Machine Learning - Week 6

Oct 3, 2022 - 7:43 PM

Where We’ve Been

Not familiar with BizarroMath? You’re in luck; I’ve launched a web site for it where you can get an explanation of the numbers and browse the data. Prior week entries in this series are linked below.

Week 1

Week 2

Week 3

Week 4

Week 5

All lines courtesy of DraftKings Sportsbook as of 10:00am CDT, Oct. 3, 2022.

Last Week Recap

How It’s Going

This week, BizarroMath went 2-5 ATS and 5-1-1 O/U. You people went 1-6 ATS and 3-3-1 O/U. My algorithm did better than the BHGP community in a rough week that included a lot of low-scoring upsets. Tragically, Iowa-Michigan wasn’t one of them.

Combined with the prior record of 15-16 and 11-20, respectively, the algorithm is now 22-24 ATS and 19-26-1 O/U on the season after five weeks. That’s Not Great. But, most algorithms did poorly this weekend. We’ll judge the body of work at season’s end.

Saturday, Oct. 1, 2022

#4 Michigan at Iowa

Vegas Said: MI -10.5, O/U 42.5

Actual Outcome: MI 27, IA 14 (MI cover, under)

BizarroMath: Iowa cover (MISS), under (HIT)

You People: Iowa cover (MISS), under (HIT)

Hot Take: Of all the things you expected to see in this game, where on your list was Spencer Petras having a better stat line than J.J. McCarthy and Michigan struggling to contain Iowa’s passing attack?

Illinois at Wisconsin

Vegas Said: WI -7.5, O/U 44

Actual Outcome: IL 34, WI 10 (IL cover, push)

BizarroMath: IL cover (HIT), under (n/a)

You People: WI cover (MISS), over (n/a)

Hot Take: I predicted an Illinois win here, and I said that when it happens, you could stick a fork in Wisconsin, but I did did not expect Wisconsin to stick a fork in Head Coach Paul Chryst, whom I recently and incorrectly characterized as having ceiling that is too low to win championships but too high to get him fired.

Indiana at Nebraska

Vegas Said: NE -3.5, O/U 61.5

Actual Outcome:: NE 35, IN 21 (NE cover, under)

BizarroMath: IN cover (MISS), under (HIT)

You People: IN cover (MISS), under (HIT)

Hot Take: Nebraska is tied for 1st in the B1G West with the other 5 members of the Division not named Wisconsin.

Purdue at Minnesota

Vegas Said: MN -10, O/U 67.5

Actual Outcome: MN 10, PUR 20 (PUR cover, under)

BizarroMath: PUR cover (HIT), under (HIT)

You People: MN cover (MISS), over (MISS)

Hot Take: I warned you that the The Flecking was nigh.

Rutgers at #3 Ohio State

Vegas Said: OS -41.5, O/U 59.5

Actual Outcome: OSU 49, RUT 10 (RUT cover, under)

BizarroMath: OSU cover (MISS), over (MISS)

You People: OSU cover (MISS), over (MISS)

Hot Take: My gut told me Rutgers would cover in this game, just based on how tough they played Iowa, and my algorithm has consistently gotten OSU games wrong this year.

Northwestern at #14 Penn State

Vegas Said: PSU -25, O/U 52.5

Actual Outcome: PSU 17, NW 7 (NW cover, under)

BizarroMath: PSU cover (MISS), under (HIT)

You People: PSU cover (MISS), over (MISS)

Hot Take: PSU won a game by double digits in which they turned the ball over 5 times, but you can do that when you also throw down 220 yards rushing and enjoy a 37:28 to 22:32 TOP advantage.

Michigan State at Maryland

Vegas Said: MD -7, O/U 60.5

Actual Outcome: MD 27, MSU 13 (MD cover, under)

BizarroMath: MSU cover (MISS), under (HIT)

You People: MD cover (HIT), under (HIT)

Hot Take: In statistics, the standard deviation of a sample size of 1 is infinite, which is why you don’t make decisions over the disposition of millions of dollars in Head Coach salary based on one good season.

Iowa Recap and Game-by-Game Projections - Week 6

My top takes on the Michigan game. Again, I write this up before I read the articles and listen to the podcasts, so if I’m repeating what others have also said, mea culpa.

• Shade will be thrown at Petras over this game, but he was 21-31 for 246 yards with 7.9 YPP for 1 TD and 0 picks. Those aren’t great numbers, but they’re numbers we can work with. Still too many balls thrown to a 9 foot tall ghost only he can see.
• Michigan avoided passing the ball, correctly understanding that the only way they’d lose this game was on special teams gaffes or turnovers. So. They just didn’t do that.
• Iowa had 35 net rushing yards. That’s not a winning number for Iowa football, period, much less against a Top 5 opponent.
• Iowa had 5 penalties for 59 yards. In a home game. Seemingly every flag was thrown both away from the play and after a big Iowa gain. Again, not a winning number for Iowa football against a Top 5 opponent.
• Speaking of flags, the officiating in this game was awful. Just bloody awful. I don’t think it impacted who won, Iowa was thoroughly outclassed across the board. But that was some of the worst officiating I’ve ever seen in a Division I game that did not involve an unranked Iowa State holding serve with a ranked Big 12 opponent. It makes watching the game less fun, and watching Iowa football is already pretty dull. The Big 10’s on-going quality control challenges with officiating will eventually change the outcome of a game that matters. The league just signed a ka-jillion dollar TV deal, can’t we hire a few crews of full-time officials?
• This game went exactly the way the numbers suggest. Iowa’s offense is only worth about 10-15 points a game and if the defense can’t hold opponents to less than that, we have to get scores elsewhere. If that doesn’t happen, Iowa doesn’t really have much of a chance. We saw this script play out against Wisconsin and Purdue last year and we’re going to keep seeing it this year.
• Connor Colby is struggling. He’s been struggling. I noticed it during the ISU game, he’s a step behind the defender, guys are blowing right past him. About all he can do is tackle the defender and hope he doesn’t get flagged. If I can see that, B1G defensive coordinators can see it. There’s talent there, so I’m not sure what’s going on with him, but it’s hard to watch.
• Remember during last year’s B1G Championship Game when Mike Golic observed on Twitter that Iowa seems to have a lot of 3rd down plays designed to get exactly 1 yard less than they need? We’re still using that playbook. Sam LaPorta ran multiple routes where he stopped or went horizontal 1 yard short of the sticks. Even when there was enough space to go ... one yard further. This trend goes back at least as far as Greg Davis’s first year in 2012, so it crosses coordinators. I cannot fathom that Iowa is intentionally coaching guys to NOT run far enough downfield to pick up first down yardage before turning for the catch, but this has been going for A DECADE and this still happens multiple times per game. What other explanation is there?

Iowa Game-by-Game Projections

After figuring in the Michigan game, BizarroMath has the Hawkeyes offense at 24.29 BOA (blended, opponent-adjusted) PPG, which is +4.15 from the OA PPG of 20.14, which is itself 1.39 up from the raw PPG of 18.75. Thus, last year’s comparatively gaudy BOA 28.44 is still pulling the team up, as are Iowa’s 14 points worth of defense scores. On defense, Iowa has finally played an excellent offense and held them in check (relatively speaking), and the OA numbers are starting to adjust downwards. The defense has a raw PPG of 11.75, which OAs -2.97 to 8.78 PPG. But, blended with last year’s BOA of 16.78, gets pulled back up to 12.78.

Again, the BOA numbers are used for projections, but I could be talked into providing multiple sets of projections based on comparing raw or only OA numbers without blending in last year’s data. Let me know if that’s something you’re interested in seeing. I can also do composites of those numbers.

The delta between Iowa’s offensive and defensive numbers is thin, but still positive, which is good. The raw delta is exactly 7 points, and while you might think that reflects the quality of competition, when you figure in opponents adjustments, the OA delta gets bigger, meaning Iowa offensive performance is better than it looks, given the competition. But, that is based on large part on having just played Michigan, and I think the season will bear out that their defense is overrated.

Last week, Iowa was favored in just two remaining games. You can add to that a third: Wisconsin. The Hawkeyes are favored in the math by 0.8 points, for a win% of 51.38% vs. the Badgers.

Ohio State is going to be a blowout. Just get comfortable with that.

The lines remain tight against Illinois, Purdue, and Minnesota, all of which hover in the 40’s. Those are all highly winnable games, and Iowa could easily be 0-3 or 3-0 or anything in-between in those matchups. The Hawkeyes are going to have to outplay the chalk to get to a decent bowl. Las Vegas sounds good.

Iowa Season Projections

The key column I look at here is the Floor column. This is the cumulative chance of getting to that record. The machine shows a 75-80% chance of hitting 6-6, but less than a 50/50 chance of hitting 7-5. This is where your O/U for the rest of the season is. Really no big changes or surprises here. We began the season is the 6-6 to 7-5 range and there we remain. 8-4 is your realistic ceiling, maybe 9-4 if Iowa can win a bowl game. That would be a massive achievement for this squad.

The Championship Race

Well this got weird fast. Last week I mentioned that the swings are wild once we get into heads-up division play, and this weekend bore that out. Losses by Wisconsin and Minnesota to divisional foes dropped them to the 5th and 3rd spots, respectively, and elevated the victors.

If you’re going to lose a conference game, lose to a good team from the East. Purdue now stands astride the Big 10 West, with Illinois nipping at its heels. Iowa’s 1-1 conference record technically puts the Hawkeyes in a 6-way tie for first place with everybody not named Wisconsin, but the Hawkeyes don’t have any heads-up losses to division foes yet, so Iowa’s fate is still in its hands for one more week, but a loss to Illinois next week will alter that calculus. Bigly.

The conference championship race remains largely academic in my mind. If either Michigan or Penn State can beat Ohio State, they’re in the driver’s seat, assuming they and Ohio State otherwise hold serve. But Michigan is looking rather mortal and James Franklin’s now well-established ability to field a team that is lesser than the whole of its parts continues to hold the Lions in check.

Week 6 Projections

Friday, Oct. 7, 2022

Nebraska at Rutgers

Vegas Says: NE -3.0, O/U 50.5

BizzaroMath Says: NE -9.03, O/U 56.35

You Say:

Hot Take: Rutgers is tougher than they’re getting credit for but still undisciplined enough that Nebraska might be able to win this by just leaning on the talent disparity.

Saturday, Oct.8, 2022

#4 Michigan at Indiana

Vegas Says: MI -21.5, O/U 57.5

BizzaroMath Says: MI -25.31, O/U 61.78

You Say:

Hot Take: Despite the win, I wasn’t not super impressed with Michigan after watching them play Iowa, but Indiana’s remarkable COVID is starting to look like an aberration.

Purdue at Maryland

Vegas Says: MD -4.5, O/U 59.5

BizzaroMath Says: PUR -0.8, O/U 50.88

You Say:

Hot Take: It seems like once a week we have a BizarroMath upset pick, and this week, it’s the Boilermakers to take out Maryland; but I don’t think BizMa has been right about any of these yet.

Wisconsin at Northwestern

Vegas Says: WI -9.5, O/U 44

BizzaroMath Says: WI -15.38, O/U 44.35

You Say:

Hot Take: The Badgers are 1-6 at Northwestern in the 21 Century; the Northwestern moneyline here is +300; I recite these facts together for no particular reason.

#3 Ohio State at Michigan State

Vegas Says: OSU -25.5, O/U 62

BizzaroMath Says: OSU -28.66, O/U 65.77

Hot Take: In statistics, the standard deviation of a sample size of 1 is infinite, which is why you don’t make decisions over the disposition of millions of dollars in Head Coach salary based on one good season; I’m aware that this hot take was also my recap of last week’s game.

Iowa at Illinois

Vegas Says: IL -3.5, O/U 35.5

BizzaroMath Says: IL -3.81, O/U 26.84

Hot Take: BizMa’s pre-season projections for Iowa are 5-0 on the year, and BizMa had Iowa winning this game, so let’s focus on that, rather than the current prediction.

The Fine Print

Data Source: The betting lines are from DraftKings. Depending on your degenerate web site of choice, your mileage may vary.

Gambling Caveat: I do not bet on sports. I run these numbers because I’m a nerd who loves math and football and this is my idea of fun. This should go without saying, but if you do bet, you should do so based on your own independent assessment, not the half-baked statistical musings of an Internet rando.

---

Odds/lines subject to change. T&Cs apply. See draftkings.com/sportsbook for details.