Inside Polymarket: Shares, Sets & the Order Book

Last lesson you learned the one idea that makes prediction markets click: a share’s price is the market’s implied probability, and a YES share is a coupon that pays exactly $1 if the event happens and $0 if it doesn’t. Buy YES at 62¢ and you’re saying “this is at least 62% likely” — and if you’re right, you collect a dollar.

Lovely. But where does that dollar come from, what exactly are you holding, and why does buying 5,000 shares cost you a worse price than buying 50? This lesson opens the hood. We’ll follow a real dollar from your wallet, watch it split into a YES and a NO share, see how the order book matches you with a counterparty, and learn why spread and slippage — not the headline “zero fees” — are the real cost of trading size. By the end, the abstract “62% chance” will be a concrete pile of tokens with plumbing you can name.

The plumbing — USDC dollars on Polygon

Analogy. A casino runs on chips, not loose cash: you swap dollars for chips at the cage, play in chips, and cash out at the end. Polymarket runs on digital chips called USDC — but with a twist a casino would never allow: each chip is permanently, redeemably worth exactly one US dollar, and you keep them in your own pocket, not the house’s vault.

Definition. USDC is a stablecoin — a cryptocurrency engineered to always trade at $1, backed by real dollars and short-term Treasuries held in reserve. Polymarket denominates everything in USDC so that “$1 of collateral” really is one dollar, not a coin that might be worth $1 today and 80¢ tomorrow. The market itself lives on Polygon, a layer-2 / sidechain built on top of Ethereum: it processes transactions for a tiny fraction of Ethereum’s fees (“gas”), so placing or cancelling an order costs fractions of a cent instead of dollars. And it’s non-custodial — your shares are tokens that sit in your wallet, controlled by your keys. Polymarket can’t reach in and take them.

Worked example. You want to trade. You move, say, $200 of USDC into your wallet. Each of those 200 USDC is redeemable for one dollar — not “about a dollar,” one dollar — because that’s what makes the whole probability interpretation honest. When you later buy 100 YES shares at 62¢, you spend 62 USDC = $62, and a Polygon transaction fee of perhaps a tenth of a cent rides along. Were this on Ethereum’s main layer, that same trade might cost several dollars in gas alone — which is precisely why Polymarket chose a cheap layer-2.

Pitfall. Don’t confuse a stablecoin with a volatile cryptocurrency like Bitcoin or Ether. If markets were collateralized in ETH, a YES share priced at “62% chance” could swing in dollar value just because ETH moved — the probability reading would be polluted by crypto noise. The dollar peg is load-bearing: it’s what lets price equal probability cleanly.

When it matters

The plumbing usually fades into the background — until you’re moving real size or trading fast. Then the non-custodial part matters (no exchange can freeze withdrawals of tokens you hold in your own wallet), and the cheap gas matters (a market-maker posting and cancelling hundreds of orders a day would be bankrupted by mainnet fees). The stablecoin peg matters always: it’s the silent assumption behind every “this is 62% likely.”

Outcome tokens & the complete set — the $1 conservation law

Analogy. Think of a raffle where there are only two tickets in existence for each draw: one stamped YES, one stamped NO. After the draw, exactly one ticket is a winner worth $1 and the other is worthless. So the pair — one YES plus one NO — is always worth exactly $1, no matter the odds, because between them they’re guaranteed to hold the single winning ticket.

Definition. A Polymarket market mints outcome tokens (also called conditional tokens): for a yes/no question, a YES token and a NO token. Together, 1 YES + 1 NO form a complete set, and a complete set is always redeemable for exactly $1 at resolution, because exactly one of the two will pay $1 and the other $0. That gives the market’s conservation law:

$P_{\text{YES}} + P_{\text{NO}} = 1.$

This is not a market convention or a rule someone enforces — it’s structural. The two tokens are two halves of a guaranteed dollar.

Worked example. Suppose YES trades at 62¢. The conservation law instantly pins NO at $1 − 0.62 = 38¢. If you ever saw YES at 62¢ and NO at 40¢ (summing to $1.02), you could buy a complete set for $1.02 that’s worth exactly $1 — a guaranteed loss, so nobody sells like that for long. The reverse (set costing 96¢) would be free money — a guaranteed-dollar bundle on sale for less than a dollar — and traders would pounce until prices summed back to $1. We’ll make that arbitrage rigorous later; for now, just hold onto the identity.

Pitfall. “If NO is cheap, the event is unlikely, so NO is a bad bet.” No — cheapness is the whole point. NO at 38¢ pays $1 if the event fails: that’s a profit of 62¢ on a 38¢ stake (a 163% return) when it hits. A low price means a low implied probability and a high payoff-per-dollar. The price already encodes the odds; there is no separate “bad bet” penalty.

When it matters

The complete-set identity is the backbone of every other mechanic in this lesson. Minting, merging, the two-sided order book, and the arbitrage that keeps prices honest all rest on “YES + NO = $1.” If you remember only one equation from this course, make it that one.

Minting & merging — making and unmaking the dollar

Analogy. A money-changer who works both ways for free. Hand him a $1 bill and he hands you two tokens, YES and NO. Bring back both tokens and he hands you the dollar. He never profits and never refuses — he’s just a machine that splits and fuses, enforced by code.

Definition. Minting (also splitting): deposit $1 of USDC and the smart contract mints you a complete set — 1 YES + 1 NO. Merging (also redeeming): hand back a complete set — 1 YES + 1 NO — and the contract returns $1 of USDC. Both directions are available to anyone, any time, at a fixed 1-to-1 rate. You never have to find a buyer to get your dollar back from a complete set; you can always merge it yourself.

This is the source of every share in the market. Sellers usually aren’t “the house” — they’re traders who minted a set and then sold the half they didn’t want.

Worked example — minting then selling the leg you don’t want. You’re bullish on YES and the order book quotes NO buyers at 38¢. Instead of buying YES outright, you:

1. Mint: deposit $1 USDC → receive 1 YES + 1 NO.
2. Sell the NO into the order book at 38¢ → collect $0.38.
3. Keep the YES. Your net cash outlay was $1.00 −$0.38 = $0.62.

You now hold one YES share for a net cost of 62¢ — exactly what buying YES directly at 62¢ would have cost (conservation law again). Minting-then-selling and buying directly are two routes to the same place; which is cheaper on a given day depends on where the order book is thickest. And whenever you’re done, if you hold a matching YES and NO you can merge them back into $1 rather than selling either into the market.

Play with the splitter below: toggle mint vs. merge to watch a dollar fission into the YES/NO pair (and fuse back), and drag the YES price to see NO = $1 − YES hold no matter what.

The CLOB — a central limit order book, not an AMM

Analogy. A stock exchange’s bulletin board. On one side, buyers post: “I’ll pay up to 60¢ for 500 YES.” On the other, sellers post: “I’ll sell 250 YES at 62¢.” A match happens when a buyer’s price meets a seller’s. Nobody’s bet is against “the house” — every trade is one trader meeting another on the board.

Definition. Polymarket matches trades through a CLOB — a central limit order book: for each outcome token, a live, sorted list of resting limit orders. Bids are buy orders (sorted highest-price-first); asks (or offers) are sell orders (sorted lowest-price-first). The best bid is the highest someone will pay; the best ask is the lowest someone will sell for. The gap between them is the spread, and the mid — halfway between best bid and best ask — is the market’s cleanest read of the implied probability.

$\text{mid} = \frac{\text{best bid} + \text{best ask}}{2} \approx \text{implied probability}.$

This is not an AMM. An automated market maker (AMM) like Uniswap prices trades off a formula against a shared liquidity pool — there’s no order book, just a curve. A CLOB instead has named, individual orders you can hit or join. The practical upshot: on a CLOB you can post a limit order and provide liquidity (earning the spread when someone trades against you), exactly like a market-maker, rather than only trading against a pool.

Worked example. Best bid for YES is 60¢ (someone wants to buy), best ask is 62¢ (someone wants to sell). The spread is 62 − 60 = 2¢. The mid is (60 + 62)/2 = 61¢, so the market reads the event as ~61% likely. If you place a market buy, you pay the ask (62¢); if you place a market sell, you hit the bid (60¢). To bet NO, you don’t short anything — you simply switch to the NO token’s book and buy NO shares there. No borrowing, no margin call, no locating shares to short: NO exposure is just owning NO tokens.

Pitfall. “The mid is the price I’ll get.” No — the mid is a reference, not a fill. You buy at the ask and sell at the bid; the mid is the average of the two, and you only ever transact at one side of it (worse than mid by half the spread, at least). Confusing the mid with your fill price is the first step toward being surprised by your cost.

Spread, depth & slippage — why size costs more

Analogy. Buying out a farmers’ market’s tomatoes. The first crate is cheap, but once you’ve cleared the cheapest stall you have to move to the next, pricier stall, then the next. The more you buy at once, the higher the average price you end up paying — because you eat through the cheap supply and climb into the expensive stuff.

Definition. Depth is how many shares are resting at each price level. A market order “walks the book”: it fills against the best price first, and when that level’s size is exhausted, it rolls up to the next-best price, and so on. Slippage is the gap between the price you hoped for (the best ask) and the volume-weighted average price you actually paid after climbing several levels. Thin books (little depth) produce big slippage; deep books absorb size with barely a wobble.

Worked example — walking the ask ladder. Suppose the YES ask side looks like this, and you send a market buy for 300 shares:

Your average fill price is $188.50 / 300 = 62.83¢ per share — not the 62¢ you saw quoted. The slippage is 62.83 − 62 = 0.83¢ per share (about $2.50 across the whole order) because your size forced you up into the 63¢ and 65¢ levels. Had you bought only 150 shares, you’d have filled entirely at 62¢ with zero slippage. Size is the slippage.

Drag the order-size slider below to watch a market buy eat deeper into the ask ladder, lifting your average fill and your slippage as it climbs:

Pitfall. Reading the top-of-book price and assuming the whole order fills there. On a thin market, a single large order can blow through several levels and pay cents more per share than the quote — and on a very thin book it may only partially fill (there simply aren’t enough sellers). Always check the depth, not just the best ask, before sizing up.

When it matters

For a 20-share flutter, slippage is a rounding error and you can ignore the ladder. For a serious position — or in a thin, newly-listed market — slippage and spread are the dominant cost of trading, dwarfing any headline fee. The professional move is to provide liquidity with limit orders (and earn the spread) rather than cross it with impatient market orders.

Fees & gas — where the cost actually hides

Analogy. A “free” checking account that still nickel-and-dimes you on the spread when you exchange currency. The advertised price (zero monthly fee) isn’t where the money goes — the cost is baked into the exchange rate you’re quoted. Same story here.

Definition. Polymarket runs a low-to-zero explicit trading-fee model — it does not skim a large percentage off each trade the way a traditional bookmaker’s “vig” does. And because the market lives on Polygon, the gas (network transaction fee) for placing, cancelling, or settling an order is tiny — fractions of a cent. So the headline costs are small by design. The real cost of trading size, as the last section showed, is the spread you cross plus the slippage you cause — an implicit cost set by the order book, not a fee line on a receipt.

Worked example (qualitative). Buy 300 YES shares in our ladder above and your average fill is 62.83¢ versus a 62¢ best ask. That ~0.83¢/share is your true trading cost — and it dwarfs the trivial Polygon gas you paid to submit the order. Two traders can face the “same zero fees” yet pay wildly different effective prices, purely because one crossed a wide spread into a thin book and the other patiently posted a limit order at the mid.

Pitfall. Judging a market as “cheap to trade” off its advertised fees alone. A zero-fee market with a 5¢ spread and shallow depth is expensive to trade in size; a market with a tiny fee but a 1¢ spread and deep books is cheap. Always price the spread and expected slippage into your edge — a thin edge can be entirely eaten by crossing a wide spread.

When it matters

For occasional small trades, fees and gas are genuinely negligible and you can forget them. The moment you trade frequently, in size, or in thin markets, your P&L is governed by execution cost — spread + slippage — far more than by any fee schedule. Sizing your edge against the spread is the difference between a profitable strategy and a break-even one.

Putting it together

A Polymarket trade is a tidy chain of mechanics. Your stake is USDC — dollar-pegged stablecoin — held non-custodially in your own wallet on Polygon, a cheap Ethereum layer-2. Every market mints outcome tokens: 1 YES + 1 NO is a complete set always redeemable for exactly one dollar, which is why $P_{\text{YES}} + P_{\text{NO}} = 1$ and why price reads as probability. You can mint a set from a dollar or merge one back to a dollar on demand. Trades match through a CLOB — a central limit order book of bids and asks, not an AMM pool — where the spread separates best bid from best ask and the mid is your cleanest probability read. Sizing up means a market order walks the book, climbing pricier levels and incurring slippage: the gap between the quoted best price and your volume-weighted average fill. And while explicit fees and gas are tiny, the genuine cost of trading size is the spread plus slippage — the implicit price the order book charges you for impatience.

Now you know what you’re holding, where it lives, and how it trades. But one giant question still hangs over every market: when the event is over, who decides who was right? A market can quote “62% chance the bill passes” all day, yet someone — or something — must ultimately rule YES or NO so the winning shares can pay out their dollar. Next lesson we meet resolution and the UMA optimistic oracle: a clever escalation game where anyone can propose an outcome, anyone can dispute it by putting money on the line, and a decentralized vote settles the rare contested case. It’s where a prediction market’s promise either becomes a paid-out dollar — or a controversy.