most traders should actually operate at.

The through-line: fixed-fractional and volatility-based sizing are edge-agnostic defenses (they cap how wrong any one trade can go); Kelly is the edge-weighted offense (it tries to maximize growth); and Risk of Ruin is the constraint that explains why the defensive, low-fraction rules survive and why full Kelly does not.

Sources

• Van Tharp Institute — Definitive Guide to Position Sizing; "Position sizing, not the trading system, is key to achieving your objectives" (the 60% / 90%+ figures are Tharp's emphasis, contested as precise measurement): vantharpinstitute.com
• TurtleTrader — Van Tharp profile (origin and contested standing of the "% of performance variability" claim): turtletrader.com/trader-tharp
• TradeZella — Position Size Calculator (Position Size = Dollar Risk ÷ Risk Per Share; worked example): tradezella.com/blog/position-size-calculator
• Chart Guys — Position Sizing: Risk Management for Traders (percent-risked vs percent-allocated; stop distance drives share count): chartguys.com/articles/position-sizing
• Child nodes (this section) — Fixed-Fractional Sizing, Volatility-Based Sizing, Kelly Criterion (Full vs Fractional Kelly; Estimating Edge & Odds), Risk of Ruin — for formulas, drawdown statistics, and failure modes.