Over the past few weeks, I have explained some of the background around my annual, math-based analysis of the upcoming college football season. In that process we have learned how accurate the preseason rankings are (plus-or-minus 25 spots, on average) and investigated what the prognosticators got right last year (Georgia), what they got wrong (Oklahoma and TCU), as well as the influence of ability, schedule, and luck.
With that foundation in place, it is now time to shift focus to the 2023 season. I have now inputted the data from the preseason rankings and completed 100,000 simulations of the upcoming season. I will soon go through the full details of what I learned from these simulations.
For today, I will focus exclusively on what it says about the Michigan State Spartans. We will take a close look at the Spartans' schedule from three different points of view.
Schedule Overview
The best place to start this analysis is with the inputs to the simulation. Each year there are two main parameters needed as inputs: schedule and relative team strengths. Figure 1 below summarizes this information for Michigan State and the Spartans' opponents in 2023.
 |
| Figure 1: Summary of the teams on the Michigan State football schedule in 2023, sorted by average preseason rankings. |
The bars give the preseason ranking for each team from four publications that were available this summer: Phil Steele, Athon Sports, Lindy's Sports, and ESPN's Football Power Index. The final labeled bar in each cluster is the consensus average of the four rankings. This is the number that I use as the basis for my full-season simulation.
The Spartans' preseason ranking has a fair amount of variance this year. ESPN's FPI is the most optimistic, ranking Michigan State as the No. 31 team in the nation, while Phil Steele puts the Spartans down at No. 66. Athlon (No. 52) and Lindys (No. 44) are in the middle. As a result, I have assigned Michigan State with a consensus ranking of No. 48.
I should note the loss of Payton Thorne and Keon Coleman in the spring is a likely source of some of the spread in this data. Phil Steele does account for these departures, but it is not clear if the other services do or do not. For this reason, it is possible that my consensus ranking of Michigan State is slightly higher than it should be.
As for Michigan State's opponents in 2023, Figure 1 starts to show why the Spartans are considered to have one of the toughest schedules in the nation. Ohio State, Michigan, and Penn State are all consensus top 10 opponents, and Washington is also ranked in the top 10 by two of the four publications.
Iowa is projected to be a solid top 25 team this year, and Michigan State travels to Iowa City this year, which will be a challenge. After that, Minnesota, Maryland, and Nebraska are all ranked somewhere in the 30s to the 50s. These teams are all in a similar tier to the Spartans.
Finally, Michigan State faces four teams that enter the season below Michigan State in the rankings: Rutgers, Indiana, and Central Michigan are three of those four teams. The fourth team is the FCS-level Richmond Spiders. Data is harder to come by for FCS team in the preseason, but it is safe to assume that the Spiders are no better than the three lower tier teams mentioned above.
Projected Point Spreads
A glance at Figure 1 provides a qualitative look at the schedule for Michigan State this fall, but what does all this mean for the individual games? In Figure 2 below, I take the information from Figure 1 and translate it into projected point spreads and the odds that the Spartans will win each of the 12 games on the schedule.
 |
| Figure 2: Projected point spreads and victory probabilities for all 12 games for the 2023 Michigan State Spartans. The odds are calculated in two ways: a) assuming that the preseason rankings are accurate and b) using the results of the 100,000 cycle Monte Carlo simulation, which accounts for the known uncertainty of the preseason rankings. |
Figure 2 contains two sets of odds. The solid green bars are the odds derived from the 100,000 Monte Carlo simulation of the full season. This simulation uses the preseason rankings to project preseason power rankings. I use these power rankings to generate point spreads and win probabilities for any potential match-up.
Each simulation cycle is a series of digital dice throws to determine the winner of every game in the full season. The unique aspect of my simulation is that I also account for the uncertainty of the preseason rankings as an input parameter.
The striped, blue bars are the odds generated using the preseason rankings with no correction for the known uncertainty. Each bar also shows a projected point spread which correlates to those odds. The point spreads from these calculations agree within a few points with the early point spreads that I see online for Michigan State's games against Washington (-12 on Draftkings), Michigan (-19), Ohio State (-25.5), and Penn State (-13).
Figure 2 brings into focus the challenge that the Spartans are facing this season. Michigan State currently projects to be a double-digit underdog in five games this year (Washington, at Iowa, Michigan, at Ohio State, and versus Penn State at Ford Field) and over a touchdown underdog for the game at Minnesota. The Spartans have, at best, only a 30% chance to win a full half of the games on the schedule.
Of the six games where Michigan State projects to be favored, four of those games (Maryland, at Rutgers, Nebraska, and at Indiana) all currently project to have spread under five points. The Spartan project to have no better than a 60% chance to win each of those four games.
The only two games on the schedule where Michigan State projects as a clear favorite are the first two games on the schedule against Central Michigan and Richmond. In addition, the actual opening line for the Spartans' Labor Day weekend home opener against the Chippewas (+14.5) is lower than my computer's projection of 19 points. Michigan State has about a one-in-six chance of a catastrophic beginning to the 2023 season.
Win Distribution
Figure 2 shows us the odds that Michigan State will win or lose each of the 12 games on the schedule. But what does this tell us about total wins and losses? There are over 4,000 different ways for the season to play out for the Spartans. Michigan State could win anywhere from zero to all 12 games. Figure 3 gives the odds for each of those 13 potential outcomes.
 |
| Figure 3: Win distributions for the 2023 Michigan State Spartans calculated in two ways: a) using the preseason rankings as is and b) using the results of the 100,000 cycle Monte Carlo simulation which accounts for the uncertainty of the preseason rankings. |
Similar to Figure 2, the results in Figure 3 are calculated using two different methods. The striped, blue bars are the data assuming that the preseason rankings are correct. The green bars are the more accurate odds which account for ranking uncertainty (of both Michigan State and all 12 of the Spartans' opponents).
At a first glance, the two sets of data tell the same story. The Spartans will most likely win five or six games with five being slightly more likely. Four wins is also slightly more likely than seven wins in both sets of data.
This information makes sense if we look back at Figure 2. The Spartans are large underdogs in six games, slight favorites in four, and big favorites in just two. If we sum up the probabilities in Figure 3, the total expected value is between 5.3 and 5.5 wins. This is exactly the win total over/under that has been published online for Michigan State this summer (5.5). The math checks out.
Based on this analysis, a 6-6 record for the Spartans in 2023 would be slightly overachieving, but what is the path to this record? The math suggests that Michigan State will most likely lose five of the six tough games in Figure 2 where the Spartans project to be big underdogs. However, more likely than not, the Spartans will score at least one upset win.
Can the Spartan steal a win at Iowa or Minnesota? If not, is a win against Michigan, Washington, or Penn State at Ford Field a possibility? The odds suggest that the Spartans will claim victory in one of those five games.
The math also suggests that the Spartan will mostly likely win the first two games against Central Michigan and Richmond to start the season 2-0. Therefore, the fate of the Spartans in 2023 will likely hinge on the four games where Michigan State is projected as a narrow favorite: Maryland, Nebraska, Rutgers, and at Indiana.
My calculations say that the expected value of wins in this set of four games is 2.4. The Spartans are likely to drop one of those four games. The good news (maybe) is that the home games against Nebraska and Maryland are the most vulnerable. If the home crowd can carry the Green and White to victory over at least one of those two teams, the odds of making a bowl game increase.
Another way to look at the schedule is that the Spartans need to have a winning record going into the bye week in early October. This means Michigan State will need to (most likely) win the first two games and get a win in one of the next three contests against Washington, Maryland, and at Iowa.
After the bye, Michigan State needs to get the win at Rutgers to get to 4-2. The back half of the schedule is still exceedingly difficult, but the math suggests that winning two games in that stretch is manageable by either beating both Nebraska and Indiana or by stealing a win at Minnesota or over one of the Big Ten East leviathans this year.
Other Scenarios for Michigan State
If Michigan State and all of the Spartans' opponents are as good as expected in the preseason, then the stripped blue bars in Figure 3 are an accurate prediction of what is to come in 2023: there is about a 50% chance that the Spartans will win five or six games and an 85% chance that the Spartans will win between four and seven games.
But there are several other plausible scenarios.
If Michigan State is better than expected and if a few of the Spartans' opponents wind up being worse than expected, Michigan State could see a rebound like the one in 2021. Figure 3 shows that the Spartans have about a 15% chance of winning eight or nine games and a three percent chance of winning 10 or more. A lot would need to break Michigan State's way for this to happen, but the odds of winning eight or more games are higher than the odds of losing to Central Michigan on the first weekend.
There is also a darker, opposite side of this coin for Spartan fans. It is equally possible that Michigan State is going to be worse than the preseason publications expect. In this scenario with this year's schedule, the Spartans could be looking at three wins or fewer. The odds for that scenario are approximately 19%.
We have now explored Michigan State's schedule in detail, but what does this mean for the overall Big Ten race? Where are the Spartans most likely to finish? That will be the topic of the next installment of this series. Stay tuned.
Comments
Post a Comment