The four steps
01
Drop a screenshot
Paste an image straight from your clipboard, upload a file, or skip the image entirely and type the market question. Any venue works — if the market is visible on your screen, it can be analysed. Polymarket and Kalshi are also available directly from the Live board, which hands the market to the analyzer with price and close date already filled in.
02
The desk reads the image
Vision extraction pulls the trade-relevant fields out of the picture: the venue, the exact contract wording, the category, the quoted YES price, traded volume and the resolution date. You see what it read before anything is analysed, and you can correct any field it got wrong.
03
The model estimates fair value
The analyzer estimates the probability that the market resolves YES, independently of what the market is currently charging. It then compares that estimate with the quoted price. The difference is the edge — the reason to take the trade, or the reason to leave it alone.
04
You place the trade
Polysight never holds your funds and never submits an order you have not approved. The verdict is stored in your history with the original screenshot, so you can revisit the call later and see whether the reasoning held up.
Every field in the verdict
A verdict is only useful if you know what each number means. Here is the whole output.
- Side
- YES or NO — which side of the contract the estimate favours. When fair value sits close to the quoted price, the honest answer is no side, and the desk will say so rather than manufacture a trade.
- Fair value
- The estimated probability the market resolves YES, shown as a percentage. A fair value of 27% means the desk thinks the outcome happens roughly once in four comparable situations.
- Edge
- Fair value minus the quoted price, in percentage points. A market quoted at 38% with a 27% fair value carries an 11-point edge on the NO side. Small edges get eaten by fees and spread; large ones usually mean you have missed something, so read the risks.
- Confidence
- How much evidence sits behind the estimate, from 0 to 100. Low confidence with a large edge is a research prompt, not a signal. High confidence with a small edge is often the better trade.
- Drivers
- The specific factors pushing the estimate away from the market price — base rates, comparable historical outcomes, recent news, structural features of the resolution criteria.
- Risks
- What would make the thesis wrong. These are the conditions to watch after you enter, and the reason the exit plan exists.
- Entry band
- The price range where the trade still makes sense. Above it the edge has gone, and chasing a moved market is how a good read becomes a bad position.
- Exit plan
- Where to take profit and where to cut. Defined before entry, so the decision is made while you are still objective.
What it will not do
It will not promise a win rate, and it will not pretend to find an edge in a market that is fairly priced. It does not hold your funds, and no order reaches a venue unless you approve it. Prediction markets carry the risk of total loss, and nothing here is financial advice.
See pricing