Polymarket probabilities and odds

A midpoint averages the best buy and sell offers; a last trade records a completed transaction. Identify which value is displayed. Comparing yesterday’s last trade with today’s midpoint as if the measurement method were unchanged can produce a misleading impression of movement.

For a percentage probability p, decimal odds equal 100 / p. Thus 25% corresponds to 4, 40% to 2.5 and 80% to 1.25. Reverse the calculation with 100 / decimal odds. Zero percent has no finite equivalent, while 100% corresponds to 1. These are mathematical examples, not available offers.

An empty input does not mean zero probability. Before calculating, distinguish missing input from a valid number or a range error.

An illustrative 40% estimate corresponds to a per-share price scale of 0.40 and mathematical decimal odds of 100 / 40 = 2.5. Ten shares actually acquired at 0.40 cost 4. With a standard confirmed payout of 1 for each of the 10 retained shares, gross receipt is 10 and the payout-to-acquisition ratio is 10 / 4 = 2.5. Profit before expenses is 10 − 4 = 6; the 40% estimate is not the amount of profit.

That ratio has conditions: a card can display a midpoint or last trade while an order fills differently. A payout of 0.50 per share returns 5 for the same 10 shares, giving a ratio of 1.25; a losing outcome returns zero. An earlier sale uses actual proceeds instead. Fees, spreads and partial fills alter the calculation. The converter therefore verifies neither the true probability, an executable quote nor a future return.

An increase from 30% to 45% is 15 percentage points. The relative increase in the number is 50%, but that is a different measure. OddsAtlas uses percentage points to make the movement understandable. The selected period must have genuine comparable observations; a missing earlier point cannot be replaced with zero.

The converter changes scales rather than verifying a forecast. Its calculation excludes fees, spreads, available quantity, rule ambiguity and data delays. Before comparing two sources, match the question, deadline and result source. Even a careful match does not establish guaranteed profit or prove that the market estimate is correct.