Hot Reads: Score Another One for Simple Ranking Systems
Photo Credit: John S. Peterson

Hot Reads: Score Another One for Simple Ranking Systems

December 17, 2018

I love a nice and simple rating system. I look at the complicated ones too, but when the output is just a number –– Alabama is a 103.05 in Sagarin right now for example –– I think you lose a little something. But if you can come up with a rating system that's accurate and also produces a number you could tally using some basic stats and a calculator, then you're really hitting on something.

Such is the case with VolleyMetrics' Conference-Adjusted Combined Offensive-Defense rating (CACOD). I wrote about this system last year at this time as it had national-champion Nebraska as the third-best team in the country entering the NCAA Tournament instead of the five-seed the Huskers actually received. The only team in the top four of last year's CACOD that didn't make the Final Four was No. 2 Texas (which was also a victim of being under-seeded and lost to Stanford in the Regional).

This year's pre-tournament CACOD had Nebraska as the best team in the country but the Huskers received the seven seed in the tournament. Stanford was No. 2 (top overall seed), BYU No. 3 (four-seed) and Illinois No. 4 (three-seed). Those ended up being the Final Four teams in 2018. Nice job, CACOD.

But here's the real beauty of this system, it's just a measure of offensive ability relative to defensive ability, as measured by hitting percentage, with a conference-strength adjustment. The first part of that equation is elegantly simple. You're looking for the ratio of hitting percentage to hitting percentage allowed. If your favorite team hit .300 for the season and allowed opponents to hit .150 it would a) be a really good team, and b) have an unadjusted ratio of 2.0.

The only tricky part here is coming up with the conference-strength adjustments, but that could be achieved with some general knowledge of the volleyball landscape and a little trial and error. For example, BYU had the highest unadjusted ratio in the country entering the tournament. The Cougars outhit their opponents .318-.139, a ratio of 2.29. Playing in the West Coast Conference, however, gave that raw ratio no boost (a multiplier of 1.0), so 2.29 it was.

Nebraska had an unadjusted ratio of 1.99 (.259-.130), but the Big Ten, the toughest volleyball conference in the country, has a 1.25 multiplier in this system. That gave Nebraska an adjusted ratio of 2.49, best in the country entering the tournament. The Huskers, with their seven-seed and a 24-6 record entering the tournament, would've been a slight favorite to come out of their Regional, beat Illinois in the semifinal and then beat Stanford in the final. It almost went down exactly like that.

But the takeaway here isn't that CACOD was really accurate in 2018, and darn good in 2017, too (though I do need to remember to look at this next year before the tournament instead of after, as I did the past two seasons). Rather, the simplicity of this system underscores a key point about how we talk about stats in general.

As Oklahoma made its late-season push for a spot in the College Football Playoff, you no doubt heard that the Sooners allowed more than 32 points per game (they rank 96th at 32.4 currently). That doesn't sound like championship-level stuff, but it's all relative. Oklahoma averages 49.5 points per game; 17.1 points per game is a good differential no matter where it falls on the scale. How many points the Sooners allow isn’t irrelevant, but it’s only half the game.

Lately I've been toying around with a points-per-play differential as a simple way to measure teams. Like CACOD and its use of hitting percentage, making this a per-play rather than per-game measure expands the sample size. Instead of measuring 12 or 13 games, you're measuring nearly 1,000 plays on each side of the ball. The question we're answering here is pretty simple: How much better (or worse) is a team than its opponent on every play on average?

I did that for the 2018 football season before the bowl games started. Here was the top 10:

1. Alabama .492 (.719 points per play on offense, .227 on defense)
2. Clemson .429
3. Utah State .354
4. Fresno State .316
5. Oklahoma .313
6. Appalachian State .307
7. UCF .303
8. Georgia .296
9. Mississippi State .272
T-10. Penn State/Michigan .239

Come up with a strength-of-conference or, more likely for football, a strength-of-schedule adjustment and you might have something there. Take out the G5 teams for a moment — though the top three here all won their bowl games on Saturday going away — and the top four P5 teams via this method are Alabama, Clemson, Oklahoma and Georgia. That was almost the Playoff. Notre Dame, which made it in, ranked 12th overall with a points-per-play differential of .222. Ohio State, which was No. 6 in the final CFP rankings, was all the way down at 28th (.156).

I plan to play around with this more in the offseason, not because the world needs another metric for ranking teams or because this one is unique, but because it gets to what I think is the central question: How dominant is a team on average?

That's what CACOD gets at and the last two years at least it was telling you that Nebraska volleyball was advancing out of its Region even if the seeding indicated otherwise.

The Grab Bag

Today’s Song of Today

  • Never miss the latest news from Hail Varsity!

    Join our free email list by signing up below.

Share via
Copy link
Powered by Social Snap