Last March I put 40% of my Kalshi balance on a single Fed decision contract. I was confident. The market had Powell at 78% to hold rates, and I thought it should be closer to 92%. I was right about the direction. The Fed held. But that position size meant I spent the entire FOMC week checking my phone every twenty minutes, unable to think about anything else. The trade made money. The process was garbage. That's when I started taking position sizing seriously.
Most new traders on prediction markets obsess over finding the right side of a contract. They spend hours reading polls, studying economic data, watching cable news. Then they throw some arbitrary amount at it. Maybe $50 because it feels small. Maybe $500 because they're excited. This is backwards.
Position sizing on Kalshi isn't just risk management. It's the difference between being able to trade next month and blowing up your account on one bad call. Binary contracts have a brutal feature: you can be directionally correct about probability and still lose everything you bet. A 70% favorite loses 30% of the time. That's not rare. That's almost one in three.
The math is unforgiving. If you bet 50% of your bankroll on a series of 60% edges, you will eventually go broke. Guaranteed. The only question is how many trades it takes.
Here's where I started, and it's a decent baseline for anyone new to prediction markets:
This sounds conservative. It is. But conservative is how you stay in the game long enough to actually develop skill. I've watched people in the Telegram channel I run blow through their entire deposit in a week because they kept doubling down on "sure things."
There are no sure things. Even when a contract is trading at 95 cents, you're paying a premium that limits your upside. And the 5% of the time it goes against you, it goes to zero.
If you want to get more precise about position sizing on Kalshi, you need to know about the Kelly Criterion. It's a formula developed in the 1950s for optimal bet sizing when you have an edge.
The basic formula is:
Kelly % = (bp - q) / b
Where:
Let's say the KXCPIYOY market has a contract trading at 35 cents, and you think the true probability is 50%. Your Kelly bet would be: (1.86 × 0.50 - 0.50) / 1.86 = 0.215, or about 21% of your bankroll.
That's way too aggressive for most people. Which is why almost everyone who uses Kelly uses "fractional Kelly," typically half or quarter Kelly. So that 21% becomes 5-10%.
Full Kelly assumes you know your exact edge. You don't. I spent years on a CME equity index desk with access to actual data and sophisticated models, and we still got probabilities wrong constantly. On Kalshi, you're estimating based on polls, vibes, and whatever you read on Twitter. Your edge estimate is noisy. Quarter Kelly accounts for that uncertainty.
Not all Kalshi markets deserve the same sizing. Here's how I think about different categories:
These resolve based on objective numbers. CPI prints, rate decisions, jobs reports. I'm willing to size up slightly here (maybe 6-8% of bankroll) because the resolution is clean and my background gives me some edge in interpreting Fed communication. Markets like KXFEDDECISION contracts tend to be liquid and well-priced.
KXPRESPARTY and election markets are where I'm most conservative. Polling error is real. Turnout models are guesswork. I've seen people confidently size up on candidates and get destroyed by a three-point polling miss. Keep these at 2-4% max unless you genuinely have information the market doesn't.
Temperature markets, hurricane predictions, awards shows. These are fun but the edge is hard to quantify. I treat them as entertainment and size at 1-2%. The goal is to stay engaged and learn the mechanics, not to build a P&L.
Theory is nice. Here's what I actually do with my own money:
Here's something the quant textbooks don't tell you: correct position sizing makes you a better trader because it keeps you calm. When I had 40% of my account on that Fed trade, I couldn't think clearly. I was checking prices constantly. I was reading into every word of every Fed governor speech.
With 5% positions, I can watch a contract move against me and actually evaluate whether my thesis is still valid. I can sleep. I can hold through normal volatility without panic-selling. Position sizing isn't just about math. It's about staying rational enough to use your edge.
Kalshi is regulated by the CFTC and requires KYC, so you're not going to lose your money to a rug pull. But you can absolutely lose it to your own bad sizing decisions. The platform will let you bet your entire balance on one contract. That doesn't mean you should.
Position sizing matters at any account size, but the math gets practical around $500 or more. Below that, a 5% position is $25, which limits your contract selection. You can still practice good habits with smaller amounts, but you'll feel the minimum tick sizes more. Start where you're comfortable losing everything as a learning experience.
Yes, the same principles apply. What matters is your edge and the potential payout, not which side you're taking. A NO position at 60 cents is mathematically equivalent to a YES position at 40 cents on the opposite outcome. Size based on your estimated edge, not the direction of the bet.
You don't, at least not at first. This is why fractional Kelly exists. Track your trades for a few months. If you're consistently profitable, your edge estimates are probably in the right ballpark. If you're losing, either your estimates are wrong or you don't have an edge at all. Most people overestimate their edge. Assume you're one of them.
The core math is similar, but Kalshi contracts have some unique features. You can often exit positions before resolution, which changes risk management. Liquidity varies widely between markets. And multi-day or multi-week contracts create exposure you need to monitor differently than a single game or hand. The principles transfer, but the application needs adjustment.
Not financial advice. I trade my own money and you can lose yours. Do your own research.