Strip the vig, four ways.

On a coin-flip market the four methods agree almost exactly. Skew the market — a heavy favorite against a longshot — and they split, because each makes a different assumption about where the book hides its margin. Multiplicative spreads the vig in proportion to each side's size. Power and Shin both assume the longshot is shaded harder — books know casual money overpays for big payouts, so the +210 side carries more than its share — and Shin models it explicitly as the book defending against informed bettors. Worst-case doesn't apportion at all: it dumps every basis point of vig on the side you're betting and hands you the floor. The wider the four answers spread, the more the fair line depends on assumptions — which is itself a signal that the market is uncertain.

What is vig, exactly?

The vig (or juice/overround) is the book's margin, baked into the prices. Convert both sides of any market to implied probabilities and they'll sum past 100% — that excess is the vig. De-vigging redistributes it away to estimate the true probabilities.

Why four methods instead of one?

Because books don't spread their margin evenly. Multiplicative assumes they do; power and Shin assume the longshot side carries more of it (which matches how books actually price); worst-case assumes YOUR side carries all of it. Running all four shows you the range of honest answers.

Which method should I trust?

On near-even markets they agree to a fraction of a point, so it barely matters. On skewed markets, Shin and power are usually closer to reality than multiplicative, and worst-case is the conservative floor. Whet's scanner uses a weighted consensus across methods and books.

Why don't the worst-case numbers sum to 100%?

By design. Worst-case answers a different question: 'what's the lowest fair probability my side could have if the book put all its margin on me?' Each side gets its own floor, so the two floors together sum to less than 100%.