Prediction markets & the ORACLE pillar.

What a prediction market actually is

A prediction market is a venue where you buy and sell shares in the outcome of a future event. Instead of trading the price of an asset, you trade the probability of something being true. "Will the Fed cut rates at the next meeting?" "Will this candidate win?" "Will this team lift the trophy?" Each of these becomes a tradable market, and the price of a share tells you the crowd's collective estimate of how likely the answer is YES.

The mechanism is honest in a way few financial instruments are: nobody pays out unless the real-world event actually resolves. There's no funding rate, no carry, no perpetual position to roll. You buy a claim on a known outcome, you wait for reality to decide, and you get paid — or you don't. Polymarket is the largest on-chain venue for this, settling in USDC on Polygon, and it's the venue BABA's ORACLE pillar trades.

How a binary YES/NO market works

The cleanest prediction market is binary: exactly two outcomes, YES and NO. Each market issues two tokens — a YES token and a NO token. The rule that makes everything tick is simple: the YES token and the NO token always sum to $1. If YES trades at $0.73, NO must trade at $0.27. They're two halves of the same dollar, because one of them will be worth exactly $1 at resolution and the other worth exactly $0.

So when you pay $0.73 for a YES token, you're risking $0.73 to win $1 if the event resolves YES — a profit of $0.27, or about a 37% return. If it resolves NO, your token is worth $0 and you lose the $0.73. The price is the implied probability: $0.73 means the market thinks there's a 73% chance of YES. That single fact — price equals probability — is the whole game.

Reading a Polymarket market

Open any market on Polymarket and you'll see four things worth knowing. The question and its resolution criteria — the exact, written rule for how YES and NO are decided, including the source of truth and the deadline. The current price — the implied probability, the headline number. The order book — the live stack of bids (people willing to buy YES at a price) and asks (people willing to sell), exactly like a normal exchange. And the volume and liquidity — how much money is actually behind the market.

Read the resolution criteria before you read the price. A market that looks mispriced is often just precisely worded — "wins the popular vote" is a different bet from "wins the election," and the price reflects the exact wording, not your loose mental version of it. The order book matters too: a thinly-traded market can show an attractive price that you can't actually fill at size without moving it against yourself.

> Always read the resolution rule first, the price second. Most "free money" markets are not mispriced — they're just worded more carefully than you read them.

Where mispricings actually come from

If price equals probability and the crowd is usually right, how do you ever make money? Because the crowd is usually right, not always. Edges show up in specific, recurring places. Stale prices — a market hasn't updated to fresh news, and you saw the headline first. Emotional skew — partisan or fan markets where people buy with their hearts, pushing a price away from the true odds. Thin liquidity — a low-volume market where one large, careless order leaves a temporary gap. Resolution-rule confusion — the crowd is pricing the wrong question.

Your edge is never "I think YES will happen." Everyone thinks something. Your edge is "the market is paying me $0.73 for a thing that's really 85% likely" — a gap between the price and your honest, defensible estimate of the true probability. No gap, no trade. That discipline is exactly what separates a prediction-market trader from a gambler.

Resolution and claiming your winnings

When the event happens, the market resolves. An oracle — a designated source of truth, often backed by a dispute-and-bond mechanism so a wrong resolution can be challenged — reports the outcome on-chain. The winning token becomes redeemable for exactly $1; the losing token becomes worth $0. You then claim (redeem) your winning tokens, converting them back to USDC in your wallet. Nothing is held for you; settlement is on-chain and yours to claim.

One practical note: resolution can lag the event by hours or, in disputed cases, longer. Your capital is locked in the position until the market settles. Factor that into sizing — money committed to a slow-resolving market is money you can't deploy elsewhere.

Time-decay and why timing matters

As an event approaches, prices move toward certainty — they drift toward $1 or $0 as uncertainty resolves. This is the prediction-market version of time-decay: the closer you are to resolution, the less room there is for the price to be wrong, and the less you stand to gain from being right early. Buying YES at $0.73 a month out, when there's still room for surprises, is a different trade from buying it at $0.95 the night before, when almost everything is already priced in.

Timing cuts both ways. Buy too early and you tie up capital while the price chops around for weeks. Buy too late and the edge has already been competed away. The sweet spot is the window where you have an information or judgement advantage but the market hasn't fully caught up — and that window closes as resolution nears.

Jurisdiction — read this before anything else

Prediction markets are regulated very differently around the world, and Polymarket is restricted in a number of jurisdictions, including the United States. US persons are not permitted to trade on it, and other countries have their own rules. This is not optional fine print — it is the first thing to check before you fund anything. Whether you may lawfully use Polymarket at all depends entirely on where you are.

> Warning. Polymarket is restricted in some jurisdictions, including the United States. Confirm it's lawful where you are before funding a wallet. See the Risk Disclosure. If in doubt, skip this venue — none of the other pillars depend on it.

The Gnosis Safe proxy setup

Polymarket doesn't trade directly from a standard wallet. It uses a Gnosis Safe proxy — a smart-contract wallet that sits between your signing wallet and the market. In plain terms: your everyday wallet (your signer) holds the keys, but trades execute through a dedicated Safe contract that holds your USDC collateral and your positions. It's a little more setup than a plain wallet, but it's well-documented in Polymarket's own onboarding flow, and once it exists you fund the Safe with USDC and trade from there.

The reason this matters for self-custody: the Safe is yours. You control the signer; no company custodies your funds. The proxy is just the contract layer that lets Polymarket route orders and settle on-chain. Keep your signer key secure exactly as you would any wallet — it's the master key to the Safe.

How BABA's ORACLE pillar trades them

ORACLE is the BABA pillar that trades prediction markets — and it's built around a hard truth: a single model, including a single AI, is overconfident and wrong often enough to lose money. So ORACLE doesn't trust one opinion. It runs a multi-model committee — Claude, Grok, and Gemini — and each model independently reads the live world (news feeds, event context, the market's own resolution rule) and judges whether the current market price is fair, cheap, or expensive relative to its own probability estimate.

A trade only happens on high-conviction agreement. If the committee is split, or only mildly confident, ORACLE does nothing — and doing nothing is the correct, most common outcome. On top of the committee vote sit two guardrails. A ≥48-hour resolution gate: ORACLE won't enter a market resolving in under 48 hours, because too close to the deadline the edge is gone and the time-decay works against you. And a post-event guard: if the underlying event has effectively already happened, ORACLE stands down rather than chasing a price that's already collapsed toward certainty.

How this differs from perps

If you've done Chapter 4, the contrast is sharp. A perp is continuous and leveraged: price can go anywhere, you can be liquidated, and a stop-loss is the discipline that saves you. A prediction-market position is binary and event-driven: there are only two outcomes, your maximum loss is exactly what you paid, and there is no leverage and no liquidation — your token simply settles at $1 or $0 when reality decides.

That makes prediction markets a different kind of risk, not a smaller one. You can't get stopped out, but you also can't cut a loser — once the event resolves against you, it's gone, full stop. The skill isn't managing an open position tick by tick; it's getting the probability right before you enter, and sizing for the chance you're wrong. Perps reward execution. Prediction markets reward judgement.