Five Factor Box Scores: Baylor Beats Duke 52% of the Time

Every week, Bill Connelly (the creator of S&P+, the best against-the-spread picker that exists) releases what he calls his “Five Factor Box Scores,” which are the five statistics most significantly correlated with whether a team wins a game. You can read all about that here.

Those stats are:

1. Yards-per-play, which functions as a measure of explosiveness.
2. Success Rate, which functions as a measure of efficiency. Success Rate is the % of the time a team gets “50% of needed yards on first down, 70% of needed yards on second down, or 100% of needed yards on third or fourth down.”
3. Average starting field position
4. Points per trip inside opponent’s 40 yard line
5. Turnovers

Finally, these stats encapsulate both offense and defense, and the important thing is advantages over the other team. I.e., you can have amazing field position starting at the 40 yard line every drive, but if your opponent does too, it doesn’t really matter. What matters is the difference. Same with explosiveness, etc.

So, looking at these stats, the numbers gave Baylor a 51.9% chance of winning yesterday. What this effectively means is this: if you go back in history and chart every game where the numbers look like this, the team with Baylor’s stats win 51.9% of the time, and the team with Duke’s stats win 48.1% of the time. It was a coin flip, basically.

But how? As we know, a major reason Baylor lost is turnovers. Baylor had two of them, one which was returned for a touchdown and the other which gave Duke the ball inside Baylor’s 40 yard line. And when teams win the turnover battle by a margin of two? They, on average, win 78.6% of the time.

Which leads to the next area where Baylor got trounced: average starting field position. Duke started on the 37 and Baylor averaged starting on the 22. That’s a huge disparity. Teams, like Duke did, that have an average starting field position of 15-20 yards higher than their opponent win on average win ~90% of the time.

And finally, as we all know, Baylor only had 5 scoring opportunities, compared to Duke’s 8. Furthermore, Duke scored 4.3 points per scoring trip, while Baylor scored 4.2. When teams, like Duke, have more scoring opportunities and a higher scoring average? They win 98.4% of the time.

OK, so what gives? Duke destroyed Baylor in 3 areas of the game, how are the numbers saying that Baylor, more often than not, should have won?

First, even though Duke had the big-big plays in this game , explosiveness was basically a wash. Duke averaged 5.39 points per play, and Baylor 5.41. Maybe it is a hyper-focus on negativity, but we as fans only remember Duke’s big plays and forget that we had several of our own. My other theory is that another team’s big plays are much more salient and memorable when your defense is otherwise bottling them up. Anyway, the explosiveness measure didn’t really favor one team or the other; so again, what gives?

Baylor had a huge advantage over Duke in efficiency: Baylor had a success rate of 46% (very good) while Duke had a success rate of 35% (middling, below average). Recall that success rate is “50% of needed yards on first down, 70% of needed yards on second down, or 100% of needed yards on third or fourth down.” On average, when one team has a success rate 10-20% higher than the opposition, they win the game 91.5% of the time.

This is pretty incredible when you realize that these numbers include the 5 or 6 drops that absolutely handicapped the Baylor offense. And it furthers my theory that the drops were so enraging because the Baylor passing game was the only thing Baylor had going for it this game. There was an implicit pressure on the WRs, because without them we were doomed (and indeed, the drops likely doomed any “real” chance Baylor had at winning the game). And not just within the offensive context, but the entire game. Baylor’s field position and turnovers had put the team in a massive hole, and only the passing game could carry it out.

One caveat before I sign out here. These stats aren’t meant to show that Baylor should have won the game, or as an excuse for poor play, etc. etc. Stats are great because they are dispassionate, but no statistics are perfect. They can’t account for everything, such as when the turnovers occur, and when you give up explosive plays, etc. I just thought this was an interesting result, because as I predicted in the post-game thread yesterday, I figured Baylor would have a post-game win expectancy of ~40%. The 52% figure is merely interesting, and hopefully something we can look at to realize that yesterday wasn’t all bad.

Baylor has a lot to clean up. They can’t rely on such disparities in offensive efficiency for every game. If Baylor continues to turn the ball over (while getting no turnovers itself), have crappy starting field position, and shoot itself in the foot with unforced errors, they’re not going to win many more football games this year. But the flipside is that if Baylor can cut down on its turnovers, force a few of its own, and reduce unforced errors, the offense should be able to carry this team to quite a few more victories this year.