Losing is fun
Will Kalshi and Polymarket lose users as they realize insiders tend to win? I'd bet against it
Back in the bad old days of crypto trading, when there were hardly enough buyers and sellers out there to repopulate an American ghost town, a once major exchange called Poloniex had a feature called the “troll box.” In the troll box, users could discuss their trades and their perspectives on trading with each other.
As one redditor described it, “It looks like pack of ranting teenagers on steroids.”
The wisdom of crowds!
A new paper has evaluated Polymarket, the world’s largest prediction market, and found that it their accuracy does not come from lots of users. Their accuracy comes from a few highly informed traders (just 3% of them).
The authors don’t offer an upshot for their research, but I think it’s implied that people participating in prediction markets are suckers. Or, as the authors put it:
“Prediction markets are remarkably accurate, but the source of this accuracy is not the crowd.”
Does playing prediction markets and losing money actually make bettors losers, though?
Could people losing money there be playing a different game?
The game
First of all, how prediction markets work. These applications get written about a lot but the real world mechanics seldom get broken down.
On Polymarket, people can create a market around any outcome that can be objectively measured. Like: Will Brady Dale be named king of Britain in 2026? In a just universe, that should clearly be a “YES,” but even I would bet against it. (I would love to meet the person who would take the other side of that bet.)
The cool thing about these markets is that the sides have prices and they are always less than a dollar. Buying a “YES” on the market above would probably show up at less than 1¢. “NO,” by contrast, would sell for more than 99¢, but less than $1.
At the start of 2027, the market would resolve, and all the people who bought “NO” would get $1 for each “NO” they bought. That wouldn’t be much of a profit of course, but the idealists buying “YES” would get nothing. A total loss!
The appeal of a market like this is: hey, maybe 2026 is really crazy and somehow I do become king? Like in that 1991 John Goodman vehicle, King Ralph (which, for some reason, I really did see in the theater).
If the outcome of a market ends up being ambiguous, it gets turned over to UMA, another crypto product. It doesn’t really matter for our purposes though, but there do tend to be a lot of ambiguous endings in markets. Language is hard.
Losers
The paper uses a lot of math and sophisticated analysis to show that a small handful of people make money on the site. A few people get lucky. A few other people look like insider traders.
Most people stink and lose.
The researchers write:
“The persistence of skill and anti-skill in prediction markets is unusually high.”
Translation, people are fooling themselves. And just in case that’s not bleak enough, it says in the next paragraph:
“Luck is not persistent.”
However, here’s what I’d like to see: How much are individual traders actually losing?
I expect it’s not very much. In fact, according to TRM Labs, the typical bet is about $12.
Other games
I’m not a prediction market trader. I have taken out less than $100 worth of trades ever — all on politics. I’ve only had one market resolve (I lost all my $7 — or whatever it was).
But I suspect people trade on Polymarket (and Kalshi) for other reasons.
First, to sharpen thinking. It’s one thing to spout an opinion, but it’s another thing to refine that into an objective bet?
Like, you can say that you believe that it’s good to reduce regulation, for example. It’s another thing to make some kind of call about what will happen in an industry when regulation gets reduced. Do you think employment will go up? Do you think it’s market cap will increase?
Look through the markets. Place a bet. Even if you lose money, you’ll learn something by deciding which one to choose and then watching the result unfold.
Second, to put skin in the game. Many of us are prone to spouting prognostications all the time. It’s easy to forget your old opinions or predictions though. How are you ever going to know if you are any good at predictions if you don’t keep track.
Prediction markets give you an actual scoreboard for your predictions. Even if you never bet more than $1 at a time, you have this record you can look at to see how your worldview holds up.
Lastly, it’s fun. I think journalists really underestimate how important this is in markets. Prediction markets are social media with better bragging rights. When you get a call right, you can brag.
(And when you get it wrong… you don’t need to talk about that one.)
To that point: If you haven’t taken a look at these sites, they basically all have their own built in troll boxes. Look at this one for the market, FBI Director “Kash Patel out by…”:
Are you telling me this doesn’t look kinda fun?
Fun is a good coin. I pay good money for fun all the time, at zero dollar-denominated profit. I bet the authors of this paper do, too.