What is position sizing, and why does it decide everything?

Dan was right about the stock. His thesis, that a beaten-down logistics company was due a recovery, played out almost exactly as he wrote it in his journal, and if you check the chart today the vindication is printed on it. Dan did not profit from being right, though, because being sure, he had put a third of his account into the position, and three weeks before the recovery arrived, a routine dip, the kind this stock produced several times a year, took his oversized stake down far enough that he sold to stop the bleeding. The market then rose without him. His experience is so common it might as well be printed on the account-opening forms. He made a good call and an unsurvivable bet on it, and the bet decided.

That is the whole subject of position sizing, in short. Every beginner believes trading skill lives in the picking: which stock, which direction, when. Position sizing is the unglamorous decision that comes after the picking, how much, and the case this article makes, with arithmetic, a sixty-year-old proof, and one regulator’s intervention, is that the how much decides more than the which. Not because entries do not matter, but because size is the term that multiplies everything else. Size sets the price of being wrong, and being wrong is routine.

What position sizing actually is

Strip the vocabulary and position sizing is one decision: of the money in your account, how much is exposed to this single idea. Not how much you could buy, which is what a buying-power number invites you to think, but how much you are choosing to put at risk of loss, and the difference between those framings is the difference between sizing by appetite and sizing by policy.

Sized by policy, the decision runs backward from the loss. A trader decides, first, the dollars they will pay if this specific idea turns out to be wrong; second, where the market would have to go for wrong to be established; and only then, dividing one by the other, how many shares or contracts that permits. The count of shares falls out at the end, as a consequence. Sized by appetite, the process runs forward from confidence: this one feels strong, so buy a lot. Dan sized by appetite. The market billed him for the difference.

Notice what the backward version quietly requires: a written answer to what would make this trade wrong, before entry. Sizing and exits are welded together, which is why they belong to one discipline, risk, rather than two topics. How to choose that concession point well, and how professionals turn the whole loop into a repeatable system, is genuinely deep craft; the concept, though, fits in the sentence you just read, and the concept is what this question was really asking.

The arithmetic that makes size decide

Why does the how-much outrank the which? Start with the least negotiable fact in trading, which is not psychological at all. Losses and gains are not symmetric. Lose 10% and you need about 11% to get back to even. The hole grows faster than the ladder.

Now add the second fact: for any trader with rules, losing trades arrive constantly and in streaks, the way tails arrive in coin flips, even when the strategy is sound. Sizing is what determines whether a normal streak is an expense or an ending. The trader risking a small fixed fraction per trade meets a five-loss streak and pays a single-digit percentage of the account, annoying and survivable. The trader with a third of the account on each idea can be functionally finished by two bad calls, and, per the bars above, mathematically crippled by them even if skill returns immediately. Same market, same streak, same skill. The size chose which trader kept existing, which is what I mean by size deciding everything: at the wrong size, none of your other decisions get enough repetitions to matter.

The 1956 proof that caution compounds faster

The deep version of this argument was written before most of today’s markets existed, by a Bell Labs physicist thinking about gambling on noisy information. J. L. Kelly’s 1956 paper, A New Interpretation of Information Rate, examined a bettor with a genuine, persistent edge and asked how much of his capital he should stake each round. The startling part is what Kelly proved about the boldest answer. Betting everything each time actually maximizes the expected value of your wealth, on paper, and yet the bettor who does it, in Kelly’s words, “would probably be broke and, in fact, would be broke with probability one if he continued indefinitely.” Kelly’s alternative arrives in the very next sentence, and it is the whole of this article’s subject: assume instead that he bets a fixed fraction of his capital each time.

Sit with that sentence, because it is the mathematics of trading in one line. The strategy that looks best by expected value guarantees ruin, certainty, not risk, for anyone who keeps playing. Growth that compounds belongs exclusively to fraction bettors, and Kelly derived the exact fraction, set by the size of your edge, at which capital grows fastest. Above that fraction, and this is the part every oversized trader is living out, more aggression produces less long-run growth, sliding toward guaranteed loss. There is such a thing as mathematically too big, it arrives long before all-in, and it applies to people with real edges. For people without one, and the measured record says that describes most active traders for most of their careers, every size is too big except the one small enough to survive the education.

When traders would not size down, a regulator did it for them

If sizing were merely advice, it would not appear in securities law. It does. In 2018, the European Securities and Markets Authority did something regulators almost never do: it capped the size of retail positions directly, by limiting the leverage brokers may offer on contracts for difference, the high-leverage instruments through which much of Europe trades. The evidence behind the intervention was the analyses of national regulators showing that “74-89% of retail accounts typically lose money on their investments, with average losses per client ranging from €1,600 to €29,000.” The caps scale by the wildness of the underlying market: 30:1 on major currency pairs, 10:1 on most commodities, 5:1 on individual stocks, 2:1 on crypto assets, joined by a rule that closes positions when margin runs low and a guarantee that a client cannot lose more than the account holds. ESMA’s chair at the time, Steven Maijoor, summarized the purpose: “The new measures on CFDs will for the first time ensure that investors cannot lose more money than they put in, restrict the use of leverage and incentives, and provide a risk warning for investors.”

Read the design and you realize what it is: position sizing, imposed from above. A leverage cap is a ceiling on how large a position a given account can carry; a margin close-out is a stop on the account itself. An entire continent’s market regulator looked at retail outcomes and concluded the decisive variable was not what people traded but how big, relative to their money, they were allowed to trade it. That is the same conclusion Kelly reached from information theory and Dan reached from experience, arriving this time with the force of law, and it is worth noticing that no regulator anywhere has ever needed to cap how good your entries are.

The policy that replaces conviction

Two refinements complete the concept, and both follow from what sizing is for. The first is that markets are not equally wild, so a fixed dollar exposure is not a fixed risk: a position in something that routinely swings five percent a day carries several times the risk of the same dollars in something that drifts, which is why practitioners scale positions down as an instrument’s typical movement scales up. The measuring of typical movement has its own tools, but the principle needs none of them: wilder gets smaller, or the risk per trade is not actually fixed. The second refinement is that positions that move together are, for sizing purposes, one position. Three energy stocks sized separately at a careful fraction each are a single triple-sized energy bet wearing three tickers, and the day the sector falls, it falls on all of them at once. A sizing policy that counts positions but not their overlap has a loophole exactly where the streaks come from.

All of which returns to the practical shape of the thing. Position sizing, done as the professionals this section keeps citing do it, is a policy, a small set of written numbers that exist before any particular trade shows up: the fraction of the account one idea may risk, and the streak the account must be able to absorb. The policy’s entire job is to be immune to how tonight’s setup feels, because how the setup feels is precisely the input that sized Dan’s position. Conviction is real and sometimes even correct. It is also the most expensive sizing algorithm ever deployed by retail traders, because it concentrates money on exactly the trades where being wrong was unthinkable, and unthinkable is not a probability. Writing that policy in units that hold steady across every market you touch is what Chapter 4 of The Complete Trader, thinking in R, teaches you to do.

Run Dan’s trade again under a policy. Same thesis, same entry, same routine dip, but the position is a twentieth of the account instead of a third. The dip now costs something he has already priced, his concession point is set by his written exit rather than by nausea, and he is still holding, or calmly re-entering, when the recovery he correctly predicted arrives. Nothing about his analysis improved. The size stopped punishing him for the market’s ordinary weather, which is all sizing ever does: it keeps you solvent and sane enough for your other decisions to get their repetitions. That is the entire, unexciting reason it decides everything.