Help

Polymarket is a prediction market. People bet real money on questions about real events, like “Will it rain in London on Friday?”

Each possible answer (“Yes”, “No”, or a name in a multi-choice market) is sold as shares. A share costs between 1¢ and 99¢. When the market ends, a share of the right answer pays $1. A share of a wrong answer pays $0. If the market ends as a 50/50 tie, each share pays 50¢.

The price is roughly the crowd's guess of the chance. A price of 30¢ means the crowd gives that answer about a 30% chance.

We watch every trade on Polymarket. Most traders lose money over time, but a small group has a good record. We find that group, which we call smart money, and track what they buy, sell and hold.

Then we look for markets where smart money is betting one way but the price has not moved there yet. Those markets become ideas. We also run a simulated book that follows the ideas, so you can see if they work.

We do not trade for you, and we do not know the future. We show data and our best cautious estimate.

The score looks at things like:

• Profit compared to the risk taken, not only total profit.
• Consistency: many good results, not one lucky win.
• Wins compared to the price paid. Winning a 90¢ bet is easy; winning many 30¢ bets is harder.
• Activity: how many different markets the wallet trades.

Each wallet gets a tier: Top (a proven record, counts most), Mid (good but less proven) or Probation (too new or too uneven, counts least).

We leave out market makers. These are traders who buy and sell both sides all day to earn small price gaps. Their trades do not show an opinion about the result.

Quality score

A 0-1 score summarizing how skilled this wallet has been: consistent profits, good risk-adjusted returns, strong returns per dollar put at risk, and a track record of being right when the odds were against them. Higher is better.

Entry edge

How good this wallet's entry prices were on resolved markets: the average gap between what happened (win or lose) and the price it paid. 50% is break-even; consistently above that means it wins more than its entry prices implied.

Breadth (90d)

The number of different markets this wallet has traded in over the last 90 days — a wallet spread across many markets is less likely to be a one-off lucky bet.

P&L

Realized profit or loss in dollars: cash from trades plus payouts from markets that have resolved, over this time window.

For each market we add up what smart wallets hold and what they bought or sold recently. From that we get:

• Smart money leans: the answer they favor.
• Smart wallets (backers): how many smart wallets back that answer.
• Consensus: how much they agree. 50% is an even split; 100% means nearly all back one answer. It is agreement, not a chance of winning.
• Confidence:low, medium or high, based on how much real money stands behind it. Low shows as “Thin data”. Medium shows as “Low stake”.

Then we make our own model estimate of the chance. It starts at the market price and moves only partof the way toward what smart money's bets suggest. On purpose, it stays close to the price. It shrinks again when:

• the price already moved since smart money bought (they already got that gain, shown as “priced in”), or
• the signal is old (no fresh smart-money buying).

Divergence (or edge) is our estimate minus the price. Positive means we think the price is too low for that answer.

The model is labelled “unvalidated” until it beats the market price on markets that have ended. Until then, treat it as a rough guide, not a real probability.

Each idea has a signal label (Strong, Moderate, Watch or Weak) and a number from 0 to 100. The number is the idea's rank among all current ideas. It combines the edge, how many wallets back it, the money at stake, the market's liquidity and how fresh the signal is.

The tabs at the top (presets) change the sort and filters in one click. A red “Smart money exiting” badge means the backers are now selling.

Divergence

The model's estimate minus the market price, both at the same snapshot. The estimate starts from how big a share of each wallet's bankroll is committed — a $5 lottery ticket implies almost nothing, a third of a bankroll implies a lot — then is shrunk toward the price. Positive means their money says the price is too low. The model is unvalidated until it beats the price on resolved markets.

Consensus

The share of tracked smart-money weight behind the leading outcome. On a two-outcome market it runs from ~50% (split down the middle) to 100% (nearly unanimous) — an even split does NOT read 0%. This measures agreement among these wallets, not the probability they'd assign to the outcome.

Model estimate

Two numbers for the outcome smart money leans toward, from the same snapshot: the market price (a 70% price means the market gives it a 70% chance), and the model's estimate. The estimate starts from how much of their bankrolls smart wallets committed, then is heavily shrunk toward the price and decayed for signal age. Until the model is validated against resolved markets it is a heuristic, not a probability.

Smart wallets

The number of historically profitable wallets currently holding or buying this outcome (tiny sub-$10 positions don't count — and wallets betting the other way or selling don't either). More wallets means the signal is less likely to be one trader's lucky guess.

At stake

The total dollar value these smart wallets currently have riding on this outcome.

Flow 24h

Net flow of smart-wallet money into (positive) or out of (negative) this outcome in the last 24 hours — a leading indicator of where sentiment is heading right now.

Smart money entered

The stake-weighted average price smart wallets paid for this outcome, against the price at the same snapshot. If the price has already run far past their entry, most of their edge is priced in.

Edge left

How far the model's estimate still sits above the price (max of 0 and divergence). It shrinks as the price catches up to where smart money bought — an idea whose edge is gone is late, however strong the backing.

Last smart buy

Time since the newest smart-money buy backing this outcome. Older signals decay: the information may already be in the price.

Resolves

When the market is scheduled to resolve. Money in a long-dated market is locked up until then.

Strongest signal

Ideas ranked by overall signal strength — divergence backed by wallet breadth, real dollars at stake, liquidity, and recent activity.

Likely winners

Ideas the market already rates as more likely than not (win chance ≥ 50%), ranked by signal strength weighted by win chance. High chance of winning AND smart money still sees extra edge.

Longshots

Low-priced outcomes where smart money's bet sizing implies far better odds than the market price (divergence ≥ 10pp). Big payoffs if right, but most will lose — position accordingly.

Heavily backed

Only ideas with at least 10 smart wallets holding or buying the consensus outcome — the broadest agreement, least likely to be one trader's hunch.

Most promising (sort)

Signal strength re-weighted by win chance: signal strength × market-implied probability, so well-backed likely winners rank above extreme longshots.

It sizes each bet with fractional Kelly. Kelly is a formula that says how much to bet when you think you have an edge. The full amount is very risky if the estimate is wrong. So we bet only a small fraction of it.

It also has safety limits:

• A cap on how much goes into any one market.
• One bet per event, so you do not bet twice on the same thing.
• A cap per sector and on closely linked markets.
• Real order-book prices and Polymarket taker fees, so the price you see is the price you would pay.
• A minimum edge that must remain after those costs.

Often it suggests few bets, or none. That is on purpose. The sizes are illustrations from an unvalidated model, not advice.

The track record is a paper book: a pretend account that started with $10,000. About every hour it follows the ideas on its own. It buys at real order-book prices, pays fees, and adds a little extra cost (slippage) to be safe. It sells when the smart-money signal fades. Otherwise it holds until the market ends.

Two benchmark books bet the same amounts without the signal:

• Favorite benchmark: the same stake on the answer the market already favors, in the same market.
• Placebo benchmark: the same stake on a random market at a similar price.

The follower is only useful if it beats both benchmarks. If it cannot beat the placebo, the signal adds nothing.

Sample size matters. A few weeks, or a few dozen settled bets, is not enough to know. Luck is large in small samples: a strategy with no skill wins 18 or more of 30 even-odds bets about 1 time in 5. Look at results over many settled bets before you draw a conclusion.

How to read the numbers:

Equity

Cash plus what the open positions could sell for now.

ROI

Total gain or loss compared to the starting $10,000, including open positions.

Realized / unrealized

Realized is profit or loss already locked in. Unrealized is the paper gain or loss on positions still open.

Max drawdown

The biggest drop from a high point. It shows how bad a bad stretch got.

Hit rate

The share of finished positions that made money.

Free users see the track record with a one-day delay: both the positions and the totals show the book as it was 24 hours ago.

Free: you can browse ideas, markets, events, wallets and the track record. The smart-money data is 1 hour behind, so you see what smart money did an hour ago. A “Delayed 1h” badge shows this. On the track record, the newest positions appear after a day.

Pro:live data (marked “Live”), the portfolio builder, and full API access. Manage your plan and API keys on Account.

Press ⌘K (or Ctrl K) anywhere to search for a market, event or wallet.

• You can lose your whole stake. A share of the wrong answer pays $0.
• Not proven. The strategy has not shown that it makes money. Past results of any wallet, or of our paper book, do not predict future results.
• Surprises. Markets can end in a way nobody expected. Results can also be disputed or delayed.
• Low liquidity. In small markets, your own order can move the price, and selling early can be hard.
• Costs. Fees and the gap between buy and sell prices eat into small edges.
• Lag. When you copy smart money, you are always later than they were. The price may already have moved.
• Rules. Polymarket restricts trading in some countries, including the US.

TailSlip is an analytics tool. Nothing here is financial advice or a recommendation to trade. Only bet money you can afford to lose.

Share

A unit of one answer in a market. It pays $1 if that answer happens and $0 if not.

Price

What one share costs now, from 1¢ to 99¢. Roughly the crowd's chance for that answer.

Resolution

The moment the market ends and the result is decided. Then shares pay out.

Liquidity

How much money is waiting to trade. More liquidity means you can buy or sell without moving the price much.

Edge / divergence

Our estimate of the chance minus the price. +5 points means we think the answer is 5 percentage points more likely than the price says.

Backers

The smart wallets that hold or buy the answer smart money favors. Shown as “Smart wallets”.

Kelly

A formula for bet size based on your edge. We use a small fraction of it to limit risk.

Slippage

Paying a worse price than the one shown, because your order uses up the best offers first.

Fee

A charge Polymarket takes on some trades from the buyer who takes an existing offer (a “taker fee”).

Paper trading

Trading with pretend money but real prices, to test a strategy without risk.

Drawdown

A drop from the highest value to a later low. Max drawdown is the biggest such drop.