Skip to main content
Prediction markets exist because people want to bet “X will happen” or “X will not happen.” The idea is simple; the implementation is less obvious and often misunderstood. Spot and perpetual markets quote prices that can run freely upward (with exchange-specific rules). A pure “will this happen?” bet needs a fixed probability scale: 0% to 100%. Venues map that to money as $0.00 to $1.00 per share. Outcome prices from $0 to $1 as 0% to 100% probability Polymarket, Kalshi, and Hyperliquid HIP-4 binary markets all work the same way at the economic level. Each outcome has two tokens: Yes and No.

​
Token creation

Like futures, these markets are zero-sum among traders: what one side gains, the other loses (fees aside). At the start there are no Yes or No tokens. You can always create a pair by locking $1 of collateral (for example USDC on Hyperliquid): that mints 1 Yes + 1 No. The reverse is a merge: return one Yes and one No, unlock $1. There is always the same number of Yes tokens as No tokens in existence, because they are only born and destroyed as pairs. That outstanding count is open interest. Split $1 into Yes and No tokens; merge back to $1 A user’s position is simply the tokens they hold (and the mark value of those tokens). When the market settles, one side wins: Yes pays out near $1 and No near $0, or the reverse. Locked collateral is paid to winners.

​
Order books and matching

Each token is a separate asset with its own order book. The books are linked: they can cross-match. That link is what makes prediction-market order flow different from a normal single-asset spot book. Limit and market orders work as on spot (a market order is an aggressive take). What differs is which counterparty you can hit. There are four useful combinations: Four match types: mint, transfer Yes, transfer No, merge Example (mint across books):
  1. User A rests a bid to buy Yes at 0.60.
  2. User B sends a market buy No.
  3. If there is no better No ask than 0.40 (= 1 − 0.60), B can match across books against A: a new Yes+No pair is created from $1. A gets Yes, B gets No. Open interest rises by 1.
Yes interest at p is economically the same as No interest at 1 − p. Engines still apply priority (for example HIP-4’s price–side–time rule when both sides sit at the same merged level).

​
Merged order book

Economically, Yes and No are one market expressed twice. A bid to buy Yes at p is the same commitment as an offer to sell No at 1 − p (and the duals for the other three corners). HIP-4 states that merge rule explicitly; Kalshi’s API can return only bids on Yes and No and still describe the full market, because a Yes bid implies a No ask at one minus that price. Buy Yes at 0.60 is equivalent to sell No at 0.40

​
Why merge for reading

If you keep two native books, “where is size?” is split across two price axes that run in opposite directions. Flipping No prices with p → 1 − p puts everything on a single Yes-odds axis:
  • No asks land on the same levels as Yes bids
  • No bids land on the same levels as Yes asks
  • Depth at one economic level can be summed: Yes resting interest plus flipped No resting interest
That merged ladder is what venues and tools mean by a merged order book. Matching still respects priority rules (for example HIP-4’s price–side–time at a shared level); the merge is how you see and aggregate liquidity, not a different asset. Separate Yes and No books aligned onto one Yes-price ladder by flipping No prices

​
How MarketLens shows it

MarketLens treats HIP-4 (and the same binary model) as an aggregated order book that evolves over time, not as two unrelated tickers you must eyeball side by side. To put Yes and No on one chart:
  1. Take the No book.
  2. Reverse every No price: p → 1 − p.
  3. Map sides so flipped No lines up with Yes (No asks with Yes bids, No bids with Yes asks).
  4. Stack / sum size at each shared level.
That aligned history is what you see in both order-flow modes: MarketLens merged Yes/No prediction market order flow — app UI So when you read “size at 0.62” on a MarketLens prediction market, that level can include Yes interest posted at 0.62 and No interest that was posted at 0.38 and flipped. Cross-book mint and merge flow still print as trades; the merge is how resting depth is drawn as one coherent market.

​
References