The Microstructure of Wealth Transfer in Prediction Markets

I mentioned this on a different post - the biggest problem with prediction markets is not the gambling or dumb people losing money. Its the fact that it gives very powerful people a vehicle to make lobsided bets on outcomes they control.

A small example of this would be NFL / NBA Refs fixing playoff games with a bad call or two. This actually happened 20 years ago, an NBA ref went to prison over being bribed just $2000 per game.

The much worse example is the fact that you can make 100-1 odds on whether the US airstrikes Iran today... or How many times Pam Bondi says the word "China" in a press conference.

I'm getting some really skeezy ads for prediction markets on TikTok at the moment, the message is effectively "hey, are you broke? earn $50+/day on Kalshi!"

I'm a little confused by the "Yes" versus "No" asymmetry.

For example, one of the top trending ~~bets~~ markets right now is on whether Miami or Indiana will win the NCAA football championship tonight. You can either take "Yes" on Indiana at 74c, or "No" at 27c, or you can take "Yes" on Miami at 27c or "No" at 74c. Or, there's another potential outcome - you can also bet on a tie at 10c yes/91c no.

Is this research suggesting that an optimistic Miami fan can somehow get a better return by buying "No" on Indiana than a "Yes" on Miami?

Why is Kalshi structured with these yes vs. no options for all outcomes?

Something that appears to be missing: Certain events attract "advertising" types of bets. E.g. There is value in making a candidate appear to be a leader, so dedicating dollars to swinging the market is more of a form of advertising than an intelligent bet.

So it would be interesting to measure the inefficiencies of various bets vs the total market value in that bet.

e: Although full disclosure, I did not pick apart the entire paper. Maybe it's buried in there.

The analysis is interesting, but I think it ignores a few factors:

1. The article mentions the bid/ask spread for contracts, but I believe that Kalshi also has its own fee structure. Small edges (an expected loss of 0.57¢ on a 1¢ contract implies an expected gain of 0.43¢ on a 99¢ contract, or a 5.75ppt edge) can be easily eaten by even small fees, and liquidity provision is all about small edges.

2. The article ignores the time value of money, and contracts take time to resolve. If a contract won't resolve for six months and the risk-free rate is 5%, then buying a "sure thing" over 97.5¢ is a loss net of otherwise earnable interest.

3. Long shots offer greater implied leverage to bettors, making them more attractive. This is still (sometimes) an exploitable mispricing, but it's closer to the well-understood "bet against beta" factor.

(Edit to add) Also, I think their explanation of the non-returns on finance is lacking:

> Why is Finance efficient? The likely explanation is participant selection; financial questions attract traders who think in probabilities and expected values rather than fans betting on their favorite team or partisans betting on a preferred candidate. The questions themselves are dry ("Will the S&P close above 6000?"), which filters out emotional bettors.

Financial contracts are the ones that are most perfectly hedges with existing markets. "Will the S&P close above X?" is a binary option, after all, so it's comparatively easy for a market-maker to almost perfectly offset their Kalshi positions with opposite positions in traditional markets.

Has anyone noticed a lot of polymarket posts on their X (formerly known as twitter) feed claiming to be making a fortune? It makes me feel like its some kind of coordinated guerilla marketing scheme.

How do prediction markets account for interest rates? I feel like I should be willing to pay no more than ~96 cents for a contract that will definitely resolve to a dollar in a year. Who puts up the other 4 cents?

I hope we manage to leverage prediction markets to actually achieve goals rather than just making a casino out of it.

For instance, if you spot malware in a commit you could bet heavily against it being merged, and that would attract the maintainers' attention, and they'll see what you see and not merge it, and you get paid for the code review--that money would come from whoever bet that it would get merged, which you could require be the author of the malware. I haven't worked it out entirely but it seems that there are opportunities to build games that reward dilligence and transparency and penalize deception and spam.

I’m surprised by the somewhat positive comments. I thought this was just a chance for insider trading without repercussions. If I work at $corp and know the hotly anticipated whatever is announced tomorrow, I can finally cash in on that knowledge.

And the people losing their life savings on gambling now have one more tool.

But what do I know. I’m probably oblivious to what greatness those Truth Engines will enable.

I feel we need a term for these attempts to paint gambling as something other than gambling. Or just proper enforcement for gambling platforms like "prediction markets". Personally I find it disappointing to see so many people wasting their time on this stuff. I'm sure Coplan, for example, could be a productive member of society, but instead chooses to waste his time on stupid stuff like Polymarket.

I have no background in financial markets at all, but it strikes me that in markets like this, the "house" should be insiders right? The Maduro capture had an insider profit something like 400k. How would one go about understanding how that impact efficient markets?

Could you use inefficient markets as a predictor of great volumes of insider trading?

tl;dr

dataset: 72.1m trades and $18.26b volume on kalshi (2021-2025)

core findings:

longshot bias: well documented longshot bias is present on kalshi. low probability contracts are systematically overpriced. contracts trading at 5 cents only win 4.18% of the time.

wealth transfer: liquidity takers lose money (-1.12% excess return) while liquidity makers earn it (+1.12%).

optimism tax: the losses are driven by a preference for "yes" outcomes. buying "yes" at 1 cent has a -41% expected value. buying "no" at 1 cent has a +23% expected value.

category variation: finance markets are efficient (0.17% maker-taker gap) while high-engagement categories like media and world events are inefficient (>7% gap).

mechanism: makers do not win by out-forecasting takers. they win by passively selling "yes" contracts to optimistic bettors

This article lacks even the most basic understanding of probability and statistics. Slot machines "93 cents on the dollar" return is a statistical certainty of 7% loss. You are playing a repeated game which by the law of large numbers will converge to the 93% probability.

In prediction markets if the markets are fully efficiently priced, in the absence of transaction costs you WILL get 100% back in the long run.

Slots are also unskilled games, prediction markets clearly some participants have a clear market edge, thus not efficiently priced.

To me this is all the more reason to get regulatory gatekeeping out of the financial markets

If the odds in some financial products are worse than gambling while everyone can access gambling, then people should stop making a distinction under the guise of protecting investors

it just drives investors to actual gambling because they cant get the exposure they were already looking for

I wonder how much of the activity on prediction markets these days is competing LLM scripts? I would guess the overlap in prediction market punters and AI boomers is high.

[dead]

[dead]