Position Sizing
Position Sizing
Position sizing is the single most consequential decision a premium seller makes, and experienced sellers are blunt about it: how much you trade matters more than what you trade, because over-sizing is the fastest way to destroy an account. The core method is to size by defined maximum loss / buying-power reduction (not by the credit collected), keep each trade to a small fraction of net liquidity, and run many small, uncorrelated occurrences so the law of large numbers — not any single bet — drives the result. Every substantive claim below is labeled by evidence grade, confidence, and nature per the Project Charter.
---
1. Size by Defined Risk / BPR — Never by Credit Collected
The foundational error in sizing is anchoring to the premium received. A $1.00 credit feels identical whether it came from a tight defined-risk spread or a naked strangle, but the risk behind it can differ by an order of magnitude. The rule here is to size every position by what it can actually lose — the defined max loss for spreads, or a realistic tail-loss estimate for undefined-risk trades — and by the buying-power reduction (BPR) the broker assesses, because BPR is what actually constrains the account.
Margin is not maximum loss
For undefined-risk trades, the margin requirement is the collateral the broker holds — typically calibrated to roughly a 2-standard-deviation move (about 95% certainty the loss won't exceed it) — not the worst case. A move beyond 2 SD can lose multiples of the posted margin. The canonical illustration: with Feb crude trading ~63.50, selling the naked 55 put (the ~1 SD strike) carried a margin requirement of about $1,900, while the risk out to the 2 SD strike (49) was roughly $6,000 — over three times the margin. Sizing off the $1,900 number badly understates the real exposure.
The practical test: compare initial margin against your estimated loss at 2 SD. If a single 2 SD adverse move would blow past a tolerable share of net liq, the position is too large regardless of how attractive the credit looks.
Why defined risk makes sizing honest
Defined-risk structures (verticals, iron condors, defined-width strangles) convert sizing into arithmetic: max loss = width − credit, full stop. You can never lose more than you can compute at entry, which is why defined-risk spreads are the default sizing vehicle for smaller accounts.
---
2. Per-Trade Risk and Total Capital Deployment
Per-trade risk: roughly 1–5% of net liq
It helps to frame per-trade risk in terms of how many consecutive losses the account can absorb. If you risk too much per occurrence, a perfectly ordinary losing streak — which high-probability traders will hit — becomes terminal. The widely-taught band is to keep the capital at risk in any single position to a low single-digit percentage of net liquidity (roughly 1–5%), with the larger figure reserved for the smallest, most defined-risk positions and the smaller figure as accounts grow. Risking on the order of 20% per trade is explicitly held up as the way to get wiped out.
A simple consecutive-loss lens makes the point concrete:
Total deployment: capital-allocation %, notional, and the cash buffer
Two portfolio-level gauges bound the aggregate:
- Capital allocation % = buying power used ÷ net liq. The discipline is to deploy only a moderate share and hold the rest as a buffer against buying-power expansion (margin requirements grow in selloffs as IV spikes). Commonly-taught guidance is roughly 25–35% deployed in normal volatility, scaling up toward ~50% when IV is high and premium is rich — because high-IV environments offer more reward per unit of risk, and because deploying from a low-vol, all-time-high market is when BPR expands the most against you.
- Notional value relative to net liq. Because options are leveraged, BPR alone can hide how much underlying you effectively control. A rough orientation for a margin account is total notional value on the order of ~3× net liq — a leverage ceiling that keeps a portfolio from being secretly enormous behind a modest BPR figure.
Conflict to flag. The "deploy more in high IV" guidance and the "keep a large cash buffer" rule are in tension precisely when markets are stressed — exactly when BPR expansion bites hardest. The resolution most active sellers live with: add carefully in high IV, but never deploy so much that a further volatility spike could force liquidation. The cash buffer outranks the urge to harvest rich premium.
---
3. Number of Occurrences and the Law of Large Numbers
The premium-selling edge rests entirely on a probabilistic claim: a high-probability premium sale has a small positive expected value per trade, but the realized win rate converges to the theoretical probability only over a large number of independent occurrences. Over a handful of trades, variance dominates and you can easily lose even with the odds in your favor; over hundreds, the average outcome tightens around expectancy. This is the literal law of large numbers, and it is the reason "trade small, trade often" is the house mantra.
The corollary is decisive for sizing: many small trades beat a few large ones for the same total capital at risk. Small size lets you (a) accumulate enough occurrences for probabilities to play out, and (b) survive the inevitable losing clusters along the way. Large size does the opposite — it caps your occurrence count and raises the chance a normal streak ends the account before expectancy can assert itself.
Load-bearing caveat — occurrences must be independent. The law of large numbers only smooths outcomes if the trades are reasonably uncorrelated. Twenty short-premium positions across twenty highly-correlated mega-cap tech names is one big bet wearing twenty costumes, not twenty occurrences. Maximizing the count of trades helps only to the extent they are genuinely diversified (see §6 and 06_portfolio_management).
---
4. Kelly Criterion: Use a Fraction, Not the Full Bet
The Kelly criterion answers "what fraction of capital maximizes long-run geometric growth?" For a bet at b-to-1 odds with win probability p, the optimal fraction is:
*f\ = (bp − (1 − p)) / b** = edge ÷ odds.
Kelly maximizes the expected log of wealth and, asymptotically, the growth rate.
Why traders use a fraction of full Kelly
Full Kelly is mathematically growth-optimal but brutally volatile — it tolerates deep drawdowns on the way to that growth, and it assumes you know your true edge precisely. In trading you do not: win probability and payoff ratio are estimates (from IV, backtests, or models), and overestimating your edge causes Kelly to over-bet, which is far more punishing than under-betting. The standard remedy is fractional Kelly — commonly half-Kelly — which captures roughly three-quarters of the optimal growth rate while sharply cutting drawdown and the risk of ruin.
The intuition dovetails exactly with the 1–5% per-trade band of §2: those small fractions are, in effect, a heavily fractional Kelly bet, deliberately under-sized to absorb estimation error and variance. Over-sizing is the #1 account killer — not bad trade selection — because even a positive-expectancy strategy bet too large will eventually hit the losing streak that ruins it.
Limitation. Kelly assumes a known, stationary edge and independent bets — neither holds cleanly for options portfolios with shifting IV and correlated names. Treat Kelly as intuition for sizing direction (bet more when edge is larger and odds shorter; always bet a fraction), not as a precise dial.
---
5. Scaling With Account Size — Micro Futures for Granularity
Position sizing is relative to net liq, so the same 1–5% rule means different absolute sizes at different account levels. The hard problem is granularity: in a small account, the minimum tradeable size of many products is already too large to keep any one trade under a few percent of net liq. The solutions, in order of preference for small accounts: defined-risk spreads, low-priced underlyings/ETFs, and micro futures.
Micro futures are 1/10th the size of their E-mini/standard counterparts, which lets a small account take genuine futures exposure (and diversify into uncorrelated asset classes) while keeping per-trade risk inside the band:
Because each micro is one-tenth the notional, it both lowers the minimum bite and lets you scale in increments — adding contracts as the account grows rather than jumping from zero to a full-size position. This granularity is precisely why micros are the bridge between a small account and the multi-occurrence, diversified book the strategy assumes.
Caveat. "Micro" reduces size, not the nature of the risk — futures are still leveraged and undefined-risk by default, so the §1 rule (size by realistic 2 SD loss, not by margin) applies with full force.
---
6. Beta-Weighting and Notional Sizing Across Underlyings
Raw deltas and raw contract counts are not comparable across underlyings — one delta of a $600 stock is not one delta of a $40 stock, and a "1-lot" means wildly different dollar risk in different names. To size across a portfolio, the method reduces everything to a common yardstick:
- Beta-weighting to SPY converts each position's delta into the equivalent delta of the benchmark, giving an apples-to-apples read of directional risk per name and for the whole book. This lets you see whether any single underlying is contributing an outsized share of portfolio delta and trim or cap it accordingly.
- Notional sizing bounds per-name exposure by the dollar value controlled, not the premium or lot count. A 50-strike short put controls $5,000 of notional (100 shares × $50); summing notional across names — and keeping any one name a modest slice of the ~3× net-liq total — prevents a single underlying from dominating risk even when its BPR looks small.
The combined discipline: equalize risk per name by beta-weighted delta and notional, rather than by number of contracts or credit. Five lots of a low-beta utility and one lot of a high-beta growth name may carry equal portfolio risk — sizing by contract count would badly misjudge both. This is what turns "many occurrences" (§3) into genuinely diversified occurrences.
---
Key Takeaways
- Size by defined max loss / BPR, never by credit. For undefined-risk trades, margin ≈ a 2 SD move and understates the tail — check estimated loss at 2 SD before entry.
- Keep per-trade risk to ~1–5% of net liq. Risking ~20% per trade is a documented way to get wiped out by an ordinary losing streak.
- Deploy ~25–35% of net liq in normal vol, up to ~50% in high IV, cap notional near ~3× net liq, and always hold a buffer against buying-power expansion.
- Many small, independent occurrences beat a few large ones — the law of large numbers only works with enough uncorrelated trades.
- Use a fraction of full Kelly (e.g., half-Kelly). Over-sizing — not stock-picking — is the #1 account killer; under-betting protects against estimation error and variance.
- Scale with account size; use micro futures (/MES, /MCL, /MGC) at 1/10th size for granularity in small accounts.
- Equalize risk per name with beta-weighted delta and notional, not by contract count or credit.
---
Common Misconceptions
- "A big credit means a good trade." The credit is reward only; identical credits can hide 10× different risk. Size on max loss / 2 SD loss.
- "Margin is my worst case." Margin is ~2 SD collateral, not max loss — a tail move can lose multiples of it (the $1,900 margin vs ~$6,000 2 SD oil example).
- "I should bet full Kelly to grow fastest." Full Kelly's drawdowns are brutal and it assumes you know your edge exactly; fractional Kelly captures most of the growth with far less ruin risk.
- "More tickers automatically means more occurrences." Correlated names are one bet; only uncorrelated trades count toward the law of large numbers.
- "Idle cash is wasted." The cash buffer is the position that survives buying-power expansion and lets you add at high IV.
---
Review Questions
1. Two trades each collect a $1.00 credit. One is a 5-wide defined-risk spread; the other is a naked strangle. Why should you size these completely differently, and what number do you size each off of?
2. For an undefined-risk short put, the broker margin is $1,900 but the loss out to 2 standard deviations is ~$6,000. Which figure should drive your sizing decision, and why is using the other one dangerous?
3. State the per-trade risk band the premium-selling method teaches and explain it through the lens of consecutive losses. Roughly how many straight losers does a 5%-risk trader survive versus a 20%-risk trader before halving the account?
4. Explain why "many small trades beat a few large ones" follows from the law of large numbers, and give the one condition under which a large number of trades still fails to deliver the benefit.
5. Why do traders deliberately use a fraction of full Kelly (e.g., half-Kelly) rather than the growth-optimal full bet? Connect this to the claim that over-sizing is the #1 account killer.
6. A small account can't keep a full-size /ES position under 5% of net liq. Name the micro alternative, state its size relative to /ES, and explain how beta-weighting/notional would let you compare its risk to an equity position in a different name.
---
Related Sections
- 03_implied_volatility — IV regime drives how much capital to deploy (lean light in low IV, scale up in high IV).
- 05_trade_management — managing winners early changes realized risk per occurrence.
- 06_portfolio_management — beta-weighted delta, capital-allocation %, notional, and buying-power expansion at the portfolio level.
- 08_defined_risk — defined-risk structures that make max-loss sizing exact.
- 09_strangles — the canonical undefined-risk trade whose sizing requires the 2 SD check.
- 15_futures_options — futures and micro-futures sizing and margin mechanics.
- 16_small_accounts — granularity, product selection, and scaling from a small base.
- 18_research_findings — the "stay small" and number-of-occurrences studies.
- 19_risk_management — defining and capping max loss, buffers, and tail risk.
- 02_probability — expectancy, the law of large numbers, and independence.
---
Sources
- industry research — Position Sizing | Defined Risk (2015-09-10): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — Position Sizing & Margin (2014-12-10): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — Sizing | Leverage (2015-03-31): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — Evidence for Staying Small (2016-11-03): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — Measures of Portfolio Risk (2017-03-10): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — The Law of Large Numbers: Trade Occurrences (2013-08-28): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — Probability and Number of Occurrences (2018-04-16): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — The Theory Behind Trade Small, Trade Often (2015-12-14): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- industry research — The Kelly Criterion (2021-04-14): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- this approach — Available Futures Products (learn): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- this approach — Beta Weight Your Portfolio to a Symbol (Beta-weighted Deltas) (support): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
- this approach — How Is Net Liq Calculated (support): https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document
_Evidence-labeled per the Project Charter. Education only, not financial advice._