A Kalshi contract at 97¢ that settles in three days is not really a bet. It is a short-duration bond: you pay 97, you collect 100, unless it defaults. Three percent in three days, which annualizes to a number so large it stops meaning anything.
Which raises the only question that matters, and the one nobody asks: do 97¢ markets actually settle yes more than 97% of the time? Because if they settle 95% of the time, buying them is a slow, confident way to lose money, and it will feel completely safe the entire time you are doing it. Kalshi does not publish that number. We measured it.
A heavy favourite is not a bond
This is the distinction the whole page turns on. A bond is a market the world has already decided but Kalshi has not settled: the game is over, the index closed above the strike, and it sits at 97¢ because there is nothing left to find out. You are paid for settlement lag, not for risk. “Indiana wins by over 6.5 points”, fourteen days out, at 86¢ is not that. It is a bet, with a fortnight of real outcome risk. And our settled record says markets at 80-89¢ go on to settle yes only about 85% of the time, so buying one at 86¢ is negative expected value. Calling it yield would be the exact trap this page exists to expose. So we keep those off the desk. What is left is markets that are genuinely decided, or close to it, and priced accordingly.
The break-even is just the price
There is no clever maths here, which is what makes it so easy to miss. A contract at 97¢ needs a 97% settle rate to break even. At 92¢ you need 92%. Every cent of the yield you are being paid is compensation for a default risk that you cannot see and Kalshi does not quote. The fee then eats into it: even a market that settles at exactly its price loses you money once the taker fee is paid.
Pricing every favourite as a bond…
How the ratings work
ContractTax rates every contract on the Bond Desk the way a credit agency rates a bond, from AAA down to B, using three measured inputs: how decided the outcome looks (price level, time to settlement, and whether the price has been swinging), the measured default rate (how often settled markets in the same category and price band actually paid, from our archive of settled Kalshi markets), and whether the yield covers the risk (net expected value per contract after Kalshi’s taker fee).
AAA is settlement lag dressed as a market: near-certain, stable, days from paying, and history at that price pays what it promises. AA and A carry real but small outcome risk, or a duration haircut. BBB is the lowest investment grade: the measured settle rate only just covers the price. BB and B are wagers wearing a bond’s price. NR means we do not hold enough settled history to grade it, and we would rather say so than guess. Ratings are recomputed every few minutes from live prices and our settled record; a rating can fall between visits, exactly like a downgrade.
What this is and is not
The measured settle rate is an average across a price band, not a forecast for any single market in front of you. A specific 97¢ contract might be a lock or a landmine, and the band tells you nothing about which. What the band does tell you is whether the strategy of systematically buying favourites at that price has historically paid, which is the thing people are actually doing when they buy them. Capital is also tied up until settlement, so the annualized figure is real only if you can redeploy immediately and repeatedly. Not financial advice.
The full price-by-price curve lives in the Truth Machine. Looking for a genuinely risk-free trade instead? Impossible Prices finds ladders where the book has broken its own logic.