Filed under:

# Soccer: How does game-level passing predict shot totals?

I've posted some tables of total passes in world cup games so far, but I'm sure many of you are left wondering what they mean. After all, teams can work the ball around in their own zone, or in non-dangerous locations and rack up big pass totals without even threatening to score.

I looked at the relationship between game-level passing ratios [Passes For/(Passes For + Passes Against)] and shot ratios [SF/(SF+SA)] for different pass locations. This table shows the percentage of shot ratio explained by the pass ratio:

 All Def Zone Off Zone Weighted R^2 0.33 0.07 0.39 0.44

The columns are all passes, defensive zone passes, offensive zone passes, and passes weighted by the likelihood of completing them. Defensive zone passes - as you probably expected - don't really contribute to shooting. But we see a slight improvement in the predictive power of passes as we go from 'all passes' to 'offensive-zone passes' to the final probability-weighted version.

So overall, completed passes are a reasonable predictor of the outcome of the game. If we look at EPL games over the last four years and split them into deciles by weighted pass-percentage (the best predictor from above), we can see that the general trend is for teams that outpass to also outshoot and outscore their opponents:

 %ile Pass% Shot% GF/G GA/G Win% 5th 32.2 38.3 1.16 1.38 415 15th 40.0 44.8 1.07 1.18 448 25th 44.7 48.5 1.47 1.06 659 35th 47.9 52.3 1.29 0.99 628 45th 51.0 55.6 1.44 0.94 701 55th 54.0 55.8 1.45 1.05 654 65th 57.1 59.6 1.53 0.95 721 75th 60.4 62.7 1.49 0.96 707 85th 65.0 67.2 1.70 0.82 811 95th 72.3 74.3 2.00 0.78 868

This chart is from the home team's perspective and they tend to win even when they get outshoot - home-field advantage is huge. Games where the passing totals are close are obviously tossups, but if the passes are lopsided, the game outcome should be too.