In Nimb's docs, alpha means informational edge: something you know or have read in the data that the price does not yet reflect. It is not a promised return. Most apparent edge is noise, or too small to survive the spread and fees, so treat every edge as a claim to test, not a reason to trade.
What is alpha in a prediction market?
In a prediction market the price of a Yes share is the crowd's chance that the event happens: a share at 41% costs $0.41 and pays $1 if it happens. You have an edge only if your own, honest chance differs from 41% by more than it costs you to act, and only if you are right more often than chance alone would explain. "I think it will happen" is not an edge. "The market is at 41%, I put it at 50%, and here is the fact the price has missed" might be.
Edge is also relative to the price you can actually get, not the price on the screen. If the ask is 42% while the headline price is 41%, your edge has to clear 42% plus fees. Fees, spread, noise and resolution risk are the four limits this page keeps coming back to.
Where can a retail trader have an edge?
Prediction markets are not efficient everywhere. Prices on thin, niche or fast-moving markets can lag what a careful reader knows. Four sources are worth taking seriously.
| Source | Example | Its limit |
|---|---|---|
| Domain knowledge | You follow a field closely: a national election's local polls, a sports league's injury reports, a regulator's calendar. You read primary sources before they reach headlines. | Others with the same knowledge trade too. Knowledge is an edge only where the market is thin or the field is obscure. |
| Reading the rules better | The title says "Will X happen by Dec 31?" but the rules name one resolution source, a time zone and edge cases. A market can price the headline while the rules say something narrower. | Rules can be clarified after trading starts, and a disputed result can take days. See how markets resolve. |
| Reacting to news | A scheduled announcement lands and the price has not moved yet, or has moved too far. | Automated traders react in seconds. By the time a human reads a headline, the easy part of the move is usually gone. |
| Logical inconsistencies | A date ladder where "by June" trades at 30% and "by December" at 27%, although December includes June. Or a multi-outcome event whose outcomes add up to well over 100%. | Fixing an inconsistency needs two trades, so you pay two spreads and two sets of fees. Small gaps are often exactly the cost of closing them. |
The common thread: the edge comes from your reading of the event, not from the price chart alone. That is what market reading is for: working out what the price, flow and holders say, and how much to trust them. The guide to reading a market covers the method.
Where does edge not come from?
Three habits feel like edge and usually are not. Nimb's own tests on its data (snapshot of 2026-11-07, before fixes) give the numbers.
- Following whales. A large holder is not an informed one. In those tests, the lean of wallets in profit matched the 30-day price direction in 63% of events: their profit came from the move, so the signal was circular. Large positions are also often hedged: 93 wallet-market pairs held both Yes and No, and 58.3M gross shares carried only 0.55M of net exposure. Nimb now nets these out, but what you see is still a position, not a reason. See holders and wallet lines. "Smart money" is a search term, not a category Nimb can verify.
- Copying flow. Net buying shows that someone was willing to pay, not that they were right. By the time flow is visible, the price has already absorbed it.
- Chasing moves. A 6-point jump often follows news you have not read yet, and sometimes reverses. Patterns in prices are often noise: before Nimb tightened its rules, an expected 43% to 71% of the "moves with / against" links it flagged were false positives. The panel says it plainly: "Unusual moves often follow news: check what happened before you follow it."
How do fees and spread eat alpha?
Every trade has a break-even probability: the real chance above which buying at this price pays off on average, fees included. On Nimb it is the price per share you actually pay, so it sits a little above the market price.
| Market | Rate | Shares | If right | Break-even |
|---|---|---|---|---|
| Politics market | 0.04 | 237 | $237 | 42.2% |
| Crypto market | 0.07 | 233 | $233 | 42.9% |
Illustrative figures for a $100 Yes order at 41%, from the ticket's formula, with all fees (Polymarket's taker fee and Nimb's builder fee) included in the $100. In a politics market, the "Before you trade" check puts it in one line: "Buying Yes at 41% only pays off if the real chance is above 42.2% (fees included)." If you think the chance is 42%, you have no edge in a politics market at this price, only a cost.
The spread adds to this. If you buy at the ask and the mid price is a point lower, you start one point behind. Thin order books make it worse: a large order walks up the book and pays more per share. Full detail is in fees and all-in pricing.
Then comes noise. A binary contract either pays $1 or nothing, so a single result tells you almost nothing about your edge. A worked case: you buy Yes at an all-in 42.2% and the true chance really is 45%. That is a real 2.8-point edge, worth about 6.7% of the stake per trade on average. The outcome swing is so large that you need roughly 1,200 independent trades like it before the edge shows up at two standard errors. At a true 50% it takes about 160; at a true 43%, more than 14,000. Small edges are real in theory and invisible in practice.
How do I know if I have an edge or am just guessing?
Write your chance down before you trade, and score it after the market resolves. A reading log turns "I felt right" into a number you can compare with the market.
| Market | You | Market | Result |
|---|---|---|---|
| Ceasefire by Dec 31 | 50% | 41% | – |
| Rate cut in December | 30% | 34% | – |
Both rows are illustrative, logged on 2026-11-07, with the result entered as 1 or 0 once each market resolves. Keep the date and a one-line reason beside each entry: here, "the rules count a signed text, not a holding truce" and "the latest inflation print".
Score both columns with the Brier score: the average of (chance − outcome)², with chance as a decimal and outcome 1 or 0. Lower is better; saying 50% on everything scores 0.25. Over the same set of resolved markets, compare your Brier score with the market's. If yours is not clearly lower after 50 or more markets, you are guessing, and the market is the better forecaster. How accurate prediction markets are sets the bar you are trying to beat.
Three rules keep the log honest: log every reading, including the ones you did not trade; log before you see the outcome; and never edit an entry. Selective memory is the most common source of imagined edge.
The plain line: most traders do not beat the market after costs. Assume you are one of them until your log says otherwise.
Can Nimb give me alpha?
No. Nimb shows Polymarket's public data, which every other trader can see. A circle on the map, a top holder or a flow chart is not edge by itself: if it were, the price would already reflect it. Nimb gives no financial advice and makes no recommendations.
What Nimb does do:
- Puts price, chart, money flow, holders and rules in one panel, so reading the market is faster.
- Shows the costs before you act: the "Before you trade" checks flag unusual moves, thin books, concentrated holders and the break-even price. See Before you trade.
- States its own limits: hedged positions are netted, weak correlations are hidden, and documented weaknesses are published.
What Nimb does not do: tell you what to buy, rank markets by expected profit, or claim that any wallet, theme or pattern predicts outcomes. Any edge is yours to find, and yours to test.
From reading to execution
When a reading survives the checks, the next step is the ticket. Read and execute describes the workflow: read the market in the analysis panel, check the break-even, then act in the order ticket from your own wallet. Most good readings end without a trade, and that is the workflow working.
Prediction markets are risky; you can lose your whole stake. Nothing on Nimb is financial advice. Read the risks of prediction markets before you trade.
Last reviewed 2026-11-07 · Data from Polymarket · Not financial advice · Nimb is not affiliated with Polymarket.