Being Right Early Should Pay More Than 11%

You've been right early before. It has never paid you what it was worth.

Not on Polymarket. Not on Kalshi.

The moment your view was worth something, the price that would have paid you for it stopped being available. And it stopped being available because your view was worth something.

Here's the shape of it, and you've lived through it.

News breaks. A market sitting at 50 cents is now a 90% certainty. You see it first. You click. You fill at 90.

You made 11%. Being right was worth 50 cents.

The forty cents in between wasn't taken by somebody faster than you. Nobody got it. It was on the screen for as long as it took one person to press a key, because the 50-cent price was a quote: a promise that costs nothing to withdraw and everything to honor. The market maker did what any of us would do. He canceled.

That isn't bad luck, and it isn't latency. It's the mechanism doing what it was built to do.

On an order book, every price belongs to somebody who is allowed to take it back, and who will, precisely when it becomes worth taking.

The best price you can get on a book is the price at which a professional no longer minds selling to you. You're allowed to trade where you have no edge.

The same thing happens, more quietly, at the other end of a market's life. A book can't open without quotes. So by the time a new market is visible to you, somebody has already read it, priced it, and posted the number that suits them. You can be the first to post a price on a book. You can't be the first to take one. And a price you post tends to fill when somebody knows something you don't.

Two moments in the life of every market where the price should be generous to whoever shows up first. On a book, both shut before you arrive.

BASIS didn't put the price in an order book.

A pool has no quotes to cancel. A pooled price isn't an offer somebody is making. It's a consequence of where money already sits. Nobody can pull it, widen it, or step away from it. It moves only when money actually lands, and every price it passes through on the way is a price somebody really got.

Which means both of those windows stay open long enough to walk through.

BASIS does have an order book, and it's worth being precise about what it's for. It's peer-to-peer: a holder can list shares they already own at a price they choose, which gives an early way out before the market resolves. It isn't what sets the price. Nobody has to quote for a market to open, and nobody can withdraw the board.

---

Every market that works pays the early

Look at where risk capital actually goes, and the same rule turns up everywhere.

A seed round prices below a Series D. An IPO allocation prices below the open. An opening line pays more than a closing one. In none of these is the discount a courtesy. It's the price of showing up while information was still thin, the outcome was still live, and being wrong was a real possibility.

That premium does two jobs, and a market needs both.

It recruits the informed. It pays people to do the expensive, uncomfortable work of forming a view before a consensus exists, and their doing it is the reason the price ends up meaning anything. A venue that doesn't pay the early doesn't get the early. It gets the crowd, arriving late, agreeing with what everybody already knows.

It recruits everybody else. What moves a crowd isn't a fee table. Nobody has ever chosen a venue by comparing its rake. It's a specific person who turned a small position into a large one: proof it can be done, and an invitation to try. Crypto didn't reach the trillions on the strength of its architecture. It got there because the stories of the early were loud, repeatable and believable.

Prediction markets are the one speculative venue that has never really paid that premium. The machine is built so it can't. Being right before the world caught up was worth fifty cents, and it handed you ten.

That makes the category's central claim quietly self-defeating. Prediction markets sell themselves as the most accurate truth-telling instrument ever built, then decline, by design, to pay the people who would supply the truth first. It also leaves them unable to make their own best advertisement. A book's big winners turn up wherever the crowd happened to be wrong: nowhere in particular, and never on request. You can promote a winner afterwards. You can't promise one in advance.

A pool can promise the opening price. Every market on BASIS opens flat, every outcome at the same price, and that best price on the board is there for whoever shows up first. It happens again on the next market, and the one after that. Nobody can tell you who the next winner will be. A pool can tell you when the early buyers get in, and at what price.

The seats fill because the winner looks reachable. The winners here are everybody who moved before the crowd finished arriving, and that share is measurable. We measure it below.

None of this is a promotion. Nobody at BASIS decided the early should be paid more. It falls out of how a pool pays.

---

Why a pool pays the early differently

Start with the one rule that governs everything else.

When a pooled market resolves, the winners split the whole pot in proportion to the shares they hold. Two things set what you get:

How many shares each dollar bought. That's set by your entry price, at the moment you buy, and nothing that happens afterwards changes it.

What each winning share is worth. That's the pot divided by all the winning shares. It's the same for everybody who backed the winner, whenever they arrived.

So, inside one market, returns move inversely with entry price. Buy the winner at 30 cents and each dollar holds twice the shares it would have at 60 cents, so it pays out twice as much, in the same pot, at the same resolution.

That isn't a BASIS feature. It's what a pool is. What BASIS changes is how long the good prices survive. On a book, they rarely survive long enough to be reached.

There's a second effect worth naming. On a book, the asks across all outcomes add up to more than 1, and that excess, the overround, is what makers charge for standing there. Measured on the trades themselves across 111 settled Polymarket markets, weighted by the money that changed hands, it runs from 1.3% on three-to-six outcome fields to 7.9% at thirteen to thirty outcomes, and 7.6% blended. It isn't the venue's fee. It's the cost of having a professional counterparty at all.

A pool has no maker to pay. Winners are paid by losers, and BASIS takes a 0.60% base fee on every bet, plus a creator fee of 0% to 1% set by the market's creator, whether the market has three outcomes or a hundred and four.

That is the deeper version of everything above. Every venue has to decide who stands behind the price, and that one decision shapes the rest.

An order book needs a market maker. A maker who can't withdraw gets picked off by anybody with better information, so the right to cancel isn't a loophole in the design. It's what makes the design possible. The cancellation you resent is what keeps the book open.

A standard AMM needs a liquidity provider, and an LP can't reprice in time. The quotes stay live on both sides at whatever the curve said before the news, and pulling out is itself a transaction, slower than the arbitrageur already on his way. That's why LPs bleed. The effect is documented well enough to have a name, loss-versus-rebalancing, and a literature.

A pool needs neither. Nobody stands behind the price, because the price isn't an offer. It's a record of where money already sits. Your counterparty is the rest of the field. Nothing has to be protected from you, so nothing has to be taken away from you when you turn out to be right.

---

Window one: the market nobody has priced yet

A new market on BASIS opens flat. Ten outcomes, ten cents each. Forty outcomes, two and a half cents each. Nobody's opinion is in the price yet, because no money is in the pool yet.

That's a real gap between the price and the truth, and you're allowed to buy it.

There's no equivalent moment on a book. The first price you can see is already somebody's considered opinion, posted at the level that suits them. Here, the first price is nobody's opinion. It's one divided by the number of outcomes, waiting for the first person with a view.

We measured what that difference is worth. We took 111 settled Polymarket markets, every trade on every outcome, $422 million of notional across fields of 3 to 104 outcomes, and replayed the order flow under pool rules: what would each position have been paid from a pot instead of a capped book?

Across the whole dataset, 61% of winning wallets, holding 71% of the winning money, would have finished ahead of what the order book actually paid them. On the 77 markets where a longshot won, the winning side would have collected 1.33x what the book paid it.

The early numbers are the point of this piece. For money arriving in the first 5% of a market's life, in markets the favorite won, the replay pays by field size:

On fields of 31 or more, the pool beat the book in every single market in the dataset. Blended, early money comes to 1.85x, ahead of the book in 79% of favorite-won markets, but the blend mostly reports on small fields. The big edge is in the wide ones, for the same reason the book's costs are: that's where the book was charging the most, so there was the most room underneath it.

The markets the book wins are inside these figures, not left out of them. On a three-way market the book's embedded cost is about 1.3%, close to a pool's fee, and there's no edge worth claiming there. The advantage sits where it has been all along: early entry, wide fields and outsiders.

Every figure above is the pool alone. Most BASIS markets also have a linked Predict+ token, which you can hold alongside a bet or instead of one. It trades on a Stable+ curve, so its price can only hold or rise in its base asset, and it tracks how much action the market gets rather than who wins. You can sell it at any time, or borrow 100% of its value against it without selling. So an early view doesn't have to sit idle while it waits for a result.

One more property exists only on this side. A book caps a winning share at one dollar, because a share is a claim on one dollar of collateral. A pool has no such cap. When the market resolves, the winners split the whole pot, including the money that went in on the outcomes that lost. On a market where the crowd was badly wrong, that missing ceiling is exactly where the money is. (The same rule means a heavily backed favorite can pay less than a dollar a share. A pool share is a slice, not a fixed amount.)

---

Window two: the gap after the news

The second window is the one we started with, and it's the more interesting of the two, because it reopens every time the world moves.

Something happens. A resignation, a data release, a court filing, a goal. The true probability of an outcome jumps from 50% to 90%.

On a book, that gap closes with a keystroke. The best price you can get is roughly the truth plus half a tick, call it 90.5 cents, which pays about 1.105x for being right about something the whole world just watched happen.

In a pool, the gap has to be bought closed. There's no button. The price only reaches 90 cents if enough money arrives to put it there, and all of that money enters at prices below 90 cents.

So the question becomes: how far does the crowd push it before pushing further stops being worth it?

We modeled it. A binary market, a $100,000 pot, the board at 50 cents, then news that makes the outcome a 90% shot. Rational buyers pile in and stop where the pool stops beating the book. In the model they stop at 70.5 cents, a little over halfway to the truth. The money that fit into that window was $72,989, about 73% of the pot as it stood before the news.

Here's what each rung of that ladder pays in the model, fully diluted, against the 1.105x the book was offering:

Every one of those rungs was real and buyable. None of them would have existed on the book for long enough to hit.

That distinction is the whole argument. On a book, the move from 50 to 90 costs nothing to make: a maker cancels at 50 and reposts at 90, and the forty cents in between is crossed without a single trade inside it. A pool can't skip. The only route to 90 cents is money entering at 51, and 55, and 60, and every level in between. The path has to be bought, and every step on it is somebody's entry.

That money doesn't leave, either. It lands in the pot the earlier buyer already holds a claim on. It isn't a gift to them: it grows the pot and the winning share count together, which is why the ladder above is quoted fully diluted. What it can't touch is the earlier buyer's entry price, fixed the moment they bought. That's where the advantage lives.

In the model, nobody inside the window did worse than they would have on the book. That follows from how the window is defined: it's the range where the pool still wins, and the buying stops when it stops winning.

The window isn't a small-market quirk. We ran the model from $25,000 pots to $1.6 million pots, and it held at 71 to 78% of the pre-news pot throughout. Its size in dollars tracks the size of the pot, with roughly the same shape at every size.

---

Wide fields leave more on the table

The stall point also moves with the number of outcomes. In the four field sizes we modeled, it moved in the direction you'd want if you're the one arriving early.

On the forty-way field, the crowd closed about 31% of the gap. Sixty-nine per cent of it was still there.

The reason is fragmented liquidity, and it's worth understanding because it doesn't wear off as BASIS grows. On a book, forty outcomes means forty separate order books, each needing its own maker, its own depth and its own two-sided quotes. Most of them won't get any. In a pool, forty outcomes means one pot. The money isn't divided. Every outcome is priced against the same pot, and every outcome is buyable from day one.

That's why BASIS lists fields a book struggles with. Winner markets run up to 150 outcomes today, alongside Podium and UP/DOWN markets. Exact order, Parlay and Survivor are coming. More outcomes isn't the point. Wide fields are where the early edge has the most room.

---

How to actually play it

Don't wait to be diluted. If you already hold a position and news breaks your way, the instinct is to sit still and let the pot grow. Think twice. The people arriving behind you are buying at worse prices than yours, and inside the window every dollar they add is a dollar you could have bought at a better price than theirs. It's one of the few moments when adding to a position that has already moved your way costs less than it will cost the people who follow.

Late isn't the same as too late. In the modeled example, somebody arriving at 65 cents, after the crowd had closed most of the distance it was going to close, still beat the best book price by 8%. On the forty-way field, where the crowd stopped at 31%, there was far more room left.

Size against the pot, not against your conviction. In the modeled binary, the window held about three-quarters of the pre-news pot. Knowing roughly where it ends is how you use it without overpaying.

Watch the new listings. The flat open is the one price on the board that's nobody's opinion. It happens on every market, and on a book it doesn't exist.

---

A book's prices are promises, and a promise can be withdrawn for free. A pool's prices are consequences of where money already sits, and a consequence can't be withdrawn. That's the whole difference, and everything above is what it's worth.

You've been right early before. This is a venue where the price that would have paid you for it is still on the board when you arrive.

---

Replay figures are from 111 settled Polymarket markets: every trade on every outcome, $422 million of notional, fields of 3 to 104 outcomes, replayed under pool rules with a 1.5% pool fee and with unfavorable results included. The method is set out in our "Why BASIS chose pools" evidence note in the data room. The news-gap example is a model, with its parameters stated above. Past market behavior does not predict future market behavior. Every figure here describes a specific dataset or a specific modeled market, not what will happen in yours.