Sizing: the decision that decides whether you're still here

Strike selection, timing, and volatility decide how often you win. Sizing decides whether you're still trading after the losses that all three of those things guarantee. It's the least discussed input and the only one that can end the account.

Risk is the notional, not the premium

A short $45 put collects $120 and obligates you to buy $4,500 of stock. The premium is the reward; the notional is the risk. Five contracts is $22,500 of obligation. If the broker's margin requirement is $620 per contract, five of them "cost" $3,100, and that's the number people size by. Assignment ignores it. Size by what you'd be forced to buy.

For defined-risk trades, the maximum loss is the number: a $5-wide credit spread sold for $1.20 risks $380 per spread, and ten of them risk $3,800. That's defined, and it's still $3,800.

The arithmetic of streaks

An 80% win rate sounds like four wins for every loss, evenly spaced. It isn't. Across 100 trades at 80%, back-to-back losses are almost certain and three in a row show up close to half the time. The losses are also correlated: in a market drop, the put that loses isn't one put, it's every put you have open. A one-in-five event across eight positions at once is one event, not eight.

Work it backward. Decide the biggest drawdown you could sit through without changing how you trade; for most people that's somewhere between 10% and 20% of the account. Then assume three max-loss outcomes land in the same week and size so that hurts without reaching that line. If the answer is fewer contracts than you wanted, that's the answer.

A working set of limits

Starting points, not rules:

• Any single defined-risk position's maximum loss: no more than 2 to 5% of the account.
• Any single undefined-risk position's notional: no more than 10 to 15% of the account, so that assignment is affordable and survivable.
• Total notional across every short put: an amount you could actually own, in cash, if all of them were assigned in the same week.
• Correlated names count as one position. Five semiconductor puts are one semiconductor put five times as large.

What sizing shouldn't respond to

Confidence. The trade you're surest about is the one where you most want to size up, and the certainty is a feeling, not information. Recent wins. A good month tempts larger size at exactly the point in the cycle where volatility is lowest and the premium thinnest. Available margin. Buying power is the broker's estimate of what you can carry today, and it shrinks in the same moments the positions grow.

What it should respond to

Account size, mechanically: the same percentage as the account grows or shrinks. Volatility: bigger expected moves mean fewer contracts for the same risk. The number of open positions: each new one raises the total, and the total is what a bad week tests. The single most useful habit is to write down, before every trade, the dollar loss if the stock fell 30% overnight, and to keep a running total of that number across everything that's open.