Methodology

Find the winning profile. Then choose the bet.

Touchline learns from settled matches and the information available before those outcomes. It compares current bets with those historical profiles to estimate their chance of a winning settlement. The strongest qualified profiles lead the ranking; current prices and fees refine the choice across Kalshi and Polymarket US.

Winston’s card is the same for every reader. It has four sections, each with its own bar, one line per verified match and at most eight lines in all. It never fills to a number: a section with nothing that clears its bar says so. The public record starts on 8 September 2026 and grades each section on its own line.

Bankers (up to three): the estimated win chance is at least 90%, the all-in price is at most 96¢, and the estimate is above the price. Full size. Near-certain and paying a few cents, boring on purpose. The price is wrong (up to three): an estimated win chance from 50% to under 90% that is at least ten points above the all-in price, with the lower estimate still at or above the price. Half size. Some of these lose; the price is why they are there. The angle (up to two): a margin, half, total, team total, both teams to score, first team to score, correct score or corners line with an estimated win chance of at least 80% and a lower estimate at or above the all-in price. Half size. In play: live calls on their own clock, held to their own record. Each section stands or falls on its own settled record: twenty settled picks across ten independent matches with a losing match-balanced return stop that section’s stakes until it recovers. These are selection bars, not guarantees or a claim that any cutoff is an optimal threshold. Since 14 September 2026 the card has no reader-set risk or payout controls.

What each number means

• Estimated win chance is the model’s estimate that the selected side settles as a winner.
• Lower estimate is a lower statistical bound reflecting uncertainty in the model and comparable historical evidence. It is shown separately from the estimated win chance; it is not a guaranteed outcome or an additional minimum win threshold.
• Purchase price uses the real quoted quantities and prices needed for the full intended stake, including all order fees. The calculation can span several order-book levels. If a stake is unknown, the estimate uses a $25 planned stake, not an assumed account balance.
• Payout after fees is the settlement payout divided by the whole purchase cost, including fees. Payout includes the original stake; profit is payout minus cost. The whole intended purchase ($25 when no bankroll is known) must pay more than its section’s bar after fees: for a banker, one over the smaller of the estimated win chance and 96%; for a wrong price, one over the smaller of the estimate less ten points and the lower estimate; for an angle, one over the lower estimate. At a 90% estimated win chance, for example, a banker’s payout must exceed 1.11×; a purchase that pays less than its bar does not qualify, whatever its win chance.
• Estimated advantage is the estimated win chance minus the average cost per contract, including fees. It is secondary to the win profile and purchase requirements.
• Comparable settled matches is the historical sample supporting the selected profile, separate from the current model’s pick record.

How a selection reaches the card

The model checks match data, settlement rules, current quotes and real available quantities. BET means the winning profile, a section’s bar, full-stake liquidity and the section’s after-fee payout requirement all qualify. WATCH identifies a winning profile that is blocked by return, available size or current execution checks; PASS fails the winning-profile or section requirements. Quotes expire, so a recommendation must remain current before its Buy link is shown. The estimate describes the quoted purchase, not a guaranteed fill. You choose whether to place an order on the exchange; Touchline does not place it for you.