Summer is here and the time is right...for crunching some numbers related to college football. Over the years, I have developed a rather complex system for analyzing an upcoming college football season. Now that we are into the month of July and the preseason magazines have hit the stands, it is time to break out the slide rule and see what insights we can gain about the 2021 season.
I have explained my methodology in detail previously. Briefly, I have developed a power ranking system for college football that does a decent job of projecting point spreads, and therefore also projecting win probabilities. Using this data, it is possible to (among other things) simulate an entire college football season in order to estimate various season odds. Systems such as ESPN’s FPI make similar calculations in the preseason.
One of the key inputs to my model is an estimate of relative strengths of each team, which in the preseason I create from the collection of preseason magazines and websites that rank all 130 FBS teams. My calculations include a subtle, but important twist that I believe make my predictions more accurate and useful than the ones generated by the FPI and other systems. I consider the historical uncertainty in the preseason rankings themselves.
I have recently updated my spreadsheets and started running the numbers for the 2021 season. Over the coming weeks, I will report here on what I have found. Today, I will dig into the simulation results for the Michigan State Spartans. After that, I will take a broader look at the Big Ten Conference. I then plan to summarize the rest of the Power Five, then the Group of Five and predictions of the teams that will make the College Football Playoff and play in the New Year’s Six bowls games. Finally, I will offer this season’s first dose of Bad Betting Advice based on some of the season odds.
What to Expect from Michigan State in 2021
Well, right off the bat I have to say that the preseason magazines are not at all optimistic about the Spartans’ chance in Coach Mel Tucker’s sophomore campaign. As a result, the math is not particularly kind.
The preseason rankings for the Spartans ranges, as ESPN’s FPI and Bill Connelly’s S&P+ system have Michigan State in the mid-50s, while Athlon and Lindy’s have MSU in the mid 70s. Phil Steele is the least optimistic and has MSU ranked No. 87 out of 130 FBS teams. As a result, my “consensus” preseason ranking for the Spartans is No. 74. Not great, but it is what it is.
Based on this preseason ranking, and the rankings of all of Michigan State opponents, I am able to project point spreads and win probabilities for all 12 of MSU’s scheduled regular season games. Note that these calculations do take into account the known uncertainty of the preseason rankings. I therefore consider these numbers to be the most accurate ones that we can generate, knowing what we know today (which is honestly not much). Nonetheless, here is Figure 1.
Based on this Figure, Michigan State’s games can be roughly divided into four different groups. First, there are the two games (and I must note, only two games) where MSU is projected to be the favorite: the Week Two game against FCS foe Youngstown State (MSU -23, 95 percent odds to win), and the Week Five game against Western Kentucky (-10, 77 percent). MSU projects as a double digit favorite in both games, but that WKU spread is a bit tighter than one might expect. I should mention here that I do not track FCS power rankings at all, so the Youngstown State line is very speculative.
The next cluster of games is composed of three contests where MSU projects to be a narrow underdog with a line of less than five points and with odds of a victory between roughly 40 and 50 percent. Those games are the Week Four clash with Nebraska (+4.0, 39 percent), the Week Six game at Rutgers (+2.5, 43 percent) and the Week 11 game versus Maryland (+0.5, 49 percent or basically a push).
While it would obviously be much better if Michigan State was projected to be the favorite in each of the five games mentioned so far, each of those games is clearly winnable, even if the Spartans actually are just the 74th best team in the country. The odds shown above suggest that MSU is mostly like to win three of those five games listed so far.
The next batch of games form another trio. In this case, it includes games where MSU is projected to be between a five and 10 point underdog, with odds to win in upset fashion between roughly 25 and 30 percent. Those include the Week One road trip to Evanston, Illinois to face Northwestern (+9.0, 26 percent), the Week Eight game against archrival Michigan (+7.0, 32 percent) and the Week Nine contest at Purdue (also +7.5, 31 percent).
The early lines for Michigan State’s game against Northwestern are closer to seven points, which suggests that Vegas has a slightly higher opinion of MSU than the preseason magazines. Or, perhaps they have less faith in Northwestern. It is hard to say. Either way, the probabilities do suggest MSU is likely to win one of the three games listed above. I will leave it for you to guess which game has my vote.
Finally, in the remainder of MSU’s games, the Spartans project to be double-digit underdogs in four games. Those include the Week Three game at Miami (+17.5, 11 percent), the Week Seven game at Indiana (+15, 15 percent), and the final two games of the season at Ohio State (+22.5, six percent) and versus Penn State (+13.5, 18 percent). Despite the long odds, the math suggests the Spartans have a 50-50 shot to steal a victory somewhere in this group of four games.
The same set of data from Figure 1 can be parsed in a different way to show the odds that MSU may win anywhere from zero to 12 games. The result of this calculation is shown in Figure 2.
In this case, I show two sets of data. The solid blue bars are derived by taking the uncertainty of the preseason rankings into account. In other words, it uses the same data used in Figure 1. The other set of data (the green striped bars) assumes that the preseason rankings are correct. Estimations that you may see from places like ESPN will resemble the striped green bars. My solid blue bars are likely more accurate.
In both cases, the total expected numbers of wins (the peaks of the distributions) are very similar. If the preseason rankings are correct, MSU would be expected to win a total of 4.1 games. When I factor in the uncertainty of these rankings, MSU expected win total increases slightly to 4.4 wins (because there is more upward potential when a team is projected to be below average). The distribution also broadens noticeably.
This data is very consistent with the analysis that I gave for Figure 1. MSU is most likely to win three of the easiest five games on the schedule, and one of the games between Northwestern, Purdue and Michigan to yield a final record of 4-8.
But, the data in Figure 2 allows us to give some additional season odds. For example, if the preseason rankings are accurate, the Spartans would need to win four games in upset fashion in order to become bowl eligible. I calculate those odds at only 14 percent. In this scenario, the odds of MSU winning seven games or more is less than four percent, and the odds of going 8-4 or better are less than one percent.
If you are a fan of the Spartans, those numbers are a bit depressing. Fortunately, Figure 2 gives some insight into the likelihood of some more optimistic scenarios. When the historical uncertainty of the rankings are considered (i.e. we consider the odds that Michigan State will be better than expected) MSU still has about a 50 percent chance to win between three and five games. However, the odds for bowl eligibility double to 28 percent.
The odds for Michigan State to win seven or more games are around 16 percent and an 8-4 record or better has odds of eight percent. Even a 9-3 record has a three percent chance. So, if you happen to be an optimist, there is the fuel for your fire. For the pessimists out there, my simulation also suggests that MSU has about an 18 percent chance to win two games or less.
The analysis of Figure 2 begs some additional questions. If we allow ourselves to imagine scenarios where Michigan State is better (or worse) than the No. 74 team in the country. How good does MSU need to be such that a 6-6 record (or better) can be expected? With the power of #math, we can answer that question as well, as I have shown below in Figure 3.
Here, I am recalculating the expected win total for Michigan State if I artificially increase (or decrease) the Spartans’ power ranking to match that of other notable Big Ten or Power Five teams. As discussed in detail above, the nominal Spartan preseason ranking of No. 74 projects MSU to be favored in just two games, but with an expected win total of just over four games (see the largest data point in Figure 3).
The encouraging news is MSU does not have to be significantly better than predicted in order to put themselves into position to be bowl eligible. If the Spartans can play more like a top-50 team (such as where Purdue is projected) the expected win total rises to over 5.0. The “break-even point” for bowl eligibility is for the Spartans to be a borderline top-40 team, similar in strength to Northwestern.
This make the opening weekend game in Evanston particular intriguing. If MSU can win that game or at least be very competitive, it will give me a lot more confidence that this team get to or above .500. If the Spartans struggle, it could be a very long year...unless, of course, the Wildcats are a lot better than we think now.
If we allow ourselves to think even more optimistically, a 7-5 record begins to look more feasible if the Spartans can play like a top-35 team (such as Minnesota), but the curve gets a bit steeper after that. In order to expect a record of 8-4 or better, the Spartans would need to be a top-20 team or better.
A quick look back at Michigan State’s schedule illustrates why. The Spartans face four teams projected in the top-25 (Miami, Indiana, Penn State and Ohio State) and must play three of those teams on the road. Even a team as strong as borderline top-10 Wisconsin squad might struggle to get to 10-2 with MSU’s schedule and even Alabama’s expected win total just clears 10.0 with MSU’s schedule.
For members of Team Pessimism, I will note that even if Michigan State is actually a little worse than predicted, the expected win total still stays above 3.0, even if MSU flirts with mere top-100 status.
Now that we have a clear understanding of what to expect from the Michigan State Spartans, the next most logical step is to take a closer look at the rest of the Big Ten. Until next time, enjoy, stay tuned, and Go Green.