Recruiting Talent vs. Conference Standings
1-0-9 is an understatement, really. It's eight consecutive years in which we've finished AT LEAST four spots lower than the raw talent level on the roster would indicate.
We've recruited well enough, won battles for sufficiently highly ranked players, to make the tournament in every single one of those years, and be a genuine conference title contender in a bunch of them. That was the objectively reasonable expectation of our rosters over the past decade.
There have been glaring personnel gaps at certain positions in certain years (sadly that will be the case this year as well), there has also been the occasional injury crisis, and there's been, y'know, Jereme Richmond. But of course that's true of every other team in the conference to some extent, no one is immune from that stuff.
But it has been a constant refrain on this board over the years that "we have to recruit better to compete, we won't have the talent to win unless we can break through to a higher recruiting level". What I hope that narrative's advocates will do is sit with this data for a second and wrestle with how uniformly, categorically incorrect that has always been. It would be great to recruit better, but when you can take the single most talented team in the B1G and one of the four most talented in the entire conference over the last 11 years and end up a bubble team who lost to a 12-20 Indiana and a 7-24 UIC, that suggests that you ought to look at other issues.
Here are those four most talented teams btw:
2008-09 Michigan State: Big Ten Champions, NCAA Runner-Up
2009-10 Michigan State: Big Ten Champions, Final Four
2010-11 Illinois: 9-9 in Big Ten, Second Round
2011-12 Ohio State: Big Ten Champions, Final Four
When your car is burning a quart of oil a day, your first thought should not be "I just don't have enough oil, I'd better save up my money so I can buy as much oil as possible."
It would be cool know what the floor and ceiling is. Like what is the maximum amount you could vary both up and down from your recruiting ranking. Like we have finished 2-4 levels below our talent ranking on the regular. Is it just as possible to finish 2-4 levels above?
What is the maximum reasonable swing in each direction?
What is the maximum reasonable swing in each direction?
Your oil/car analogy is excellent! But, I would add, “that newer better synthetic oil will surely make this clunker go better.”
Last edited:
I'm just eyeballing it so apologies if I miss something.
The biggest overperformance in the sample is 2016-17 Wisconsin who finished 2nd with the 12th best talent. Not surprisingly a team built around four seniors who had been playing together in the same system forever.
The biggest underperformance is 2016-17 Ohio State, who finish 10th with the #1 talent. KBD got hurt and it was the lowest overall talent level ever to be #1 in the conference, but they did end up firing their coach over it in the end.
The year everyone got hurt for us we were a -9, 12th with the 3rd best talent.
But I gotta give the "Worst Coaching" award to Tom Crean for 2013-14 Indiana in which he took the #1 talent and finished 9th even with full health among his best players. Yes that team had lost a ton from the previous year and was very young, but taking Yogi Ferrell and Noah Vonleh and not even making the NIT is something even the Illini can't match over this period. Congrats Tom!
TownieMatt
CU Expat
- Chicago
Good stuff. I did something extremely similar to this a season or two ago. Conclusion was the same, we've greatly underperformed for the talent we've had.
Edit: Found my post, it was from when we were hiring a new coach. Shows how Groce underperformed given the talent he had.
Edit: Found my post, it was from when we were hiring a new coach. Shows how Groce underperformed given the talent he had.
Last edited:
There appear to be seven seasons in the sample that are a -7 or worse. We have 3, Indiana has 2, and Ohio State and Nebraska have 1 each.
It's The Year That Got Weber Fired, The Year That Got Groce Fired, The Year That Got Crean Fired, The Year That Got Matta Fired, our injury-ravaged year (where we still should have been better than we were), the year at Nebraska where James Palmer and Isaac Copeland were sitting out as transfers, and the Tom Crean year previously mentioned.
The Groce vs. Miles example seems instructive. Both schools had a season where they just didn't have access to a bunch of their best players, and for that reason a well-liked coach on the hot seat was given a reprieve for the next year. That next year both teams had all of their guys and had good luck with health, they both really got to play with a full deck of experienced players.
Miles' Nebraska bounced back from 6-12 to 13-5. Groce's Illinois bounced back from 5-13 to 8-10.
It's The Year That Got Weber Fired, The Year That Got Groce Fired, The Year That Got Crean Fired, The Year That Got Matta Fired, our injury-ravaged year (where we still should have been better than we were), the year at Nebraska where James Palmer and Isaac Copeland were sitting out as transfers, and the Tom Crean year previously mentioned.
The Groce vs. Miles example seems instructive. Both schools had a season where they just didn't have access to a bunch of their best players, and for that reason a well-liked coach on the hot seat was given a reprieve for the next year. That next year both teams had all of their guys and had good luck with health, they both really got to play with a full deck of experienced players.
Miles' Nebraska bounced back from 6-12 to 13-5. Groce's Illinois bounced back from 5-13 to 8-10.
Last edited:
First of all, great work Soupy, good analysis and appreciate the time spent on this. A few comments and one suggestion for improvement.
1) It is not really "talent" but more accurately "HS ranking" (in this case as measure by 247). While there is indeed correlation between rankings and talent (correlation decreases as one moves towards the lower end, since now 247 even ranks players even in the 250+ range), rankings are really more about "potential" based on HS/AAU. The higher the ranking, the higher the probability of individual success (not team success) in college, and that relationship is stronger at the higher end and gets weaker as one progress lower in the rankings. But talent extends beyond rankings into individual performance in college; it does not stop at HS/AAU. Therefore, there are new "rankings" (indicative of potential) as players move into the next level (i.e., ranking players in the NBA draft, post college, etc.) and so on.
2) A critical element not captured is positional talent. A great part of our problem post-Dee has been our inability to have B1G talent level at PG. Since Dee left (2006), Illinois has had only one (1) true B1G level PG until Frazier arrived on campus (i.e., Demetri McCamey). That is the most critical position. A secondary positional gap has been inside presence. You can't have such positional gaps (no matter how much posters argue about positionless basketball) unless you have exceptional talent at the other positions to compensate for. Depth is a critical element as well. A low ranked player should not hurt a team’s talent ranking (Duke, UK, Kansas and others have even had walkons at the end of the bench) unless the team has no depth at particular positions and are forced to play those players significant minutes.
3) Furthermore, HS rankings alone do not account for whether that HS ranking/potential was used or not (multiple players had injuries, suspensions, off court problems, etc. etc.). I believe you also identified that as a limitation. If one counts Michael Porter Jr. on Mizzou (who never really played), we may come to the conclusion that Mizzou was way more talented than it actually was. Same with Tracy, Thorne, Mike Shaw, and even Richmond (who missed games and playing time due to off court issues/failed tests, etc.). By the same token, the ranking of the 12th, 13th, 14th player on a team (even walkons) is irrelevant as they usually have really no effect and rarely play. The average HS ranking unfairly treats every player the same, whether they play or not, or how much they play. That is a critical limitation as it treats the 1st player on the team exactly the same as the 13th player on a team.
A way of actually addressing some of those gaps is weighting 247 rankings by the minutes each player played in the B1G (to stay consistent) and creating an Effective Ranking for EACH player, then adding them together (not averaging them) to create a composite Effective Ranking score for each team. While not perfect, that would IMO be a much better surrogate measure.
For example, Trent Frazier's Effective Ranking during the 2017-2018 season would be:
0.9392 x 533 = 500.5936
whereas, Mark Smith's Effective Ranking during the 2017-2018 season would be:
0.9629 x 279 = 268.6491
Which would much more accurately capture talent (weighted by utilization) on the team last year as measured by rankings. It also correctly compensates for players not playing for whatever reason (injuries, RS, off and on-court problems, suspensions, skipping practice, not getting along with coaches, etc. etc.), despite their rankings.
If we add (not average) the effective rankings of each player on the team, we would get an Effective Ranking composite for each team in the B1G, which will be a much better surrogate of the actual talent on B1G teams each year. And as with rankings, I believe that measure, and other similar measures, will show that the impact is more pronounced between high and low end, as it is within groups, especially at low end. That means that effective ranking would be more meaningful on its correlation of competing for B1G or making NCAA, e.g., between the 1st, or 2nd projected team and teams at the lower end, rather than projecting whether the 11th ranked team actually finished 9th or 13th which IMO is less important (similarly, rankings on the average show greater differences between the 15th ranked player and the 130th ranked player, rather than differences between players ranked 170th and 250th).
1) It is not really "talent" but more accurately "HS ranking" (in this case as measure by 247). While there is indeed correlation between rankings and talent (correlation decreases as one moves towards the lower end, since now 247 even ranks players even in the 250+ range), rankings are really more about "potential" based on HS/AAU. The higher the ranking, the higher the probability of individual success (not team success) in college, and that relationship is stronger at the higher end and gets weaker as one progress lower in the rankings. But talent extends beyond rankings into individual performance in college; it does not stop at HS/AAU. Therefore, there are new "rankings" (indicative of potential) as players move into the next level (i.e., ranking players in the NBA draft, post college, etc.) and so on.
2) A critical element not captured is positional talent. A great part of our problem post-Dee has been our inability to have B1G talent level at PG. Since Dee left (2006), Illinois has had only one (1) true B1G level PG until Frazier arrived on campus (i.e., Demetri McCamey). That is the most critical position. A secondary positional gap has been inside presence. You can't have such positional gaps (no matter how much posters argue about positionless basketball) unless you have exceptional talent at the other positions to compensate for. Depth is a critical element as well. A low ranked player should not hurt a team’s talent ranking (Duke, UK, Kansas and others have even had walkons at the end of the bench) unless the team has no depth at particular positions and are forced to play those players significant minutes.
3) Furthermore, HS rankings alone do not account for whether that HS ranking/potential was used or not (multiple players had injuries, suspensions, off court problems, etc. etc.). I believe you also identified that as a limitation. If one counts Michael Porter Jr. on Mizzou (who never really played), we may come to the conclusion that Mizzou was way more talented than it actually was. Same with Tracy, Thorne, Mike Shaw, and even Richmond (who missed games and playing time due to off court issues/failed tests, etc.). By the same token, the ranking of the 12th, 13th, 14th player on a team (even walkons) is irrelevant as they usually have really no effect and rarely play. The average HS ranking unfairly treats every player the same, whether they play or not, or how much they play. That is a critical limitation as it treats the 1st player on the team exactly the same as the 13th player on a team.
A way of actually addressing some of those gaps is weighting 247 rankings by the minutes each player played in the B1G (to stay consistent) and creating an Effective Ranking for EACH player, then adding them together (not averaging them) to create a composite Effective Ranking score for each team. While not perfect, that would IMO be a much better surrogate measure.
For example, Trent Frazier's Effective Ranking during the 2017-2018 season would be:
0.9392 x 533 = 500.5936
whereas, Mark Smith's Effective Ranking during the 2017-2018 season would be:
0.9629 x 279 = 268.6491
Which would much more accurately capture talent (weighted by utilization) on the team last year as measured by rankings. It also correctly compensates for players not playing for whatever reason (injuries, RS, off and on-court problems, suspensions, skipping practice, not getting along with coaches, etc. etc.), despite their rankings.
If we add (not average) the effective rankings of each player on the team, we would get an Effective Ranking composite for each team in the B1G, which will be a much better surrogate of the actual talent on B1G teams each year. And as with rankings, I believe that measure, and other similar measures, will show that the impact is more pronounced between high and low end, as it is within groups, especially at low end. That means that effective ranking would be more meaningful on its correlation of competing for B1G or making NCAA, e.g., between the 1st, or 2nd projected team and teams at the lower end, rather than projecting whether the 11th ranked team actually finished 9th or 13th which IMO is less important (similarly, rankings on the average show greater differences between the 15th ranked player and the 130th ranked player, rather than differences between players ranked 170th and 250th).
Last edited:
FeelYourPaign
- So Cal
This is great stuff.
Trying to turn this around and look forward, the final column in the talent column is the talent that each team will have for the coming season. And just to be sure that I am reading these graphs correctly (mostly a color thing), going into the coming season, we have the 7th best talent, right?
Only PSU and Rutgers have clearly worse talent that we do, while Northwestern, Minnie, Nebraska, Iowa and Wiscy are all fairly close.
Our disadvantage is the relative youth of our team. To earn our 7th spot in the conference, we will presumably have to perform at or above our talent level because of our inexperience. Doesn't that sound right?
That makes me think that a conference record between 8-10 and 11-9 is what should be expected.
Anything outside of those bands will say something about coaching or injuries, etc. etc.
Trying to turn this around and look forward, the final column in the talent column is the talent that each team will have for the coming season. And just to be sure that I am reading these graphs correctly (mostly a color thing), going into the coming season, we have the 7th best talent, right?
Only PSU and Rutgers have clearly worse talent that we do, while Northwestern, Minnie, Nebraska, Iowa and Wiscy are all fairly close.
Our disadvantage is the relative youth of our team. To earn our 7th spot in the conference, we will presumably have to perform at or above our talent level because of our inexperience. Doesn't that sound right?
That makes me think that a conference record between 8-10 and 11-9 is what should be expected.
Anything outside of those bands will say something about coaching or injuries, etc. etc.
FeelYourPaign
- So Cal
Obelix,
While I agree that factoring playing time into the analysis of how a team performed is valid, it seems to me that this somewhat ruins the predictive power of the exercise. I say "somewhat" because some of the playing time analysis reflects the decisions that the coach makes about who to play and who to sit, but some of the analysis excuses the coaches' inability to develop that talent and therefore sidestep that deficiency.
I suppose the question is whether you want a better fit of the talent / performance data to reality or whether you want a better predictive vision of what the past suggests about the future. In the first example (looking back with a better perspective), then playing time is appropriate. I would argue that in the other example (using the past to predict the future), playing time adjustments can cloud the analysis.
While I agree that factoring playing time into the analysis of how a team performed is valid, it seems to me that this somewhat ruins the predictive power of the exercise. I say "somewhat" because some of the playing time analysis reflects the decisions that the coach makes about who to play and who to sit, but some of the analysis excuses the coaches' inability to develop that talent and therefore sidestep that deficiency.
I suppose the question is whether you want a better fit of the talent / performance data to reality or whether you want a better predictive vision of what the past suggests about the future. In the first example (looking back with a better perspective), then playing time is appropriate. I would argue that in the other example (using the past to predict the future), playing time adjustments can cloud the analysis.
Really good stuff from Bartovik. He has us at 16-13 and 10-10 with a 11th place B1G finish.
http://barttorvik.com/team.php?team=Illinois&year=2019
http://barttorvik.com/team.php?team=Illinois&year=2019
SampsonRelpenk
- Edwardsville, IL
Sign me up for that.
While the coach always has the final saying, it does not actually excuse a coach's inability to develop that talent as again, measured by rankings. The coach who will play higher ranked talent will show in the Effective Ranking, the coach who will not play them will also show that. A coach who will just play them, but not coach them/develop them well, will also show as the Effective Ranking composite of the team will not correlate well with the final performance on the team (which is I believe what we want to show).
It is a much more preferable to counting players not playing for whatever reason (injuries, RS, off and on-court problems, suspensions, skipping practice, not getting along with coaches, etc. etc.), despite their rankings. And it also avoids the problem of unfairly treating players the same, whether the top player on the team and the last person on the bench. In real situation, when we say that Kentucky is really talented (based on some measure of rankings), whether the 13th player on the team is a walkon is irrelevant, unless again they have no positional depth (rarely happens). And in the cases where they had a lot of depth 2 units (the famous in-out substitutions of all 5) that will also show, as the rankings of players who play (second unit) also corresponds to their high rankings (because they play).
It is not perfect, for sure, but less problems IMO with limitations of just averaging all rankings and counting all players the same irrespective of actually those players played or not. It is also different than saying this recruiting class is good, which is a-priori statement (i.e., denoting "expectations") than saying that were "talented" a particular year (based on rankings again, which as I said, IMO rankings=talent, just denotes correlation).
Last edited:
THAT would be a crazy scenario.
1. 15-5
2. 14-6
3. 13-7
4. 12-8
5. 11-9
6. 11-9
7. 11-9
8. 11-9
9. 11-9
10. 11-9
11. 10-10
12. 5-15
13. 3-17
14. 2-18
THAT is the type of distribution it would take for us to be 10-10 in the B1G and finish 11th. I guess three basement bottom dwellers could occur.
This method would help with the youth issues as they would normally play less minutes. It wouldn't help with the positional deficits, however, as the minutes would be taken by players playing out of position. Need to deferentiate by position at least pg, ctr and others. Difficult to do as you may be playing 2 PG's and maybe 2 PFs. Still believe the lack of a pg and true center has been a large part of our problem. Kipper was playing a four and should have been a three. Finke was playing a five and should have been a four. You can play positionless offense but you still need particular physical assets to play defense. Does our coaching staff have metrics to determine their own performance?
I agree, that is why I said it addressed some of those critical issues (I believe most), even in #2 above it addresses the depth (often as it relates to position) but not every possible case as, for example, constantly playing talented players out-of position. But again, if you have a lot of superior talent at other positions that may somehow compensate somewhat for lack of a particular position (IMO that was the case in 1989 Flyin' Illini).
Yet in those cases where limitations exist, it will also show anomalies/outliers in the correlation with performance (which is the main point of interest). But it will avoid crude generalizations and contamination in the main correlation.
Last edited:
I'm not familiar with how this site does it's predictions, but it looks like he has ~30 minutes per game from our centers, which will leave 10 minutes/game of Kipper at center. If that's the case I have to guess his Ortg and PPG estimates are probably skewed. All in all, I think he's on the high side, but not out of this world.
Interesting that the only home game loss is to MSU. Gets us to mediocre.
It all kinda depends on precisely the question you're trying to ask, right? You weight it in the way you've suggested and the question becomes "to what extent did highly ranked players contribute to the various teams?". But then as others said, if a lesser ranked player is keeping a higher ranked one on the bench (see: Trent Meacham and Alex Legion) that might become a bit deceptive depending on what you're trying to study. If you're really trying to isolate the effect of "roster quality" as separate from "coaching effects" you would surely want to incorporate what class the various players are, with seniors being more valuable than freshmen.
If you really wanted to drill down on the idea of "talent", it would probably be wise to mix in post-college results, draft picks, NBA careers, something.
All of these would be very interesting and instructive, and present an answer to a slightly different question.
What Soupy has done answers the question "How strong was the recruiting reflected in these rosters?". It's just the raw recruiting data, unconcerned with how *good* the many other factors made those players and those teams. It would be a little bit more robust if we could eliminate guys who were in redshirt sit-out years or who missed entire seasons due to injury, but that stuff happens everywhere, it probably washes out of the data relatively well. (It would also be a ton of work, and Soupy has already put a heroic amount of effort to this)
What this data just thoroughly and comprehensively demolishes is the narrative of the last decade here that "we need to recruit better than we have to succeed, we have to break into a higher level on the recruiting trail BEFORE we can reasonably expect improvement". To be totally honest, when I started reading this I was worried the data was going to be much worse for my argument. Even I have started to just accept things like "Of course 2007 RSCI #35 Kalin Lucas was an era-defining Big Ten superstar while 2009 RSCI #35 DJ Richardson was a limited, marginal player, MSU just recruits better than us". Even I need to be reminded what a total lie that is from time to time.
Obviously we're fighting against youth, inexperience, and extreme positional imbalance in turning our raw recruiting level into wins this year. But looking at these numbers, you've gotta feel pretty optimistic about our future if, IF, we can A. keep all our talented players for the future and B. add some nice pieces in the frontcourt over the next year. Especially if we can improve the staff, I just refuse to believe Underwood is going to keep up this streak of -4's. There's no way. The roster might fall apart, but not crazy underperformance of our talent levels. I realize last year refutes me, but whatever, that was just a vortex of terrible from which nothing could escape.
Last edited:
soupy17
- MI
I'm curious of how this would turn out. I do think it's a slightly different question, but it's one that kind of is a natural next step from here. We now have an idea of what talent was at each program each year, so now we want to know how that talent was used.
I'll get to work on this. Will take some time obviously but I think I'm most curious in how it might change things.
Do you (or anybody else) know of a reliable source for the conference data? sports-reference.com only has the conference stats broken out from the 2010-11 season forward. I could of course just use that season as a starting point, but then we would lose S&C's favorite season in 08-09
The biggest problems (as many of you have touched on) is
1. Even though we have had some descent ranked recruits in recent years during this comparison, our performance projection numbers are being artificially rated as higher expectations during these years because when we do get a decent ranked recruit more often than not he is falling like rock and/or WAY overrated. In many cases said recruit should have been 100 points (or more) lower in their rankings. WAY too many players like Mike Shaw, DJ Williams, Brian Carlwell, etc. along with guys that never really played for us (hurt or left program immediately before we go anything out of them) like Richmond, Leonard, C-Head, etc. I think we have had an inordinate amount of these type of kids compared to other schools and it inflates these statistical expectations compared to other schools.
2. Position imbalance. We have had soooo many years (most years?) without a good PG or a big or both it is ridiculous. Especially compared to other schools. Therefore, we tend to perform lower than recruiting projections would predict most years.
Combine 1 & 2 above and you have Illinois. Bad luck? Bad evaluations? I am guessing a lot of both. But it still comes down to recruiting. I am guessing there is a reason we got some of these higher rated "gems" that were falling like a rock in the rankings.
1. Even though we have had some descent ranked recruits in recent years during this comparison, our performance projection numbers are being artificially rated as higher expectations during these years because when we do get a decent ranked recruit more often than not he is falling like rock and/or WAY overrated. In many cases said recruit should have been 100 points (or more) lower in their rankings. WAY too many players like Mike Shaw, DJ Williams, Brian Carlwell, etc. along with guys that never really played for us (hurt or left program immediately before we go anything out of them) like Richmond, Leonard, C-Head, etc. I think we have had an inordinate amount of these type of kids compared to other schools and it inflates these statistical expectations compared to other schools.
2. Position imbalance. We have had soooo many years (most years?) without a good PG or a big or both it is ridiculous. Especially compared to other schools. Therefore, we tend to perform lower than recruiting projections would predict most years.
Combine 1 & 2 above and you have Illinois. Bad luck? Bad evaluations? I am guessing a lot of both. But it still comes down to recruiting. I am guessing there is a reason we got some of these higher rated "gems" that were falling like a rock in the rankings.
soupy17
- MI
You want a new graph? It sounds like you want a new graph.
The bottom half of that graph has a lot of orange and a lot of crimson.
soupy17
- MI
Next table is the average difference between conference rank and "talent" rank over the ten seasons.
This is awesome.
This table really makes it clear just how badly we have been underachieving over the past decade. Many of us had a gut feeling but this confirms it.