clock menu more-arrow no yes

Filed under:

Big Ten Strength of Schedule

New, 37 comments

Not all paths to Indy are created equally

NCAA Football: Big Ten Championship-Iowa vs Michigan State Brian Spurlock-USA TODAY Sports

Howdy everyone! For those of you that follow this site closely, you may have seen some of my previous posts in the Fanpost section. I have an obsession with numbers and statistics, and I enjoy performing a wide variety of sports-related analyses, with a focus on MSU. I am a research engineer by training and I try to use the same basic logic and scientific methodology to analyze sports data. In some corners of the internet, I go by the handle “Dr. Green and White.”

Recently, the editors of this fine website asked me to write for The Only Colors in a more official capacity. So, I will endeavor to deliver some interesting, math-based tidbits and analysis as we work through the upcoming football and basketball seasons.

As we move closer to the start of the 2019 College Football Season, one topic that comes up periodically is the question of strength of schedule. Does Team A have some sort of schedule advantage over Team B? As with many topics, everyone seems to have an opinion on this, but it is actually a bit tricky to try to quantify it. After all, how does one actually measure strength of schedule?

A method that I currently like is to use is based on the expected value of wins (the sum of the odds to win each individual game). I have posted here before that I have developed my own simple algorithm and formulas to project point spreads and win probabilities for any given match-up or set of match-ups in a given season. In this case, I can use this data to calculate the expected number of conference wins for each Big Ten team.

When it comes to evaluating strength of schedule, the trick I use is to manually adjust the strength each team. For example, if I adjust the power ranking of Illinois such that they are the 10th best team in the country, I can easily calculate their expected conference win total using their current schedule (using the preseason rankings to estimate the power ranking of all of their opponents). I can then do the same thing for Iowa, Indiana, and so on. This way, I can directly compare the impact of each team’s schedule on the actual number of games a team might win. That seems better to me than just saying Team A’s schedule is harder than Team B’s.

I can also do a bit of a sensitivity analysis by running this mini-simulation a few times with different values for the assumed power ranking. In the example above, I did it assuming each team was ranked #10; in other words, I am assuming that they are pretty good. But, how does this change if they are ranked #25 or if they are ranked #50? Well, I just so happened to have made just made this set of calculations for the Big Ten. The results are shown here:

I sorted the data based on the #10 ranking from the easiest schedule (most expected wins) to the toughest schedule from left to right. There is a lot to say about this Figure.

First, Nebraska and Minnesota are the obvious big winners in the Big Ten scheduling sweepstakes. For all three data sets, both teams are in the Top 3 in expected wins. This schedule advantage compared to the teams that are at the tough end of the spectrum (the bottom three in the East) is just about a full game, which out of only 9 total seems pretty significant. It is no wonder that many are labeling Nebraska as the favorite in the West and as Minnesota as a possible dark-horse. They have essentially a half-game lead out of the gate over even the #3 team overall, Purdue.

Looking at those West team’s schedules give a big hint as to why the numbers work out that way. Nebraska, Minnesota, and Purdue only play one of the “Big Four” of the East as their cross-over. Minnesota and Nebraska both draw that team at home (Penn State and Ohio State), while Purdue draws Penn State on the road. The other four West teams all play at least two of the “Big Four.” So, that all makes sense.

After those three squads, there is a plateau in the data where the next 5 teams reside where the expected win total only varies by a little over a tenth of a game. Michigan, Northwestern, Iowa, MSU, and Ohio State all have fairly comparable schedule difficulties. Michigan’s schedule is very slightly easier (as they draw OSU and MSU at home) but on balance, their schedule is only 0.1 wins easier, at least if we are assuming everyone is a Top 10 level team.

For the remaining six teams (Illinois, Penn State, Wisconsin, Indiana, Maryland, and Rutgers) their expected win total is a bit lower, by an average of 0.3 games. Four of those teams are generally pretty average to down-right bad, so it is not a surprise that they have a harder schedule. A team like Indiana, for example, has to play all four of the Big Four of the Big Ten East, AND they do not get the benefit of playing... Indiana. So it only makes sense that Indiana, Maryland, and Rutgers have depresses expected win totals.

As for Penn State and Wisconsin, their slightly lower scores are certainly indicative of a slightly harder schedule. Wisconsin is the only West team that faces one the “Big Four” in all three of their East cross-over games. As for Penn State, they also have a lot of tough road games (at Ohio State, at MSU, at Iowa, and at Minnesota.) So, it all seems quite reasonable.

It is also interesting to see what happens to the rankings as the assumed baseline ranking changes. If I run the simulation assuming everyone is ranked 25th or 50th, the expected win totals go down, but they go down at different rates for each team. The expected win total ranking for Michigan, Purdue, Northwestern, and Nebraska goes down if those teams are worse. Basically, their schedules will be relatively harder if they are actually worse than expected. It is a bit of a double-whammy. In contrast, Iowa, MSU, and Penn State’s ranking goes up in this simulation. In other words, if those team are a little bit worse than expected, their schedules are relatively easier. MSU’s actually has the 2nd easiest schedule if the baseline ranking is #50 in the country.

I believe that what this analysis is capturing is a general effect of having the tough games and easy games clustered either at home or away. For a team like Michigan, they have a lot of tough games at home, and not as many on the road. If Michigan is worse than expected, those “winnable” home games may turn into losses and those “easy” road games are no longer so easy. That can snow-ball into a result where UofM just loses a lot of games. That is essentially what happened last year to MSU. On the flip side, MSU’s schedule is the opposite with a lot of tough road games and easy home games. If MSU is worse than expected, those tough road games are probably a lost cause, but the easy road games are still quite winnable. MSU’s floor is likely higher than last year for this very reason. I think the data in that chart above bears this out.

Finally, I should point out that Minnesota seems to be the team with the overall easiest Big Ten schedule this year. While Nebraska’s expected win total is slightly higher at the Top 10 level, Minnesota’s has the easiest schedule at the other two benchmark levels. The Gophers certainly are a team to watch. That all said, ALL of this analysis assumes that the preseason rankings are accurate, and that is certainly not true. It will be an interesting exercise to revisit this analysis after the season is complete and we know how good these teams actually all are.