What a close to the regular season. Just a few weeks ago, on February 17th, the math was suggesting that MSU had only a 4.0 percent chance of winning the regular season Big Ten title and Wisconsin had only a 1.4 percent chance. But, both teams ended on a tear and (more significantly) Maryland ended in a slump, and as a result, it was a very happy weekend in both East Lansing and Madison.
I could probably go on for paragraphs about the significance of the three-peat this year. Back in the summer, I knew that this campaign was going to be a stressful, because they always are when expectations are so high. It looked for a long time like this season was going to end with extreme disappointment. But, Izzo worked his magic again, and now this team has a shot to be special and to achieve some more of those goals that were set back in April.
I could certainly continue to wax poetic, but I am a numbers guy, so I will just give these numbers:
For the full length of the Big Ten season, I has provided numerical updates to handicap the race in real time. Now that the race is over, it is time to take a brief look back at the results and then forward to the next phase of the season: the Big Ten Tournament. Let's attack this by asking three probing questions:
1) In the regular season race, which is more important, luck or schedule?
Throughout the season, I had a running tally of the "luck" accumulated by each team to that point. In this context, I defined "luck" as the difference between the actual wins for a team and the expected value of wins. This essentially measures each team's ability to win toss up games. As an example, if a team played a total of ten games in which the point spread was a pick'em (50 percent chance to win) one would expect that team to win five of those ten games. If that team instead won six games, then they would have +1.0 game(s) of luck.
As I have all season, I now present the final Big Ten standings for 2020, including the calculated luck, with the teams listed in order of their current Kenpom ranking, and numbered with their seed in the upcoming Big Ten Tournament.
There are several interesting takeaways from this table. First, if I go back and check the data from New Year's Day, the expected win total for first place MSU was 14.3 wins and the Big Ten champ was projected to go 15-5. In the final analysis, MSU's expected win total was almost a full game lower than that, which suggests the league was tougher than it was even expected to be in early January.
Second, there is quite a bit a spread in luck. The luckiest team in the conference (Illinois, +3.23) had over a five game advantage in luck over the least lucky team in the conference (Purdue, -2.14) who is actually ranked higher in Kenpom than the very lucky co-champions, the Wisconsin Badgers (+3.21). MSU finished the season almost a half-game on the lucky side, while Michigan was a bit over a game on the not-so-lucky side of the coin. While "luck" is very likely not strictly random (some might refer to it as "grit"), it certainly has a big impact on the final standings.
But, what about the effect of the schedule? Back in December, I posted an analysis comparing the strength of schedule of each team in the conference using a unique method (as far as I know). I made the same expected win calculation as in the standings above for each team, but with a twist. I calculated the expected win total for each team's schedule, assuming that schedule was played by a Big Ten team of average strength, which is the preseason was assumed to be Penn State.
In addition, I also made a similar calculation in which I adjusted the Kenpom efficiency of the average team (Penn State) to be equal to that of the team in question. This was an attempt to correct for the fact that a team like MSU benefits from not have to play MSU, while a team like Nebraska is penalized but not getting to play a team as bad as Nebraska.
Now that the season is over, I made the same calculations again using the current Kenpom efficiencies. In this case, the most average Big Ten team turned out to be Purdue. The charts below show both the preseason and postseason strength of schedule results (both corrected and uncorrected) with the easiest schedules (most expected wins) on the left.
Once again, there is a lot to learn from these graphs. First, they also suggest that the conference as a whole is tougher than expected, as basically every teams expected win total decreased from the preseason to post-season calculations.
Second, it is pretty easy from this graph to identify the teams with the easier or harder schedules. The big winner in this analysis appears to be Iowa, who had a ~0.3 game advantage in schedule over Purdue, Rutgers, and Ohio State. If we look back at the enhanced Big Ten standings shown above, this makes sense, as Iowa and Purdue have a higher expected win total than their Kenpom efficiency would suggest. MSU had a slightly (0.1 game) tougher schedule than the lead pack, while Wisconsin and to a greater extent Penn State, Illinois, and Michigan all had tougher schedules by about half a game.
Third, the entire range of strength of schedule is right at a full game. Iowa did have a better schedule than Michigan by that amount. But, as we already know, the full range of luck is closer to five games. In other words, it is better to be lucky than (to have a) good (schedule).
2) Should MSU fans be upset that we didn't get the No. 1 seed in the Big Ten Tournament?
No.
As the Big Ten race entered the final day, I was able to perform several Monte Carlo style simulations of the full Big Ten Tournament using the most like five scenarios, depending on the winners of the MSU-OSU game, Michigan-Maryland game, and the Illinois-Iowa game. The results are shown here:
I can't tell you the exact error in this simulation, but I think that it is around one percent. So, in other words, the simulations gave MSU essentially the same odds (~23 percent) to win the tournament regardless of whether MSU was the No. 1 seed, No 2. seed or the No. 3 seed.
The reason for this, I believe, is also clear from the enhanced Big Ten standings shown above. Due to the overall parity of the league, there are good teams in all parts of the bracket. Even the 12th best team in the conference (Indiana) is ranked in the Kenpom Top 40. It was going to be tough no matter where MSU landed.
3) Who is going to win the Big Ten Tournament and should I bet on it?
Great question. While I never bet myself, it you are of that persuasion, I perhaps have some data to help. The table below shows the full results of my Monte Carlo simulation of the final Big Ten Tournament bracket, with odds for each team to advance to a certain point.
Based on the odds that I calculate for the championship, I also tabulated the betting break even point for each team. If you see a money line higher than the one listed here, my analysis suggests to take that bet. As for the probabilities, MSU does have the best overall odds, but the field appears to be wide open. Except, that is, for poor Northwestern and Nebraska, who both failed to win the tournament in any of the 5,000 trials.
Finally, I pulled some of the lines for the tournament that I saw and calculated the return on investment for each team. Those results are shown here:
Interestingly, the only teams that appear to be a "good bet" are Wisconsin at +1200 and Minnesota at +4000. All other bets are "expected" to lose money. The line for MSU (+300) is very close to my calculated line. Meanwhile, one of the worst bets looks like one on Michigan at +500. This would imply that their odds to win the tournament are about 16 percent while my calculates suggest that they are 7.7% in reality. Michigan? Over-valued? I know, you're shocked.
That is all for today, until next time, go wash your hands and let's hope that let fans attend in person down in Indy. Go Green.
I could probably go on for paragraphs about the significance of the three-peat this year. Back in the summer, I knew that this campaign was going to be a stressful, because they always are when expectations are so high. It looked for a long time like this season was going to end with extreme disappointment. But, Izzo worked his magic again, and now this team has a shot to be special and to achieve some more of those goals that were set back in April.
I could certainly continue to wax poetic, but I am a numbers guy, so I will just give these numbers:
- In 25 years at MSU, Coach Izzo has won the regular season Big Ten title in ten of those years
- In six of those years (two of them without the regular season title) Coach Izzo has won a Big Ten Tournament Title
- Everyone knows that Coach Izzo also has eight total Final Fours, two of which came in years without a Big Ten regular season or tournament title.
For the full length of the Big Ten season, I has provided numerical updates to handicap the race in real time. Now that the race is over, it is time to take a brief look back at the results and then forward to the next phase of the season: the Big Ten Tournament. Let's attack this by asking three probing questions:
1) In the regular season race, which is more important, luck or schedule?
Throughout the season, I had a running tally of the "luck" accumulated by each team to that point. In this context, I defined "luck" as the difference between the actual wins for a team and the expected value of wins. This essentially measures each team's ability to win toss up games. As an example, if a team played a total of ten games in which the point spread was a pick'em (50 percent chance to win) one would expect that team to win five of those ten games. If that team instead won six games, then they would have +1.0 game(s) of luck.
As I have all season, I now present the final Big Ten standings for 2020, including the calculated luck, with the teams listed in order of their current Kenpom ranking, and numbered with their seed in the upcoming Big Ten Tournament.
There are several interesting takeaways from this table. First, if I go back and check the data from New Year's Day, the expected win total for first place MSU was 14.3 wins and the Big Ten champ was projected to go 15-5. In the final analysis, MSU's expected win total was almost a full game lower than that, which suggests the league was tougher than it was even expected to be in early January.
Second, there is quite a bit a spread in luck. The luckiest team in the conference (Illinois, +3.23) had over a five game advantage in luck over the least lucky team in the conference (Purdue, -2.14) who is actually ranked higher in Kenpom than the very lucky co-champions, the Wisconsin Badgers (+3.21). MSU finished the season almost a half-game on the lucky side, while Michigan was a bit over a game on the not-so-lucky side of the coin. While "luck" is very likely not strictly random (some might refer to it as "grit"), it certainly has a big impact on the final standings.
But, what about the effect of the schedule? Back in December, I posted an analysis comparing the strength of schedule of each team in the conference using a unique method (as far as I know). I made the same expected win calculation as in the standings above for each team, but with a twist. I calculated the expected win total for each team's schedule, assuming that schedule was played by a Big Ten team of average strength, which is the preseason was assumed to be Penn State.
In addition, I also made a similar calculation in which I adjusted the Kenpom efficiency of the average team (Penn State) to be equal to that of the team in question. This was an attempt to correct for the fact that a team like MSU benefits from not have to play MSU, while a team like Nebraska is penalized but not getting to play a team as bad as Nebraska.
Now that the season is over, I made the same calculations again using the current Kenpom efficiencies. In this case, the most average Big Ten team turned out to be Purdue. The charts below show both the preseason and postseason strength of schedule results (both corrected and uncorrected) with the easiest schedules (most expected wins) on the left.
Once again, there is a lot to learn from these graphs. First, they also suggest that the conference as a whole is tougher than expected, as basically every teams expected win total decreased from the preseason to post-season calculations.
Second, it is pretty easy from this graph to identify the teams with the easier or harder schedules. The big winner in this analysis appears to be Iowa, who had a ~0.3 game advantage in schedule over Purdue, Rutgers, and Ohio State. If we look back at the enhanced Big Ten standings shown above, this makes sense, as Iowa and Purdue have a higher expected win total than their Kenpom efficiency would suggest. MSU had a slightly (0.1 game) tougher schedule than the lead pack, while Wisconsin and to a greater extent Penn State, Illinois, and Michigan all had tougher schedules by about half a game.
Third, the entire range of strength of schedule is right at a full game. Iowa did have a better schedule than Michigan by that amount. But, as we already know, the full range of luck is closer to five games. In other words, it is better to be lucky than (to have a) good (schedule).
2) Should MSU fans be upset that we didn't get the No. 1 seed in the Big Ten Tournament?
No.
As the Big Ten race entered the final day, I was able to perform several Monte Carlo style simulations of the full Big Ten Tournament using the most like five scenarios, depending on the winners of the MSU-OSU game, Michigan-Maryland game, and the Illinois-Iowa game. The results are shown here:
I can't tell you the exact error in this simulation, but I think that it is around one percent. So, in other words, the simulations gave MSU essentially the same odds (~23 percent) to win the tournament regardless of whether MSU was the No. 1 seed, No 2. seed or the No. 3 seed.
The reason for this, I believe, is also clear from the enhanced Big Ten standings shown above. Due to the overall parity of the league, there are good teams in all parts of the bracket. Even the 12th best team in the conference (Indiana) is ranked in the Kenpom Top 40. It was going to be tough no matter where MSU landed.
3) Who is going to win the Big Ten Tournament and should I bet on it?
Great question. While I never bet myself, it you are of that persuasion, I perhaps have some data to help. The table below shows the full results of my Monte Carlo simulation of the final Big Ten Tournament bracket, with odds for each team to advance to a certain point.
Based on the odds that I calculate for the championship, I also tabulated the betting break even point for each team. If you see a money line higher than the one listed here, my analysis suggests to take that bet. As for the probabilities, MSU does have the best overall odds, but the field appears to be wide open. Except, that is, for poor Northwestern and Nebraska, who both failed to win the tournament in any of the 5,000 trials.
Finally, I pulled some of the lines for the tournament that I saw and calculated the return on investment for each team. Those results are shown here:
Interestingly, the only teams that appear to be a "good bet" are Wisconsin at +1200 and Minnesota at +4000. All other bets are "expected" to lose money. The line for MSU (+300) is very close to my calculated line. Meanwhile, one of the worst bets looks like one on Michigan at +500. This would imply that their odds to win the tournament are about 16 percent while my calculates suggest that they are 7.7% in reality. Michigan? Over-valued? I know, you're shocked.
That is all for today, until next time, go wash your hands and let's hope that let fans attend in person down in Indy. Go Green.
Last edited:
