For MSU basketball, the odds of a Big Ten regular season title are now vanishingly small (about 7%), but that doesn't mean the season is over. While the last few weeks have certainly been disappointing, MSU still has the potential to make this a season to remember. What happens in February stays in February. But, college basketball legends are made in March, and March is just around the corner.
So, the question now is what can we realistically expect out of the Spartans in March? While no one has a crystal ball to predict the future, it is possible to look back at historical data to see what attributes are necessary to build a championship team. While the number of variables that we could analyze are virtually endless, there are certain key stats that seem to be more important than other.
I am personally a huge proponent of efficiency values. For my money, Kenpom adjusted efficiency values (offensive, defensive, and margins) are the golden standard when it comes to predicting basketball outcomes. Kenpom efficiency margins correlate very well to Vegas point spreads, and Vegas point spreads correlate very robustly to victory probabilities. This is always a great place to start.
Beyond that, many experts also point to the so-called "Four Factors" in basketball to best correlate to winning. Those are (for both offense and defense): effective field goal percentage, rebounding percentage, turn-over rate, and free throw rate. For good measure, I think that it is also interesting to look at both 2-point and 3-point shooting percentages individually.
In order to understand the big picture, and the specific situation for MSU and other teams, I decided to plot the various pre-tournament statistics mentioned above for each team that has won the National Title back to 2002 (where Kenpom data is readily available). On top of that, I also plotted the same statistic for each team in the current Kenpom Top 25. Why the Top 25? No team since 2002 has won the National Title without finishing the regular season at least in the Kenpom Top 25. Furthermore, I also generated plots that compare MSU's current stats to the pre-tournament stats of every other Tom Izzo team back to 1997.
For each plot, I compared a defensive stat (such as defensive efficiency) to an offensive stat (such as offensive efficiency). I also adjusted the axes of all plots such that the upper right hand corner is always positive. So, the right side of the plots represent good offensive stats, the top of the graph represents good defensive stats, and the upper right hand corner is both. As mentioned above, the data from past years is all pre-tournament data, I feel that this is the best snap shot to focus on if the goal is to predict tournament perform.
With that background, let's dig into the data:
Efficiency
Kenpom efficiency margins are the most important set of data in predicting tournament success. Once again, no National Champion in the past almost 20 years has started the tournament outside of Top 25 in adjusted efficiency margin (defined as the difference between offensive and defensive adjusted efficiency.) But, it is actually a bit more severe than that. Actually, 15 of the past 18 Champions started the Tournament in the Top 6. That right there is perhaps the most important piece of data that I can share.
In addition to just the efficiency margin, every National Championship has posted both minimum adjusted offensive efficiency (111.4 points per 100 possessions from UCONN in 2014) and a maximum adjusted defensive efficiency (96.0 points per 100 possession from Villanova in 2018). These data points help to define a range of stats where potential National Champions can be found. I highlighted this region of the plot.
So, without further ado, here is the current efficiency plot showing the current Kenpom Top 25 as well as every champion back to 2002. The diagonal lines represent a constant efficiency margin.
The first thing to note from this plot is that the field in 2020 is surprisingly weak overall. Currently, only one team (Kansas) has an efficiency margin over 30 and only four other teams (Baylor, Duke, San Diego State, and Gonzaga) have a current margin over 25. The vast majority of past champions (14 of 18) are located above the 25 line.
As for other teams in championship region, there are currently only 8 other teams (Ohio State, MSU, Arizona, Michigan Louisville, Houston, Villanova, and Dayton) and each of those are located precariously close to the edge. Also notable are teams like West Virginia and Maryland with decent overall efficiencies margins, but who are not well balanced (in this case, both teams have weak offenses) and are therefore not inside the potential championship envelope.
But, where does MSU stack up in 2020 compared to previous Izzo teams? That data set is shown here, where each data point is also labeled with the team's performance in the NCAA tournament:
As I squint at the data, Izzo teams seem to fall into four distinct categories. First, there are the "balanced" teams located in the upper right hand corner of the plot. Based on efficiency margin, these are Izzo's four best teams, including the three Final Four teams from 1999 to 2001, the National Title team, and last year's team. All four of these teams made the Final Four. Also of note is that all four teams tended to be better on defensive than on offensive.
The next category are the good defensive teams. This cluster of 8 teams (including this year's squad) includes two additional Final Four teams, an Elite 8 team, and 3 Sweet 16 teams. The worst performer in this cluster was the 2007 team that lost to a very good UNC team in the second round. It should be noted so far that every Izzo team with a defensive efficiency better than ~91 has at least made the Sweet 16, and 6 of the 9 have at least made a Regional Final. So far, the 2020 team is also in this category.
Next is the good offensive teams. This cluster includes 6 teams, including 2 Final Four teams, 1 Elite Eight team, 1 Sweet 16 team, 1 team that lost in the 2nd round, and the ill-fated 2016 team which we shall not speak of again. While this cluster of teams had a history of success, as a general rule Izzo teams that are good defensive teams tend to do better in March than his offensive teams.
Mathematically, I tried a simple correlation of the number of tournament wins to all of the statistics mentioned so far in this post (as well as a few which are not.) The only two stats that correlated to March wins with an R-squared over 0.2 were efficiency margin (0.432) and defensive efficiency (0.430). The R-squared for offensive efficiency is only 0.064.
The final cluster of Izzo teams include teams that struggled to achieve an efficiency margin much above 15. Of those 6 teams, combined they only won a single tournament game in 2017. Four of Izzo's five teams that lost in Round 1 are located in this cluster. It seems that we perhaps should have seen those losses coming after all.
So, what does this mean for the prospects of the 2020 MSU team? Based on the data shown above, I see more reasons to be optimistic than pessimistic. That said, all of the usual Kenpom caveats still exist. This type of efficiency data is based on season averages. MSU has clearly been on a slide recently. So, the real question is: who is the "real" MSU team? Is it the team that we saw in early January or it is team that just lost four of the last six games?
At this point, I don't think that this question has a right answer. We just don't know. I think that it is certainly possible that this team turns it around and starts to play a little better or a lot better very, very soon. However, it is also possible that they are who they are now. In either case, the safest bet is to just use the data as is. Maybe they simply are who their averaged data says that they are. If nothing else, I am sure the each and every Izzo team (and non-Izzo team for that matter) has ebbs and flows throughout the season. But, when I plot up the data, I just use the average. I think that this is the only fair comparison, but it certainly is not infallible.
So, back to the 2020 Spartans. Based on the plots above, the 2020 version of MSU is a pretty good defensive team. As such, history actually suggests that this team can still make the Sweet 16 or beyond. But, they would also be highly dependent on MSU's seed and draw (which will be the subject of a future post). If MSU were to draw an 8- or 9-seed lined up against a strong 1-seed like Kansas... well, that could be a problem. But, the historical data seems promising.
The analysis may also give us a hint as to how to adjust our expectations over the next few weeks. If MSU continues to struggle, the defensive numbers will likely start to slide as well. If the defensive efficiency starts to drop, MSU will start to approach the "bad Izzo team" region of the graph. In that case, an early, perhaps 1st round exit is a very realistic expectation. However, if MSU can hold their defensive position, yet improve on offense, the team would start to resemble the team that we all thought (hoped?) that we had back in October: a Final Four and National Title contender. While time is growing short, I think both directions are still very possible.
While my comments above provide the bulk of the analysis, I actually have A LOT more data to share about some of the individual stats. However, I will try to keep my comments to a minimum. On some level, the data speaks for itself. Next, let's cover each of the four factors, once again plotted as the defensive number as a function of the offensive number, starting with:
Effective Field Goal Percentage (eFG%)
For those that are not familiar with the details of the four factors, eFG% is just a weighted field goal percentage that takes into account the fact that 3s are worth 50% more points than 2s. Similar to the graphs above, let start with the current national snapshot compared to past champions
As for teams closest to the upper right-hand corner, Kansas and San Diego State have the best overall combination of good shooting and good shooting defense. MSU is a pretty average shooting team, but is one of the best at FG% defense. In general, however, most teams in the current Top 25 reside within a region of the plot where past champions reside. That said, there are a couple of notable outliers. For example, West Virginia can't shoot, and teams like BYU, Villanova, and Iowa struggle to defend. Purdue can't do either (unless they are playing MSU at Mackey).
As for MSU's current team compared to historical teams, that comparison is shown below:
Similar to the national plot, it is clear that from a field goal percentage point of view, this is an average shooting MSU team, but a very good team defensively. Interestingly, most of Izzo's Final Four teams are right in the middle of this plot, so there is no clear trend relating shooting to March success for Izzo teams.
Turnover Rate
Below is the chart of turnover rate (turnovers per possession) for the Kenpom Top 25 and the previous National Champions.
Once again, most of the current Top 25 is clustered with past Champions, so they is no clear trend to note. That said, teams like Baylor, San Diego State, Texas Tech, and Florida State stand out a bit on the upper right portion of the graph. West Virginia not only can't shoot, but they struggle turning the ball over on offense. Interestingly, both MSU and Michigan our outliers on this graph, not because they turn the ball over on offense, but because neither team is generating TOs on the defensive end. Odd.
As for the comparison to historical MSU teams, that plot is shown here
While the 2020 MSU team has struggled with turnovers in some games this year (such as the previous game in Lincoln) believe it or not, this is one of Izzo's least turn-over prone team. As for their inability to create turnovers, a handful of Izzo's previous teams were just as bad or worse (including last year's team) and at least two of them wound up in the Final Four anyway.
Rebounding Rate
The chart below compares the offensive and defensive rebounding percentages for the current Kenpom Top 25 and the previous National Champions
In this case, the three teams that stand out are Gonzaga, Houston, and West Virginia (who at least can do something right). MSU actually looks relatively strong in this graph as well, especially on the defensive glass. As for outliers, both BYU and Creighton struggle on the offensive glass, while Villanova, Michigan and Dayton are all also outside of the Championship region due to offensive rebounding
The chart below shows the rebounding comparison of the 2020 MSU team to previous Izzo squads.
While MSU's current offensive rebounding numbers look decent compared to the rest of the Top 25, this is one of the three worst offensive rebounding teams of the Izzo era. For the pessimists out there, I will mention this: offensive rebounding is the variable that correlates 3rd highest to March success for MSU historically. The R-squared is only 0.19, but there is some evidence that the offensive rebounding is a problem.
That said, the defensive numbers look quite a bit better. While MSU currently ranks only in the 60s national in defensive rebounding, the raw number of 25.4% is actually 5th best in the Izzo era, and slightly better than the 2000 National Title team.
Free Throw Rate
The final of the four factors is free throw rate, which measures the ratio of free throw attempts to field goal attempts. It more or less measures how much a team fouls and how many fouls a team draw. Following the same format, I present below the national plot, followed by the historical MSU plot.
As for the national plot, only about half of the Top 25 seem to fall within the region of past champions. MSU is barely in that region, along with Gonzaga, Maryland, Kansas, and Colorado. As for the notable outliers, West Virginia and Houston seem to foul way too much, while Villanova, Michigan, Purdue, and BYU don't get to the line enough on offense.
As for the 2020 MSU team, it does seem to be a bit of an outlier. The 2020 team is very good at not putting opposing teams on the line. In fact they are second best in the Izzo era in this regard, with only last year's team being better. However, it is also the 2nd worst team in the Izzo era as far as drawing fouls. Only the 2016 team was worse. Odd
2-pt FG%
Just to give a little bit more data, I also have the national and MSU chart for both 2-pt and 3-pt field goal percentage. Here are the charts for 2-pointers:
The message regarding MSU is essentially the same for both plots: MSU is an average 2-pt shooting team with an above average 2-pt defense. Also notable is the 2-pt shooting of Dayton, which is off-the-charts good.
3-pt FG%
Finally, here are the last two plots comparing national and MSU 3-pt% shooting.
The 3-point line did get pushed back a bit this year, so the comparison to previous years' is not quite fair. However, I think the message is still the same. MSU has a very good 3-point defense (the best in the Izzo era, by a long shot) but MSU's 3-point shooting is also well below average. As a final data point for the pessimists, 3-pt shooting is the 4th highest correlated variable to MSU's historical success in March, with an R-squared value of 0.183. No other variable has an R-squared value over 0.1. So, those that are concerned about 3-pt shooting might have a point.
Thus ends the lesson for today. Go Green.
So, the question now is what can we realistically expect out of the Spartans in March? While no one has a crystal ball to predict the future, it is possible to look back at historical data to see what attributes are necessary to build a championship team. While the number of variables that we could analyze are virtually endless, there are certain key stats that seem to be more important than other.
I am personally a huge proponent of efficiency values. For my money, Kenpom adjusted efficiency values (offensive, defensive, and margins) are the golden standard when it comes to predicting basketball outcomes. Kenpom efficiency margins correlate very well to Vegas point spreads, and Vegas point spreads correlate very robustly to victory probabilities. This is always a great place to start.
Beyond that, many experts also point to the so-called "Four Factors" in basketball to best correlate to winning. Those are (for both offense and defense): effective field goal percentage, rebounding percentage, turn-over rate, and free throw rate. For good measure, I think that it is also interesting to look at both 2-point and 3-point shooting percentages individually.
In order to understand the big picture, and the specific situation for MSU and other teams, I decided to plot the various pre-tournament statistics mentioned above for each team that has won the National Title back to 2002 (where Kenpom data is readily available). On top of that, I also plotted the same statistic for each team in the current Kenpom Top 25. Why the Top 25? No team since 2002 has won the National Title without finishing the regular season at least in the Kenpom Top 25. Furthermore, I also generated plots that compare MSU's current stats to the pre-tournament stats of every other Tom Izzo team back to 1997.
For each plot, I compared a defensive stat (such as defensive efficiency) to an offensive stat (such as offensive efficiency). I also adjusted the axes of all plots such that the upper right hand corner is always positive. So, the right side of the plots represent good offensive stats, the top of the graph represents good defensive stats, and the upper right hand corner is both. As mentioned above, the data from past years is all pre-tournament data, I feel that this is the best snap shot to focus on if the goal is to predict tournament perform.
With that background, let's dig into the data:
Efficiency
Kenpom efficiency margins are the most important set of data in predicting tournament success. Once again, no National Champion in the past almost 20 years has started the tournament outside of Top 25 in adjusted efficiency margin (defined as the difference between offensive and defensive adjusted efficiency.) But, it is actually a bit more severe than that. Actually, 15 of the past 18 Champions started the Tournament in the Top 6. That right there is perhaps the most important piece of data that I can share.
In addition to just the efficiency margin, every National Championship has posted both minimum adjusted offensive efficiency (111.4 points per 100 possessions from UCONN in 2014) and a maximum adjusted defensive efficiency (96.0 points per 100 possession from Villanova in 2018). These data points help to define a range of stats where potential National Champions can be found. I highlighted this region of the plot.
So, without further ado, here is the current efficiency plot showing the current Kenpom Top 25 as well as every champion back to 2002. The diagonal lines represent a constant efficiency margin.
The first thing to note from this plot is that the field in 2020 is surprisingly weak overall. Currently, only one team (Kansas) has an efficiency margin over 30 and only four other teams (Baylor, Duke, San Diego State, and Gonzaga) have a current margin over 25. The vast majority of past champions (14 of 18) are located above the 25 line.
As for other teams in championship region, there are currently only 8 other teams (Ohio State, MSU, Arizona, Michigan Louisville, Houston, Villanova, and Dayton) and each of those are located precariously close to the edge. Also notable are teams like West Virginia and Maryland with decent overall efficiencies margins, but who are not well balanced (in this case, both teams have weak offenses) and are therefore not inside the potential championship envelope.
But, where does MSU stack up in 2020 compared to previous Izzo teams? That data set is shown here, where each data point is also labeled with the team's performance in the NCAA tournament:
As I squint at the data, Izzo teams seem to fall into four distinct categories. First, there are the "balanced" teams located in the upper right hand corner of the plot. Based on efficiency margin, these are Izzo's four best teams, including the three Final Four teams from 1999 to 2001, the National Title team, and last year's team. All four of these teams made the Final Four. Also of note is that all four teams tended to be better on defensive than on offensive.
The next category are the good defensive teams. This cluster of 8 teams (including this year's squad) includes two additional Final Four teams, an Elite 8 team, and 3 Sweet 16 teams. The worst performer in this cluster was the 2007 team that lost to a very good UNC team in the second round. It should be noted so far that every Izzo team with a defensive efficiency better than ~91 has at least made the Sweet 16, and 6 of the 9 have at least made a Regional Final. So far, the 2020 team is also in this category.
Next is the good offensive teams. This cluster includes 6 teams, including 2 Final Four teams, 1 Elite Eight team, 1 Sweet 16 team, 1 team that lost in the 2nd round, and the ill-fated 2016 team which we shall not speak of again. While this cluster of teams had a history of success, as a general rule Izzo teams that are good defensive teams tend to do better in March than his offensive teams.
Mathematically, I tried a simple correlation of the number of tournament wins to all of the statistics mentioned so far in this post (as well as a few which are not.) The only two stats that correlated to March wins with an R-squared over 0.2 were efficiency margin (0.432) and defensive efficiency (0.430). The R-squared for offensive efficiency is only 0.064.
The final cluster of Izzo teams include teams that struggled to achieve an efficiency margin much above 15. Of those 6 teams, combined they only won a single tournament game in 2017. Four of Izzo's five teams that lost in Round 1 are located in this cluster. It seems that we perhaps should have seen those losses coming after all.
So, what does this mean for the prospects of the 2020 MSU team? Based on the data shown above, I see more reasons to be optimistic than pessimistic. That said, all of the usual Kenpom caveats still exist. This type of efficiency data is based on season averages. MSU has clearly been on a slide recently. So, the real question is: who is the "real" MSU team? Is it the team that we saw in early January or it is team that just lost four of the last six games?
At this point, I don't think that this question has a right answer. We just don't know. I think that it is certainly possible that this team turns it around and starts to play a little better or a lot better very, very soon. However, it is also possible that they are who they are now. In either case, the safest bet is to just use the data as is. Maybe they simply are who their averaged data says that they are. If nothing else, I am sure the each and every Izzo team (and non-Izzo team for that matter) has ebbs and flows throughout the season. But, when I plot up the data, I just use the average. I think that this is the only fair comparison, but it certainly is not infallible.
So, back to the 2020 Spartans. Based on the plots above, the 2020 version of MSU is a pretty good defensive team. As such, history actually suggests that this team can still make the Sweet 16 or beyond. But, they would also be highly dependent on MSU's seed and draw (which will be the subject of a future post). If MSU were to draw an 8- or 9-seed lined up against a strong 1-seed like Kansas... well, that could be a problem. But, the historical data seems promising.
The analysis may also give us a hint as to how to adjust our expectations over the next few weeks. If MSU continues to struggle, the defensive numbers will likely start to slide as well. If the defensive efficiency starts to drop, MSU will start to approach the "bad Izzo team" region of the graph. In that case, an early, perhaps 1st round exit is a very realistic expectation. However, if MSU can hold their defensive position, yet improve on offense, the team would start to resemble the team that we all thought (hoped?) that we had back in October: a Final Four and National Title contender. While time is growing short, I think both directions are still very possible.
While my comments above provide the bulk of the analysis, I actually have A LOT more data to share about some of the individual stats. However, I will try to keep my comments to a minimum. On some level, the data speaks for itself. Next, let's cover each of the four factors, once again plotted as the defensive number as a function of the offensive number, starting with:
Effective Field Goal Percentage (eFG%)
For those that are not familiar with the details of the four factors, eFG% is just a weighted field goal percentage that takes into account the fact that 3s are worth 50% more points than 2s. Similar to the graphs above, let start with the current national snapshot compared to past champions
As for teams closest to the upper right-hand corner, Kansas and San Diego State have the best overall combination of good shooting and good shooting defense. MSU is a pretty average shooting team, but is one of the best at FG% defense. In general, however, most teams in the current Top 25 reside within a region of the plot where past champions reside. That said, there are a couple of notable outliers. For example, West Virginia can't shoot, and teams like BYU, Villanova, and Iowa struggle to defend. Purdue can't do either (unless they are playing MSU at Mackey).
As for MSU's current team compared to historical teams, that comparison is shown below:
Similar to the national plot, it is clear that from a field goal percentage point of view, this is an average shooting MSU team, but a very good team defensively. Interestingly, most of Izzo's Final Four teams are right in the middle of this plot, so there is no clear trend relating shooting to March success for Izzo teams.
Turnover Rate
Below is the chart of turnover rate (turnovers per possession) for the Kenpom Top 25 and the previous National Champions.
Once again, most of the current Top 25 is clustered with past Champions, so they is no clear trend to note. That said, teams like Baylor, San Diego State, Texas Tech, and Florida State stand out a bit on the upper right portion of the graph. West Virginia not only can't shoot, but they struggle turning the ball over on offense. Interestingly, both MSU and Michigan our outliers on this graph, not because they turn the ball over on offense, but because neither team is generating TOs on the defensive end. Odd.
As for the comparison to historical MSU teams, that plot is shown here
While the 2020 MSU team has struggled with turnovers in some games this year (such as the previous game in Lincoln) believe it or not, this is one of Izzo's least turn-over prone team. As for their inability to create turnovers, a handful of Izzo's previous teams were just as bad or worse (including last year's team) and at least two of them wound up in the Final Four anyway.
Rebounding Rate
The chart below compares the offensive and defensive rebounding percentages for the current Kenpom Top 25 and the previous National Champions
In this case, the three teams that stand out are Gonzaga, Houston, and West Virginia (who at least can do something right). MSU actually looks relatively strong in this graph as well, especially on the defensive glass. As for outliers, both BYU and Creighton struggle on the offensive glass, while Villanova, Michigan and Dayton are all also outside of the Championship region due to offensive rebounding
The chart below shows the rebounding comparison of the 2020 MSU team to previous Izzo squads.
While MSU's current offensive rebounding numbers look decent compared to the rest of the Top 25, this is one of the three worst offensive rebounding teams of the Izzo era. For the pessimists out there, I will mention this: offensive rebounding is the variable that correlates 3rd highest to March success for MSU historically. The R-squared is only 0.19, but there is some evidence that the offensive rebounding is a problem.
That said, the defensive numbers look quite a bit better. While MSU currently ranks only in the 60s national in defensive rebounding, the raw number of 25.4% is actually 5th best in the Izzo era, and slightly better than the 2000 National Title team.
Free Throw Rate
The final of the four factors is free throw rate, which measures the ratio of free throw attempts to field goal attempts. It more or less measures how much a team fouls and how many fouls a team draw. Following the same format, I present below the national plot, followed by the historical MSU plot.
As for the national plot, only about half of the Top 25 seem to fall within the region of past champions. MSU is barely in that region, along with Gonzaga, Maryland, Kansas, and Colorado. As for the notable outliers, West Virginia and Houston seem to foul way too much, while Villanova, Michigan, Purdue, and BYU don't get to the line enough on offense.
As for the 2020 MSU team, it does seem to be a bit of an outlier. The 2020 team is very good at not putting opposing teams on the line. In fact they are second best in the Izzo era in this regard, with only last year's team being better. However, it is also the 2nd worst team in the Izzo era as far as drawing fouls. Only the 2016 team was worse. Odd
2-pt FG%
Just to give a little bit more data, I also have the national and MSU chart for both 2-pt and 3-pt field goal percentage. Here are the charts for 2-pointers:
The message regarding MSU is essentially the same for both plots: MSU is an average 2-pt shooting team with an above average 2-pt defense. Also notable is the 2-pt shooting of Dayton, which is off-the-charts good.
3-pt FG%
Finally, here are the last two plots comparing national and MSU 3-pt% shooting.
The 3-point line did get pushed back a bit this year, so the comparison to previous years' is not quite fair. However, I think the message is still the same. MSU has a very good 3-point defense (the best in the Izzo era, by a long shot) but MSU's 3-point shooting is also well below average. As a final data point for the pessimists, 3-pt shooting is the 4th highest correlated variable to MSU's historical success in March, with an R-squared value of 0.183. No other variable has an R-squared value over 0.1. So, those that are concerned about 3-pt shooting might have a point.
Thus ends the lesson for today. Go Green.
Comments
Post a Comment