<p>Bloomberg reported this month that longshot bets losing 98% of the time still dominate trading on Kalshi and Polymarket. That is not random bad luck. It is a documented pattern called the kalshi longshot bias, and new research shows it is baked into how these markets price risk.</p>
<p>The short version: cheap contracts that promise huge payouts look like lottery tickets, and they get priced like one. Traders pay more for a shot at a big multiple than the real odds justify. The expensive, “boring” side of the same market tends to be slightly underpriced instead.</p>
<h2>Where the Numbers Actually Come From</h2>
<p>A working paper by economists Constantin Bürgi, Wanying Deng, and Karl Whelan, titled “Makers and Takers: The Economics of the Kalshi Prediction Market,” pulled real trade data to test this. The results are stark. <a href=”https://www.karlwhelan.com/Papers/Kalshi.pdf” rel=”noopener” target=”_blank”>The paper found that contracts priced below 10 cents lose more than 60% of the money put into them</a>, while contracts priced above 50 cents earn a small positive return on average.</p>
<p>One detail stands out. Contracts trading at 5 cents, which imply a 5% chance of winning, actually won only about 4.18% of the time in the sample. That gap sounds small. Multiplied across millions of trades, it adds up to a steady transfer of money away from longshot buyers.</p>
<p>Bloomberg’s own numbers point the same direction. Around half of all trading volume on Kalshi involves contracts offering potential returns of 1,000% or more. Since early 2025, money staked on “yes” longshots has lost about 15% on Kalshi and 27% on Polymarket. People keep buying them anyway.</p>
<h2>Makers Lose Less Than Takers</h2>
<p>The same research splits traders into two groups: makers, who post limit orders and wait, and takers, who hit existing orders to get filled right away. The split matters more than most casual traders realize.</p>
<p>Takers lose close to 32% on average when they buy longshot contracts. Makers lose closer to 10% on the same trades. The explanation is simple: takers tend to have stronger, more extreme beliefs about an outcome, so they are willing to pay worse prices to act on that conviction. Makers set the price and let it come to them, which gives them a built-in edge.</p>
<p>This lines up with what we found when we looked at why copying Polymarket’s top traders rarely works for ordinary users. Execution style and order type matter as much as picking the “right” outcome.</p>
<h2>Why a One-Cent Tick Size Makes It Worse</h2>
<p>Part of the problem is mechanical. Kalshi’s prices move in whole-cent increments, with one cent as the smallest possible step. That sounds like a technical detail, but it has real pricing consequences at the extremes.</p>
<p>A market maker normally builds in a small cushion, or vig, proportional to the contract price to cover risk. On a 50-cent contract, shading the price by a cent or two is a minor adjustment. On a 2-cent contract, a one-cent shade is a 50% change in price. Market makers cannot apply a proportional cushion at that level, so the pricing gets distorted exactly where longshot buyers are most active.</p>
<h2>Polymarket’s Version of the Same Pattern</h2>er
<p>Polymarket shows a similar bias, but a recent academic paper on arXiv found it depends heavily on how you count it. If you weight every individual contract equally, or just add up pooled dollars across contracts, longshot buyers lose money. But if you weight by the parent event rather than each underlying contract, the picture flips, and longshot buying shows a small net gain.</p>
<p>That split matters for anyone trying to generalize from one platform to another. The underlying driver is similar on both: modest disagreement among traders about true probability, plus a tendency to slightly overstate how likely rare events are, reproduces the bias pattern seen in the live data.</p>
<h2>What This Means If You Trade These Markets</h2>
<p>None of this means longshot contracts are worthless or that favorites always win. It means the odds embedded in very cheap contracts are, on average, worse than they look. A 5-cent “yes” is not really a 1-in-20 shot once fees and the bias are priced in. It is closer to a 1-in-24 shot dressed up as better value.</p>
<p>If you are weighing a cheap contract against an expensive one, treat the listed price as a rough starting point rather than a precise probability. Our guide on how to read prices as probabilities walks through the basics of that conversion, and the bias described here is one of the main reasons the conversion is never perfectly clean.
Posting as a maker rather than chasing fills as a taker also appears to narrow the gap, based on the Bürgi-Whelan-Deng data. It will not eliminate the bias, but it avoids paying the extra premium that comes from buying urgency instead of waiting for it. For anyone building a repeatable approach rather than placing one-off bets, that distinction is worth more than hunting for a single hot pick.
The bias itself is not new to prediction markets; it shows up in horse racing and sports betting too, under the same name. What is new is having millions of Kalshi and Polymarket trades to measure it against in near real time, which is exactly what let researchers quantify it down to a specific cent-by-cent breakdown instead of describing it as a vague tendency.
