Position sizing without blowing up

October 14, 2026·8 min read

Picture two investors. Same stock picks. Same starting money. One ends up rich. The other ends up broke. What made the difference? Not the picks. The size of the bets. Most investors spend hours choosing what to buy. They spend almost no time deciding how much to buy. That order is backwards. How much you risk decides whether a good idea builds wealth. It also decides whether a run of bad luck wipes you out. This guide gives a plain framework for that decision. It covers three ideas: the average outcome, the sizing rule, and the drops you must live through.

Start With Expected Value

Check the average outcome first. Every investment is a bet with a range of possible outcomes. The average outcome (expected value) weighs each outcome by its chance of happening. A big gain with a small chance counts less than a small gain with a big chance. If the average outcome is not positive, the right size is zero. No sizing method fixes a bet that loses money on average.

Nobody buys a stock because the math looks good. They buy because the story sounds good. Sound familiar? The story is the trap. A great story with a negative average outcome is still a bad bet. Run the numbers before you fall in love with the pitch.

The formula is simple. Multiply the chance of winning by the gain, then subtract the chance of losing times the loss. In symbols, EV = p(win) × gain − p(loss) × loss. Suppose a stock has a 60% chance to gain $100 and a 40% chance to lose $100. The average outcome is $20. That is 60% of $100 minus 40% of $100.

Try a trickier case. Suppose a stock has a 30% chance to gain $300 and a 70% chance to lose $100. A 30% chance sounds weak. The math disagrees. The average outcome is $20. That is 30% of $300 minus 70% of $100. The average outcome sees what the story hides.

Think of a coin flip for money. Heads you gain $100. Tails you lose $100. The coin is fair, so the average outcome is zero. The right size for a zero average outcome is zero.

Try a second case. Suppose a stock has a 40% chance to gain $100 and a 60% chance to lose $100. The average outcome is negative $20. That is 40% of $100 minus 60% of $100. The right size is zero, even if the stock feels like a sure thing.

Zero means zero. Not a small bet. Not a toe in the water. Skip the idea.

The average outcome says nothing about the ride. A bet with a positive average outcome can still lose most of the time. It can still drop hard before it pays. The average tells you the destination. It hides the road.

The math is easy. The inputs are hard. Your estimate of a 60% chance is a guess. Guesses run overconfident. Before you trust your gut, check the track record of similar ideas (base rate). Suppose you study a company for a week, read the filings, listen to the call, and feel 80% sure the stock rises. Then you check the record. Similar picks by investors like you worked 55% of the time. Your gut says 80%. History says 55%. Trust history. Move your estimate toward 55% before you size. Your gut is one data point. The track record is many. A track record beats a hunch because it counts many cases, not one feeling.

The Kelly Fraction

So the bet pays on average. Now what? How much do you put down?

Once a bet has a positive average outcome, you need a size. A sizing rule (Kelly criterion) gives the bet size that grows your money fastest over the long run. Growth builds on itself, so small gains add up over years. It assumes you have a real advantage. The rule is simple. The catch is that you should bet less than it says.

The formula is f* = (p(b+1) − 1) / b. Here p is your chance of winning. Here b is your profit per dollar staked (net odds). Take a 60% chance of winning at even money. Even money means a bet that pays $1 for each $1 staked. The formula gives f* = 20%. That means you bet 20% of your investing money. Your investing money means all the money in that account, not one trade.

Walk through the example. At even money, b is 1, so b + 1 is 2. The math is (0.6 × 2 − 1) / 1, which is 0.2. That is 20%. The formula turns your guess into a size.

Try a stronger case. Suppose you have a 70% chance of winning at even money. The math is (0.7 × 2 − 1) / 1, which is 0.4. That is 40%. The rule says bet 40% of your investing money. Feel how aggressive that sounds. That feeling is the point. The formula has no fear. You do. That gap between the formula and your nerves is where most sizing mistakes live.

The rule aims at the long run, and it assumes you place many bets over many years. One trade proves nothing. The math needs repetition to work. A single bet can go either way no matter how good the sizing is.

The formula is exact. The inputs are guesses. Professionals almost never bet the full Kelly amount. Full Kelly assumes you know p and b exactly. You do not. Errors in your guesses are normal. A small error in p or b becomes a large error in size. If you bet twice what the rule says, your growth rate falls to zero.

Suppose the true chance is 50%, but you guess 60%. The formula gives 20% from your guess. The right size at a 50% chance is zero. A 10-point error turns a zero-size bet into a 20% bet. That is why professionals shrink the full amount.

Standard practice is half-Kelly or quarter-Kelly. Half-Kelly means half the full amount. Quarter-Kelly means a quarter of the full amount. Suppose the full amount is 20%. Half-Kelly is 10%. Quarter-Kelly is 5%. You keep most of the growth with a fraction of the swings (volatility). Smaller bets also make the drops easier to live through. Half-Kelly is the professional default. Quarter-Kelly is for people who want to sleep at night.

Try it yourself with the Kelly position-size calculator. It shows full, half, and quarter Kelly side by side. It also warns about drops from a high point (drawdown). Enter your own numbers. Compare the three sizes. Watch the drawdown warning. Then decide before you risk money.

The Drawdown Reality

Even the perfect size comes with painful drops. A full-Kelly bettor should expect to lose half their money at some point. The formula does not warn you about how that feels. A drop from a high point is normal at that size. Drops of that size happen again and again, not once.

A 50% drop feels like failure. Your stomach says the sizing was wrong. Your stomach is wrong. The drop was part of the plan. Plans do not feel good at the bottom.

The real danger is not the drop. It is what you do during the drop. Many investors sell at the bottom, convinced the drop proves they were wrong, while others double down to win it back fast. Both reactions turn a planned drop into real damage. The sizing plan only works if you hold it through the ugly part.

Suppose your account peaks at $100,000. Suppose it then falls to $50,000. That $50,000 fall is a drawdown. The loss is real even when the sizing is right.

Half-Kelly cuts the drop roughly in half while giving up only a quarter of the growth, and that trade is the best deal in investing. You trade a little growth for a much smoother ride. Almost nobody takes it because it feels too conservative. It feels like leaving money on the table. It is. A little. On purpose.

Three rules of thumb survive contact with real markets. First, never risk more than you can lose without changing your behavior. A loss that changes your behavior leads to worse decisions. You sell low. You chase the loss. You quit early. Second, size every position as if your advantage is half what you think. An edge is a guess, and guesses run high. Cutting it in half is cheap insurance. Third, 10 positions at 2% each beat one 20% bet when the positions move independently (uncorrelated). One bad idea cannot sink the account when it is only 2% of your money.

Spreading money across positions (diversification) is the quiet partner of position sizing. It works with position size to keep one bad idea from sinking the portfolio. Suppose one stock falls 50%. At 2% of your money, the damage is 1% of your account. At 20%, the damage is 10%. Same fall. 10 times the pain. When one position falls, the other nine hold you up. That is why size and spread work together.