Why do some soccer bettors lose more money than others? In an efficient prediction market, each gambler should break-even before costs (but losing a constant amount after costs, reflecting the bookmaker's margin). Previous empirical studies across numerous sports betting markets show that bets on longshots tend to lose more than bets on favourites (favourite-longshot bias). We use 163,992 soccer odds from ten European leagues to test plausible hypotheses around why some soccer bettors lose more money than others. Are soccer bettors with above average losses simply biased, or are their losses driven by betting on events that are inherently unpredictable? We confirm the existence of favourite-longshot bias in soccer in this sample, but find another surprising feature of betting on longshots. As measured by the Brier score, bookmakers’ odds were better predictors of longshots than favourites, suggesting another potential channel whereby bettors’ preference for betting on longshots may cost them dearly.