← Start here · Step 7 of 8 · 9 min
Risk and position size
Why you can lose everything at a 60% win rate, what share of the bankroll to stake, and the two habits of survivors.
Bankroll
Your bankroll is the sum you set aside for prediction markets whose loss changes nothing in your life. Not "temporarily free money" but money already written off in your head. If you do not have such a sum, the markets can wait. Everything below is in percent of bankroll, not dollars.
Why 60% wins do not save you
Suppose you are right 60% of the time at a price of 0.50, an excellent result, better than most. If you stake 25% of the bankroll per trade, a run of four losses, which happens about once per 40 trades, wipes you out. A run of ten losses at a 60% win rate happens about once per 10,000 trades, so almost surely within an active year. Variance beats skill when stake size is unbounded.
How much to stake
A working rule for beginners: no more than 2% of bankroll per position and no more than 10% in total on one story, such as one match or one election, because related markets lose together. With a $500 bankroll that is $10 per position. Yes, it is boring. Boredom is the protection: at 2% per stake, ten losses in a row cost 18% of the bankroll, not all of it.
The Kelly formula you will read about online says to stake a fraction equal to your edge over the price. The problem is that beginners do not know their edge and almost always overestimate it. Half or quarter Kelly is for people with six months of records.
Correlation
"Yes" on a team winning, "Over 2.5 goals" and "Win with handicap" are three bets on the same thing. If the team plays badly, all three lose. Count such positions as one.
What not to do
- Double after a loss. Not a strategy, the fastest way to zero.
- Bet on what "cannot fail". We checked 6,951 markets: favourites at 0.90–0.99 win exactly as often as the price says. The expected return of "sure things" is zero, and the rare loss costs the whole stake.
- Trade the 5-minute crypto markets. You face programs with millisecond latency; we measured it with 32 strategies.
- Copy blindly. Even the best wallets lose 30–40% of their markets. Copy style and discipline, not every bet.
The two habits of survivors
A journal. Date, market, price, amount, why you bought, how it ended. After 50 entries you see where you earn and where you lose. Without a journal memory keeps only the wins.
A pause after a streak. Three losses in a row: a day without trades. Decisions after a loss are worse; that is not weakness, it is physiology.
Next: how to use this site to learn from the consistent and avoid repeating others' mistakes.