A 10-minute routine for checking a prediction market spread

A scanner is good at finding price differences. It is bad at telling you whether the difference is real. This is the routine we use before treating a pair as a trade.

Minute 0–2: filter the noise

Start with match confidence at 90% or higher and a minimum depth that matches your budget. Hide anything with an ROI that looks too good. A 200% return on a market that resolves tomorrow is usually a stale quote, a thin book or a rule mismatch. The default max ROI filter exists for a reason.

Minute 2–5: read the two markets

Open both markets in separate tabs. Read the resolution source, the close time and the edge cases. If the titles differ in a number, a date, a time zone or a direction word, stop and compare the exact wording. This is the part that cannot be automated away.

Minute 5–7: check the fills

Look at the size available at the quoted ask on both legs. If one side has $5 and the other has $5,000, your trade is a $5 trade. Check the fee on each venue. For Kalshi, use the fee at the actual price, not a flat percentage. For Limitless, remember the fee may be an estimate.

Minute 7–9: plan the legs

Decide which leg to place first. The thinner leg is usually the one that can move away from you, so place it first and keep the second tab ready. Do not assume both orders will fill at the quoted price. If the first leg fills and the second does not, you are holding a naked position, not an arbitrage.

Minute 9–10: write it down

Record the pair, the prices, the size, the fees, the resolution rules and the time. If the trade goes wrong, the notes are how you find out which assumption failed. If it goes right, they are how you repeat it.

Ten minutes is not a guarantee. Some pairs take an hour, and some are not worth ten seconds. The routine is there to make the fast decisions consistent, not to make every decision fast.

If you only remember one thing: the scanner finds the question. You still have to answer it.