Stop treating Polymarket like single bets. Build hedge baskets, map correlated markets, and verify cross‑market mispricings using public order book data.
Explore structured research pillars and internal link paths.
Visit Research SeriesMost Polymarket traders ask one question: “Is this market mispriced?”
Better traders ask a different one: “If I’m right, what else has to be true, and where is the cheapest place to express that view?”
That’s the portfolio lens. It turns Polymarket from a pile of isolated coin flips into a connected graph of exposures—some redundant, some contradictory, and some quietly offering free optionality.
This post breaks down:
Prediction markets don’t just price outcomes. They price flows, and flows are rarely clean.
Two markets can move together because they describe the same event from different angles. They can also diverge because:
Academic work on prediction markets has long highlighted that prices can reflect more than “objective probability,” including risk and hedging demand in the market. That’s not a bug—it’s a feature if you know how to trade it. See Wolfers & Zitzewitz (2004): https://jmvidal.cse.sc.edu/library/wolfers04a.pdf
If you want to hedge or arbitrage across markets, you first need to understand how they’re related. In practice, Polymarket “multi‑market mapping” tends to fall into three buckets.
One event gets broken into multiple markets with different granularity:
Portfolio insight: these markets often carry overlapping exposure. If you hold one, you may be unintentionally long (or short) the others.
Some markets are designed as complements. In ideal conditions, the prices should satisfy simple constraints:
Mispricings happen when liquidity is fragmented or one side becomes the “tourist” side.
This is where the portfolio lens gets sharp.
If market A is “Event happens” and market B is “Event happens by date,” then:
P(by date) ≤ P(happens at all) should hold.
Or if market C is a sub‑event that can’t occur without market A, then:
P(C) ≤ P(A) should hold.
When these inequalities break, you’ve found either a data issue or a tradeable dislocation.
The goal isn’t to build a perfect academic hedge. The goal is to trade a thesis with less fragility.
Here’s a workflow that fits how Polymarket actually trades.
Instead of “Yes is undervalued,” write:
You’re not building a story. You’re identifying dependent exposures you can price‑check.
For any “main” market you want to trade, list substitutes that capture similar information:
You’re looking for the contract that offers the best execution, not the most exciting headline.
Most retail trades fail the same way: you’re directionally right, but you’re wrong about timing, path, or resolution mechanics.
Good hedge legs often hedge:
You don’t need a full quant model to avoid common sizing mistakes.
A simple approach:
The result is usually a basket where you keep most of your upside, but your worst‑case drawdown stops being catastrophic.
In textbook markets, arbitrage is instant and riskless. In prediction markets, “arbitrage” is often:
Treat “arbitrage” as a trade class, not a guarantee.
Check A: Complement parity
If you can obtain both sides’ midpoints, track:
Spread = (P(Yes) + P(No)) − 1
If the spread persistently deviates beyond fees + typical spread, something is off.
Check B: Inequality parity
For any “subset” market:
P(subset) − P(superset) should never be positive in a clean market.
When it is, you’re either looking at:
Polymarket exposes public endpoints for market discovery and order book data. You can pull market metadata and order book snapshots directly from their APIs:
Use the Gamma / Markets API documentation to find events and their markets, then list the markets you care about:
The CLOB API provides public endpoints for order book data (bids/asks, spread, midpoints):
Once you have the market identifiers (or token/asset IDs), you can fetch snapshots over time and compute:
Polymarket also publishes a curated list of blockchain data resources (including Dune/Allium/Goldsky) that can be used to validate volume, positions, and trade history:
That’s how you separate “the market moved” from “the market got walked.”
Whales rarely express a view in exactly one place. They choose the leg that offers:
When you see a whale enter “a market,” assume there’s a second position somewhere:
This is why single‑market copy‑trading breaks: you’re copying one leg of a portfolio.
If you want context‑rich signals instead of raw prints, that’s the point of our Smart Money tooling: it’s designed to read flow as a portfolio, not as isolated trades.
Here’s a template you can reuse:
Then do one thing most traders skip: pre‑plan exits.
If you want to trade baskets, you want tooling that supports baskets:
Sources & Further Reading
Related Research
Research Series
Follow related research articles or jump to the full pillar library.