The Quick Answer
A regular trader can post a big $700,000 loss on a prediction market because a binary contract carries all-or-nothing risk at settlement: it pays $1.00 if you are right and $0 if you are wrong, with nothing in between. A price that reads like a smooth 96% probability feels like a near-certainty, so people size up hard, right up until the one time it settles at zero and takes the whole position with it. The full mechanism, a worked example with real Kalshi numbers, and how to size so a single bad outcome can't end you are below.
The line that started this was posted in a prediction-market community: "We're watching in real time a regular guy lose $700,000 today." I have watched enough of my own tickets go from "this is basically effortless" to zero to know exactly how a run like that happens. That is not stupidity, and it is rarely a rigged market. The instrument itself is the culprit. What follows is a story about how a number that looks like a probability quietly hides an all-or-nothing bet, and what to do about it before you are the cautionary post.
Where The $700,000 Loss Comes From
That $700,000 line is community-reported, not a court filing, so treat the exact figure as an anecdote rather than an audited fact. But the shape of it is not rare. The Roosevelt Institute analyzed more than 400 million trades on Kalshi and estimated that retail traders lost $583.5 million between July 2021 and May 2026, with sports markets alone accounting for $371.6 million of that. A single wipeout post is the tail of a very large, very ordinary distribution of losing accounts.
So the honest question is not "how did one unlucky person lose $700,000." It is "what about a binary contract lets a normal position balloon into a life-altering number without the trader feeling it coming?" The answer lives in how these contracts pay out, which is the promise I will pay off in the worked example a few sections down.
Why A Binary Contract Hides The Risk
On a prediction market, you are not buying a slice of anything. You are buying a coin flip with a fixed payoff. Kalshi says it plainly in its own explainer: "Each contract on Kalshi settles to $1 if your answer is right, and $0 otherwise." The price you pay, somewhere between 1 and 99 cents, is just the market's read on the odds. Kalshi's own example: "If some question has a 40% chance of being answered YES, the expected value of a Kalshi YES contract is 40¢."
Here is the trap. That 40¢, or 90¢, or 96¢ moves around all day like a stock price. You watch it tick 94, 95, 96, and your position marks up in real time. It feels continuous, like a thing you could always sell for close to what it is worth. But the mark-to-market price is a read on the future. The settlement is the fact. And the settlement only knows two numbers.
The price is a rumor about the future. The settlement is the fact. And the fact only knows two numbers, $1 or $0. Every tick between now and resolution is the market arguing over which one you get, and you get exactly one of them.
That gap between how the price feels (smooth, sellable, forgiving) and how settlement actually works (binary, final, unforgiving) is where tail risk hides. If you want to understand where your maximum loss actually lives, read our breakdown of whether you can lose more than you put in on Kalshi alongside this one, because the two ideas are the same coin.
A Worked Example: How A 96¢ "Safe" Bet Becomes A Wipeout
Say a market is trading at 96¢. The news says the outcome is all but decided, and the price agrees. You think, correctly, that this is probably going to happen. So you buy 1,000 Yes contracts at 96¢, a $960 position.
| Outcome At Settlement | Each contract pays | Total payout | Profit / loss |
|---|---|---|---|
| Settles Yes (You Were Right) | $1.00 | $1,000 | +$40 |
| Settles No (You Were Wrong) | $0.00 | $0 | −$960 |
Look at the bottom row, because that is the one everybody ignores. You are putting up $960 to win $40, roughly 24 to 1 against you in dollar terms, on a bet you will win most of the time. Win it twenty-four times in a row and you are up $960. Lose it once and you give all of it back. This is the "picking up pennies in front of a steamroller" trade, and the reason people do it over and over is that the wins feel effortless and the loss feels impossible, right up until it is not.
Now scale the psychology, not just the math. Every clean win tells you the read was right, so next time you buy 3,000 contracts, then 10,000. The position that finally settles No is the biggest one you have ever held, because your confidence and your size grew together. That is the $700,000 shape: not one insane bet, but a ladder of "safe" ones where the last rung is load-bearing and the floor is $0. For the psychology behind why 96¢ favorites get overbought in the first place, our piece on favorite-longshot bias on Kalshi walks through the pattern.
Fees Don't Blow You Up, Position Size Does
People love to blame fees for prediction-market losses, and fees are real, but on these trades they are almost a rounding error compared to the tail. Kalshi's published fee schedule charges takers roughly round up (0.07 × contracts × price × (1 − price)). That curve peaks when a contract is priced near 50¢ and shrinks toward nothing at the extremes.
Run it on our example. At 96¢, the fee on 1,000 contracts is about 0.07 × 1,000 × 0.96 × 0.04, which is around $2.69. On a $960 position facing a $960 downside, a $2.69 fee is not the thing that ends you. But notice the cruel twist: the fee is lowest exactly on the deep-favorite trades, so the one trade type that can quietly bankrupt you is also the one that costs the least to enter. Cheap to put on, catastrophic to be wrong on. If you want the full mechanics of why the fee curve looks like that, we cover it in how Kalshi makes money and the Kalshi fee breakdown.
The Moment Of Truth Is Settlement, Not The Ticker
The whole illusion holds together because you can watch the price and tell yourself you would get out before it went bad. But a binary contract does not fade to gray. At resolution it prints $1 or $0, and if the event breaks against a large deep-favorite position, there is frequently no friendly exit price on the way down, because everyone holding it is trying to sell the same suddenly-worthless contract at once.
Here is the callback to the earlier point: the price felt sellable the entire time, and then at the exact moment it mattered, it was not. This is why I treat the ladder before settlement as information, not comfort. Reading the order book tells you how thin your exit really is, which is a different question from how likely you are to be right. We get into that skill in reading the Kalshi ladder, and into the resolution mechanics themselves in what happens at settlement and the settlement-station trap.
How To Size SO One Tail Can't End You
You cannot make a binary bet stop being binary. What you can control is how much of your bankroll rides on any single $1-or-$0 outcome. A few rules I hold myself to:
- Price Your Downside, Not Your Win Rate. On a 96¢ contract, the honest question is never "will this hit," it is "am I comfortable losing 96¢ to make 4¢, and how many of these before one costs me a month of profit?" If the answer scares you, the size is wrong.
- Cap Single-Outcome Exposure. I do not let any single $1-or-$0 outcome cost me more than 2% of my bankroll, decided before I place the trade, and I never let a deep favorite quietly breach that cap just because it feels certain.
- Do Not Let Winning Trades Raise Your Size On Autopilot. The blow-up shape is confidence and size growing together. Set your unit before the streak, not during it.
- Respect The Exit You Actually Have, Not The One On The Screen. A mark-to-market price you cannot sell into at size is not a real exit.
None of this promises you will win. It is the difference between a bad day and a post other people screenshot. For the foundation under all of it, our how prediction markets work explainer and the asymmetric payoff piece cover why the math tilts the way it does.
If what you actually want is to see the reasoning behind a position before you risk anything, our analysts publish their work openly.
FAQ
Can you really lose $700,000 on a prediction market? The specific $700,000 figure is community-reported rather than documented, so treat it as an anecdote. But large losses are well documented in aggregate: retail Kalshi traders lost an estimated $583.5 million between July 2021 and May 2026, per the Roosevelt Institute. The mechanism that makes a six-figure prediction market loss possible is the same either way: all-or-nothing settlement plus growing position size.
Why does a 96¢ contract feel so safe? Because the price reads like a 96% probability and moves smoothly all day, so it feels like a near-certain, easily-sellable position. Settlement ignores all of that and pays either $1 or $0, so a 4% miss costs you the full 96¢ you risked to make 4¢.
Are fees the reason people lose big? No. On deep-favorite trades the taker fee is often only a few dollars per thousand contracts, because Kalshi's fee curve is smallest at extreme prices. The losses come from the binary payoff and position sizing, not the fee.
How do I actually protect myself? Size every position by its downside, cap how much any single $1-or-$0 outcome can cost your bankroll, and do not let a winning streak inflate your unit. You cannot remove the binary risk, only decide how much of it you carry.
The Bottom Line
The reason a "regular guy" can post a $700,000 loss is not that prediction markets are a scam or that he was uniquely reckless. The cause is simpler: a binary contract dresses an all-or-nothing bet in the clothing of a smooth, sellable probability, and the trades that feel the safest, the deep favorites, are the ones that hide the steamroller. Zoom back out and the whole thing collapses to one idea: what the screen shows is a guess in progress, but settlement pays only the two numbers at the ends, $1 or $0. Trade like that second number is always live, size so no single outcome can end you, and the figure under a screenshot of your account will never be the one people are watching in real time.
