Posts
[EDIT: After writing this, Jacob Pullen and Curtis Kelly were ruled ineligible fro Kansas State. So, umm, never mind. KSU is no longer a “good” offense, so UNLV should be fine.]
[This post is really just an excuse to use a spreadsheet I whipped up, which takes a year and a school as input, and automatically creates a chart like the one below.]
UNLV tips off against Kansas State in Kansas City a mere hour from now. The Rebels started off the year with a bit of hype, and lived up to it at first, topping Pomeroy ratings #10 Wisconsin at home by 3, and blowing out some scrubs. But they’ve stumbled over the last 10 days, following up a very understandable loss at #13 Louisville with a much less forgivable home loss to #100 UC Santa Barbara. In truth, their problems started to show up 2 games previous to the Louisville loss, when they beat a poor Nevada (#202) team by only 12, but sometimes a W can have a lipstick-on-pig effect.
It’s clear that a 12-point win over #202 isn’t fantastic (Pomeroy had predicted UNLV by 18), but sometimes it’s hard to get a feel for exactly how good/bad a performance is. We can use opponent ratings to shed some light on the issue by using Pomeroy’s efficiency prediction formula:
Predicted Offensive Efficiency = ([Team Adj Off] + HFA) * ([Opp Adj Def] + HFA) / [Lg Avg Eff]
For each game that a team has played, we can replace the predicted efficiency with the team’s actual raw efficiency in that game, plug in their opponent’s rating, the league average rating, and the appropriate home field advantage (+/- 1.4% for each team, in a normal H/A situation) , and solve for [Team Adj Off]. That gives us the team’s single game adjusted offensive efficiency rating – essentially, this is how efficient a team would have been if they played exactly the same, but were facing an average opponent on a neutral court. We can do the same for defense, and from those two numbers we can calculate the efficiency margin (which I find more intuitive to use) or Pythagorean rating (which Pomeroy uses to rank teams). As a last step, we can take the single game Pythagorean rating, pretend that’s how the team has played the whole year, and see where they would rank in the Pomeroy ratings. That allows us to say, for example, UNLV played like the #232 team in their loss to UC Santa Barbara. The chart below shows this “played like” rank for each of UNLV’s games so far. (It also shows, from left to right, the location, opponent, opponent ratings, raw game efficiencies, and adjusted game efficiency ratings.)
You can see that UNLV played great through the first 7 games, and then has really struggled over the last 5. They’re still rated #22 by Pomeroy, but taking a close look at the individual game adjusted ratings reveals something interesting. Their overall defensive rating so far is 89.7, but that’s largely because of ridiculous defense in a few games against poor offensive teams. Here are the teams they’ve managed to post a sub-90 defensive rating against so far, along with those teams’ offensive ratings (keep in mind, average is 100):
- UC Riverside (90.6)
- SE Louisiana (91.7)
- Illinois State (97.1)
- Southern Utah (90.9)
That’s it. Their 30th ranked defensive rating comes in large part from really cranking the screws on the little guys. If they want to compete in the Mountain West this year, that’s going to have to change. And if they want to win their semi-road game tonight against Kansas State (offensive rating of 106.7), that’s going to have to change. I’m not saying it definitely won’t, but I’d say KSU has a better chance than UNLV of bettering their Pomeroy prediction (KSU by 1) tonight.
