BYU's Defensive Problems, as Explained by Soccer

-- Soccer finally comes in handy

In soccer there’s a stat model called expected goals (referred to as xG).

What xG does is assign a probability of scoring to each soccer shot attempt. Let’s say there’s a shot from 30 yards away that has 4 defenders and the goalie in the path of the shot. That shot will have a probability of ending in a goal maybe 3% of the time, and therefore is worth 0.03 expected goals (xG). On the other hand let’s say there’s a shot where the goalie stumbles out of position and the shooter has a tap in right in front of the goal with no defenders able to stop him. That shot will go in say 95% of the time and therefore is worth 0.95 expected goals.

At the end of the game, you sum together all the xG created by both teams and can get an idea of which team was better at attacking and defending, regardless of what the actual score was.

The idea of xG is that goals in soccer are so rare that they can skew the game results dramatically. For example, one team can dominate the whole game, take 15 shots, and miss every shot and not score. Meanwhile the other team could get in position for only one shot the whole game but they score it to win 1-0. Who was the better team? Which team is likely to be more successful in future games?

The team that scored the goal gets the win, but it’s likely that the 15-shot team generated about 2 goals worth of shots while the 1-shot team generated only about 0.08 goals worth of shots. It just so happened that the low percentage shot went into the net, while the higher probability shots of their opponent didn’t. Or in other words the 1-shot team exceeded their goal expectation. The 15-shot team underperformed against their goal expectation.

If those teams play 10 times, you’d expect the 15-shot team to win 70 to 75 percent of the time right? You’d consider the future outlook for the 15-shot team as better than the outlook for the 1-shot team as well right? xG looks beyond the results of one isolated game — which can be skewed by one random goal — and gives a predictive element of what you can expect for a team going forward.

What does this have to do with BYU? The defense allowed 14 points. That’s great! If we could guarantee our opponents only score 14 points a game we’ll be unstoppable. But while SDSU only scored 14, the expected goals model would say most teams that gain 400 yards and get to the red zone 4 times will usually score way more than 14 points. SDSU underperformed their expected goals thanks to some factors they control (sucky QB, bad snaps), some factors BYU controlled (interception), and some random bad luck (icy field).

BYU's defense held SDSU to 14 points, yes. But replay that game 10 times, give SDSU the same offensive yardage output, and if a couple breaks go their way they score 24 points. Give a team with a good quarterback that same offensive output and maybe they score 35.

That’s why BYU fans are stressed and concerned about the defense. Points are a snapshot of one game. The trends are predictive of the future. And the trend isn’t great.