FanPost

Super Prediction Action Machine - 2018, Week 6

Well that was a pretty chalky weekend. It's also my wife's birthday so I watched less football that one might imagine. Instead, I mulched our flower beds (which took 6 hours but only cost me $12 in mulch from the local recycling center), took the wife out for Japanese, and spent the evening watching scores on my phone and drinking beer next to the fire pit. #WINNINGATLIFE. Anyway, just because the Hawkeyes didn't play, doesn't mean there can't be any change in the predicted outcomes of the games. Other teams played, and their performance can change the numbers. Let's see how this weekend's outcomes influences SPAM's predictions, shall we?

If you're new to SPAM, many years ago I built a rudimentary statistical model of how Iowa's football seasons go. It completely ignores win/loss records, which don't seem to matter at all when teams play Iowa, and instead looks at yards gained and yards surrendered, and the relationship between those numbers and total points scored. It then predicts how many points Iowa's offense will score against each opposing defense, and how many points Iowa's opponents will score against Iowa's defense.

Here's how the breakdowns have changed over the season:

Week 1: 8-4 (L to WI, MN, PSU, Purdue)

Week 2: 10-2 (L to WI, MN)

Week 3: 10-2 (L to WI, MN)

Week 4: 10-2 (L to MN, PSU)

Week 5: 9-3 (L to WI, MN, PSU)

I now present to you updated SPAM predictions before Week 6.

Again, no significant change. SPAM still thinks Iowa should have beat Wisconsin, but has the Hawkeyes losing to Minnesota in a squeaker, and losing to Penn State by doubel digits.

I may have found a flaw in SPAM's calculations, however, which cuts both far and against. Basically, Iowa, Wisconsin, and Minnesota all have lower-than-expected offensive statistics, but higher-than-expected defensive statistics because they've played fewer games. For example, go look at ESPN's total defense statistics, and you'll see that Iowa is ranked #1. However, this is because their ordinal rankings are based on total yards surrendered. Since Iowa has played 1 fewer game, there's missing yards there. I've been calculating yards per game based on total yards and what week it is, but that's producing artificially deflated numbers. Flip that for offense. I'm not going to tweak it now, but maybe next season I will.

Note also that these numbers have home field advantage baked in. Minnesota is actually a dog to Iowa, but home field flips the L to a W. Other than Wisconsin, there were no other games on the schedule where home field advantage flips the outcome.

Spam says 9-3. What say you?

  • 1-0: Iowa over Northern Illinois (CORRECT!)
  • 2-0: Iowa over Iowa State (CORRECT!)
  • 3-0: Iowa over UNI (CORRECT!)
  • 3-1: Iowa over Wisconsin (WRONG!)
  • 3-2: Minnesota over Iowa, 23-21
  • 4-2: Iowa over Indiana, 27-23
  • 5-2: Iowa over Maryland, 27-23
  • 5-3: Penn State over Iowa, 38-21
  • 6-3: Iowa over Purdue, 26-23
  • 7-3: Iowa over Northwestern, 30-18 (eighteen!?)
  • 8-3: Iowa over Illinois, 28-23
  • 9-3: Iowa over Nebraska, 35-14

Unless otherwise expressly indicated by BHGP editors, this FanPost is strictly the viewpoint of the author and is not endorsed by BHGP in any way.