A 90% win rate on a naked put — what the probability doesn't tell you
One of the things I hoped for in this beta was that testers would find uses of the tools I hadn't spelled out myself. This week one of them did — and it turned into a good illustration of why a high probability of profit, on its own, can be quietly dangerous.
The workflow a tester found
A beta tester was hunting for naked-put entries using Forecast by Options (the implied-probability tool). The steps:
Pick a ticker and an expiry
Set the target to Probability, with a lower threshold (he used 85%)
Turn backtest validation on
Read the implied-probability curve across strikes, and choose a strike sitting at a comfortable distance in the left tail (he anchored around the 15% line)
It's a disciplined way to enter: instead of guessing a strike, you let the options market's own implied distribution tell you how far out you are. I liked it enough to want to share it.
But it also surfaces the limit of choosing a strike on probability alone — so here's the second half of the workflow I'd add.
Probability is not the whole story
The implied-probability curve tells you how likely the spot is to reach a strike. It says nothing about how much you lose if it does. For a naked put, that downside is open-ended — the loss can run to many times the premium you collect. A strike that looks safe at "85% it won't get there" can still carry a heavy left tail.
So I took a representative trade through Trade Doctor: a short AMD put near the money, ~10 days out, take-profit 30%.
Here's what comes back:
POP: 90% — looks great
Expected P&L: +$6.93 — positive
IV Rank 92%, liquidity good, no earnings before expiry — all green
And then the line that changes the picture:
Loss profile: open-ended risk — average loss ~$33, worst-5% tail ~$74 — roughly 3× the credit collected
What that means
A 90% probability of profit is sitting on a tail that, when it bites, gives back about three times the premium you took in. You win often and small; you lose rarely and large. That's not automatically a bad trade — plenty of people sell premium in high IV on purpose — but it's a profile you want to see before you size it, not discover afterward.
This is the same lesson I keep coming back to: a win rate tells you how often, not how much. On a defined-risk spread the damage is capped. On a naked put it isn't, and the only honest way to judge the trade is to put a number on the tail. That's what the new Loss profile does — it shows the average loss and the worst-5% (CVaR) side by side, so the gap between them tells you how fat the tail is. When they're close, the loss is contained. When the tail is several times the average — as here — the downside is the thing to respect.
Two tools, one decision
The point isn't that Forecast by Options is wrong and Trade Doctor is right. They answer different questions, and they're meant to be used together: Forecast finds you a strike; Trade Doctor tells you what that strike risks. Probability picks the entry; the loss profile sizes the position. Used in sequence, you get a trade you've actually stress-tested rather than one that merely looked good on a single number.
Try it
This is all live in the OptionBench beta — free through the summer. If you sell premium, the Loss profile on naked and uncovered positions is the piece I'd most want your eyes on. And as always: tell me where I'm wrong. The best feedback in this beta has come from testers using the tools in ways I didn't anticipate — this post exists because of one of them.
Launch waitlist: https://www.optionbench.com