Every so often someone points at a PredictIt price sitting a few cents off the same contract on Kalshi or Polymarket and asks why nobody has closed it. The honest answer is that plenty of people have looked, run the math on what it costs to actually collect that spread, and walked away. The gap is real. Whether it is yours to take is a different question, and most of the time the answer is no, or not at the size you were hoping for.
PredictIt is a strange venue by design. It runs under a specific regulatory arrangement that came with hard constraints, and those constraints do not exist on Kalshi or on Polymarket. Once you understand what they are, the persistent price differences stop looking like free money and start looking like exactly what they are, which is a structurally capped price that cannot converge past a certain point.
The three frictions that actually matter
There are a few limits people quote, but only three of them change how you trade against the other venues.
- The per-contract trader cap. Each PredictIt market historically allows only a limited number of unique traders on a given contract, roughly in the mid-hundreds range. Once a market is full, you cannot get in even if you want to. So the marginal informed trader who would normally arbitrage a gap away simply cannot enter. The price stays wrong because the people who would fix it are locked out.
- The per-trader investment cap per contract. There is a ceiling on how much money any single account can put into one contract, and it is small, low four figures at most. This is the one that kills size. Even if you find a clean spread, you cannot lean into it. A trade that would be worth doing at real size is only available to you at pocket-change size.
- The fees on profit and on withdrawal. PredictIt takes a cut of net winnings, historically around ten percent, and then takes another cut when you move money off the platform. Kalshi and Polymarket do not work this way. So a spread that looks like it clears your costs on paper has to survive two separate haircuts before the cash reaches you.
Stack those together and the picture is clear. You are trading small, you are trading against a wall of locked-out competitors, and you are paying to leave with your winnings. None of that is a reason to never touch PredictIt. It is a reason to be extremely picky about when you do.
When a PredictIt leg is worth the friction
The trade only makes sense when the raw spread is wide enough to survive the fees and still pay you for locking up capital until the market resolves. That is the whole test. Because of the profit fee, a PredictIt leg needs a meaningfully bigger edge than the same leg on Kalshi or Polymarket would.
A rough rule I use: if the price difference does not clear the fee drag by a comfortable margin, skip it. A one or two cent gap is noise once you net out the profit cut and the withdrawal cut. You are working with something once the gap is wide enough that even after PredictIt takes its share, you are still ahead of where you would be just holding the position on the other venue.
The other thing that has to be true is that the two contracts are genuinely the same event. This sounds obvious and it is where most people get hurt. PredictIt, Kalshi, and Polymarket write their rules independently. A market that looks identical can settle on different criteria, different sources, or a different resolution date. If you treat two contracts as one position and they resolve on different terms, you do not have a hedge, you have two separate directional bets that happen to point the same way today.
So before any cross-venue trade, walk the resolution language line by line:
- What exact outcome pays out, in the platform's own words.
- Who or what is the settlement source, and can it disagree between venues.
- The resolution date and whether an early or delayed call is possible on one side but not the other.
- How each venue handles an ambiguous or contested result.
If any of those diverge, the spread you are looking at is not an arbitrage. It might still be a trade, but you are taking resolution risk, and you should price it as such.
Sizing around the constraints
Assume you found a real gap on a market that is genuinely the same event. Now you size it, and the caps do the sizing for you more than you do.
The binding constraint is almost always the per-trader investment cap. Whatever that number is on the contract, that is your maximum PredictIt leg, full stop. Your hedge on Kalshi or Polymarket gets sized to match the PredictIt side, not the other way around, because the other side has effectively unlimited room and PredictIt does not. Sizing the liquid leg first and then discovering you cannot fill the capped leg is a common and avoidable mistake.
A workable sequence looks like this:
- Confirm the market is not already full and you can actually open a position.
- Set your PredictIt size to what the per-contract cap allows, and haircut it for the profit and withdrawal fees so the net still clears your threshold.
- Only then place the matching leg on the liquid venue, sized to the net PredictIt exposure.
- Plan to hold to resolution. The withdrawal fee means round-tripping cash in and out for a small spread eats the edge, so PredictIt works best when you leave capital parked until the contract settles.
Because everything is small, the practical reality is that PredictIt is rarely the main event. It is a leg you add when the gap happens to be fat, not a venue you build a book around. Where it earns its place is inside a broader cross-venue view, watching the same events priced across Kalshi, Polymarket, and PredictIt at once and only acting when the difference is large enough to matter. That is the kind of monitoring I keep instrumented in Blockcircle, mostly so I do not waste attention on gaps that were never tradeable to begin with.
Why the gaps never fully close
Here is the part that trips people up. On a normal market, a persistent price difference is a signal that you are missing something, because arbitrageurs should have erased it. On PredictIt, the difference persists precisely because the mechanism that would erase it has been capped out of existence.
When a contract is full, no new trader can push the price. When the investment cap is low, the traders who are in cannot push it far. And the fees mean the price has to move a real distance before anyone bothers. So PredictIt prices can sit systematically offset from the more liquid venues, sometimes rich, sometimes cheap, and stay there. It is not mispricing in the usual sense. It is the honest price of a market where the correction mechanism is throttled.
Practically, that means you stop expecting convergence and start treating PredictIt as its own quote. You take the spread when it is wide enough to pay you after fees, you size to the cap, you match the hedge to the net exposure, and you hold to resolution. Most days there is nothing to do. The trade shows up a handful of times, and the discipline is in ignoring all the gaps that look tradeable but are not once the friction is priced in.