Most people who lose money on prediction markets were not wrong about the world. They were wrong about size.
That’s the part nobody covers, because it doesn’t sell. Reading prices is interesting. Spotting a mispriced market feels clever. Deciding how much to put on it feels like admin. But the size of your position decides whether a normal run of bad luck is an annoying week or the end of your account, and no amount of forecasting skill rescues a number that was too big to begin with.
The price already told you how often you lose
Here’s the thing that took me embarrassingly long to notice.
When you buy Yes at 65¢, the market is telling you something more useful than “probably.” It’s telling you that on questions priced like this one, the answer comes out No about a third of the time. That’s not pessimism. It’s the definition of the price, straight out of the price-reading note.
So the market hands you the losing frequency in advance, on every single trade, and almost nobody reads it that way. People see 65¢ and feel reassured. What they’ve actually been handed is a warning label: one in three of these will hurt.
Losing streaks are boringly normal
Suppose you’re good. Genuinely good, right 60% of the time, which would put you ahead of most people who do this.
Four losses in a row still has roughly a one in forty chance of happening at any moment. Five in a row, about one in a hundred. Those look like small numbers until you remember you’ll place many more than a hundred trades, and each new trade is another chance for the run to start. Across a few hundred trades, a streak of five or six is closer to expected than to unlucky.
This matters because of how people react to streaks. A run of losses feels like evidence that you’ve lost your touch, and that feeling arrives at the worst possible time, when your account is already smaller and the urge to make it back is strongest. It usually isn’t evidence of anything. It’s what a 40% loss rate looks like from the inside.
What a normal streak does to an account
Now put a size on it.
Bet half your account on each trade, and five losses in a row leave you with about 3% of what you started with. Not “down a lot.” Effectively gone, from a sequence that a 60% forecaster should expect to hit sooner or later.
The asymmetry underneath is simple arithmetic and it never gets less annoying: lose half your money and you need to double what’s left to get back to even. A 50% loss requires a 100% gain to undo. Losses cost more than the matching gains give you, which means the job of position sizing isn’t to maximize the good months. It’s to make sure the bad ones stay survivable.
The favorite trap
This is the part I’d tattoo on a beginner if they’d let me.
Buy a contract at 90¢ and it feels like the safe trade. It isn’t just less profitable than it looks. Run the numbers: you’re risking 90¢ to make 10¢. You need nine wins to pay for one loss.
At a true probability of 90%, that’s exactly break-even. Nine wins, one loss, zero profit. Then subtract the spread you crossed and the taker fee you paid, both covered in the costs note, and the same trade goes quietly negative.
So the market that feels safest is the one where being slightly wrong about the odds does the most damage, and it’s the one people size largest precisely because it feels certain. If you take one thing from this page, take that.
What sizing can and can’t do
I should be straight about the limits, because this gets oversold.
Sizing does not make you profitable. If your forecasts have no edge, careful sizing means you lose slowly instead of quickly, and slowly is still lose. It buys you time and the chance to find out whether you’re any good. That’s all it does, and that’s plenty.
The other honest problem: small positions are dull. That’s the real reason people abandon this, not ignorance. A $5 trade that goes your way produces almost no feeling at all, and the whole appeal of these markets is the feeling. Anyone telling you sizing is easy is selling something. It’s simple and unsatisfying, which is a harder combination than complicated.
The habits that actually help
No percentages here, because the right number depends on things I don’t know about you. The habits transfer regardless.
Decide your size before you look at the price. Conviction is a terrible input to sizing, because conviction is highest exactly when you’re most likely to be missing something.
Read every price as a loss frequency. 65¢ means one in three. 80¢ means one in five. Say the number out loud before you commit.
Assume the streak. Ask what six losses in a row would do to your account, and if the answer makes you flinch, the size is wrong. Not the market. The size.
Track why, not just what. A losing trade that was right for the right reasons is a different animal from a losing trade you talked yourself into, and after twenty of them you’ll be able to tell which kind you make.
The uncomfortable truth in all of this is that the math cares nothing for how sure you feel. A market at 90¢ will hand you nine wins and let you enjoy every one of them, and the tenth will take it all back with interest. Size for the tenth.
Nothing here is financial advice. Prediction markets involve real risk, so never trade money you can’t afford to lose.