Untitled Document
Brown Columbia Cornell Dartmouth Harvard Penn Princeton Yale



 Page 4 of 4 « First<1234
Username Post: Stefanini        (Topic#19971)
Chet Forte 
PhD Student
Posts: 1858

Reg: 03-02-08
04-24-17 11:42 AM - Post#228888    
    In response to Dr. V

Thanks Dr. V. Very interesting. While I admit that I am not objective, Smith is also an exceptional ball handler.


 
mrjames 
Postdoc
Posts: 4859

Loc: Montclair, NJ
Reg: 11-21-04
Stefanini
04-24-17 12:05 PM - Post#228889    
    In response to Chet Forte

Counting stats don't really tell you all that much. What matters more is efficiency and usage.

For the season, Bryce had the sixth-highest efficiency nationally amongst frosh with 28% usage or higher (nationally only about 130 players total used 28% or more of possessions). For the season, his shot rate was up over 29%, which is a similarly high rate to his usage. His eFG% was just over 50% for the year though his TS% was up over 56% because of his strong free-throw shooting.

Mike was a little behind in all of those metrics, but still played quite well for the year, especially for a freshman. His usage and shot rate were 24 and 26%, respectively, which indicates a central role in the offense - not quite carrying it, but certainly a big-time contributor. His eFG% and TS% were 46 and 51, respectively, which you'd like to see rise a bit next year.

What's more is that Mike was only a KP MVP in 2 games last season, whereas Bryce was in 5. That indicates more of an ability to take over a game - essentially a "where's the ceiling" metric, versus more of those season-long averages above.

Finally, looking at Win Shares, Bryce was fifth amongst frosh with 2.5, whereas Mike was eighth with 2.0.

Mike had a freshman season that would have won ROY easily back in the 2000s and even in some of the 2010 years. But he might not have even made an All-Rookie First Team last year (Brodeur, Oni, Aiken would have been locks, from there, it'd be Mike, Towns, Bruner, Lewis, Bassey) - that's how crazy last year's class was. IF this league can just stay healthy and not lose any stars in the annual Sept-Nov bloodbath, this should be a Top 10 league next year.

 
Dr. V 
PhD Student
Posts: 1356

Reg: 11-21-04
04-24-17 02:09 PM - Post#228893    
    In response to mrjames

Maybe I'm too much of a dinosaur to fully understand and appreciate what you're trying to say, but something doesn't add up. Head to head, Smith's and Aiken's stats were roughly even, though Smith's assists and steals were higher (those don't count in the new BB math anymore?).

For the year, Aiken's scoring average was a little higher, 14.5 to 13.6, but in League, Smith's was higher, 14.7 to 13.4. You say those don't count for much? OK. But how do shooting percentages not matter? Smith's for the season were FG % 401, 3-pt % 382 and FT % 821. Aiken's were FG % 392, 3-pt % 345 and FT % 876. How does that trianslate into Aiken having had a better season? (And that's what we're measuring, right? Not future ceiling height or whatever.)

Perhaps most importantly, at least to this dinosaur, Smith's assist to TO ratio was 95/48; Aiken's was 71/63. How in the world does that get washed out in the new stats? And lastly, Smith had 29 steals to Aiken's 21.

 
mrjames 
Postdoc
Posts: 4859

Loc: Montclair, NJ
Reg: 11-21-04
04-24-17 02:26 PM - Post#228894    
    In response to Dr. V

FG% is a misleading stat. Because Aiken took a much higher percentage of his shots as threes than Smith did, you'd expect his FG% to be lower. Once you adjust for the 3-point shot (eFG%), Aiken's eFG is ~4 pp higher.

Assist-to-turnover tells some of a story, but really misses the bigger point - namely that the two aren't related stats. What we care about more are the independent rates: What percent of the team's baskets that you didn't score while you were on the floor did you have an assist on and what percent of your used possessions were used by you turning the ball over.

Since Aiken used a higher percentage of possessions, you'd expect him to have more turnovers. Aiken's TO Rate was 18.8 percent, which is pretty good for a ball-dominant freshman (better than Siyani's was). Smith's was SUPERB at just under 14%. If he keeps that up, that's extremely valuable.

Their assist rates were pretty much equal: 22.1% for Aiken, 22.8% for Smith.

Their steal rates were essentially equivalent, due to the difference in numbers of oppt poss faced based on different playing time.

Rates are extremely important, as is adjusting for the three-point shot. When you do that, Smith's only real advantage over Aiken was his extreme stinginess with turnovers, while Aiken got the better of shooting and usage. That's why Aiken's efficiency for the year was slightly better than Smith's even at a higher usage rate.

 
Stuart Suss 
Masters Student
Posts: 845

Loc: Chester County, Pennsylva...
Reg: 11-21-04
04-24-17 02:48 PM - Post#228895    
    In response to mrjames

This debate is why, for reasons of clarity and transparency, I prefer to use the individual efficiency statistics.

Of the 60 players who were on the court for 140 or more minutes during the 14 regular season league games, Bryce Aiken ranked 52 and Mike Smith ranked 53 in Adj WS. The main difference is shooting efficiency. Mike Smith was a high volume shooter who made only one-third of his two point shots, 47 for 141. Mike's good assist and turnover numbers could only partially make up for that negative.

If you sort the spreadsheet to rank players by WS/40, to see who performed best on a "per 40 minute" basis, the underrated player on the Columbia roster was Lukas Meisner who ranked 4th in the entire league.

(Please note that since the column headings are on Row 1, the number 1 ranked player appears on Row 2, the number 2 ranked player on Row 3, etc . . .)


 
mrjames 
Postdoc
Posts: 4859

Loc: Montclair, NJ
Reg: 11-21-04
04-24-17 06:01 PM - Post#228898    
    In response to Stuart Suss

Any metric that has Bryce and Mike ranked 52nd and 53rd out of 60 qualified Ivy players is a pretty terrible metric. I have Bryce at 15 in win shares and Mike at 25. Bryce was 10 in hands on buckets and Mike was 3rd. Even in PER, which penalizes non-rebounds big time, they're 22nd and 26th.

On no planet were they 52nd and 53rd out of 60.

 
Tiger69 
Postdoc
Posts: 2078

Reg: 11-23-04
04-24-17 07:04 PM - Post#228900    
    In response to mrjames

Couldn't help but notice how many Tigers were in top 19 (5). Well, we were 16-0.😄 That might have had something to do with it. Also, in those 3-4 games where we were nearly upset because one or two players were being effectively controlled by good defense, some other team player always came through. Perhaps Harvard had to lean too heavily on Aiken. What little I saw of him, he seem to suffer from "freshmanitis" which, I'm sure that he'll get over. Lucky for Harvard that Chambers was around to calm him down this season. Next year he'll be on his own.

 
Stuart Suss 
Masters Student
Posts: 845

Loc: Chester County, Pennsylva...
Reg: 11-21-04
04-24-17 08:05 PM - Post#228902    
    In response to Tiger69

If you do not like the WS (Win Score) rankings, please direct your complaints to Dave Berri. He devised the weightings for the individual metrics, not me. In league games, on a per 40 minute basis, the top ranked player in conference games was Zena Edosomwan. This is consistent with Mike James’ frequent complaint that Zena was underutilized. So, the WS must not be a totally flawed evaluative method.

The WS places considerable weight on rebounding. Forwards and centers tend to score higher than guards. It also punishes inefficient shooting. Bryce ranked 48th and Mike ranked 58th in effective FG%. Neither player put himself on the foul line that much. Looking at True FG%, Bryce ranked 46th and Mike ranked 57th. So, if you are going to overcome those rankings, and you do not play at a position that produces a lot of rebounds, you better make up the shooting numbers with low turnovers. Mike Smith succeeded. He committed 25 turnovers in 477 minutes, ranking 16th. Bryce committed 36 turnovers in 404 minutes, ranking 54th.

John Hollinger and his PER do not weigh rebounding as heavily as does Dave Berri. Hollinger does not punish inefficient shooting; arguably, he rewards high volume shooting without regard to efficiency. In my chart the GS (Game Score) is Hollinger’s PER (without the norming to a league average). In the Adjusted GS, Mike Smith ranks 16th and Bryce Aiken ranks 22nd, although both rank lower on a per 40 minute basis (Bryce 28th, Mike 35th).

This whole debate turns on how much you want to reward (or punish) a high volume, low efficiency shooter. My friend, Mike, you should analyze the numbers, not engage in what you used to criticize as malleable heuristics.


 
mrjames 
Postdoc
Posts: 4859

Loc: Montclair, NJ
Reg: 11-21-04
04-24-17 08:51 PM - Post#228903    
    In response to Stuart Suss

Not a huge fan of limiting sample for linear weights models - and I see that you're using Ivy-only stats. You get a very different story for Bryce and Mike if you use full-year numbers. Also those don't take into account the usage rates, which are extremely important context for efficiency.

Linear weights models can turn the box score stats into something a little more useful as a first-blush approximation, but stopping there misses some very important context that can be picked up via other metrics both derived off the box score and those derived off the play-by-play feeds. It's rather humorous that the specific Ivy-only slice via the linear weights models is so penal specifically to Mike and Bryce, but it's certainly an inaccurate view of their value from the 2016-17 season.

 
Chet Forte 
PhD Student
Posts: 1858

Reg: 03-02-08
04-24-17 09:14 PM - Post#228904    
    In response to mrjames

Mike's two point % was adversely affected by his inability to shield the ball as he went in for layups. His shot was blocked with regularity. When he learns how to protect the ball in close and shoot at a higher angle his two point percentage will improve.


 
hoopsfan 
Masters Student
Posts: 516

Reg: 12-26-04
04-25-17 09:51 AM - Post#228914    
    In response to mrjames

Smith and Aiken are both very talented, and they clearly have different styles of play.

In comparing them however unless the analytics factor in the talents of their team mates I am not sure how accurate a comparison can be. For example, Aiken had: Corey Johnson to pass to, who shoots .40+ from 3; Chambers, a 4 time all-Ivy player to set an example; and Seth Towns, who will make a boat-load of money some day.

While Harvard wasn't a very good offensive team, it had more weapons than Columbia which impacted how often Mike Smith shot and his assist rate, at least. I'm not defending him per se because he needs no defense. He has has proven himself as has Aiken. I just think we should acknowledge who they played with in making any comparison.

 
Stuart Suss 
Masters Student
Posts: 845

Loc: Chester County, Pennsylva...
Reg: 11-21-04
04-25-17 10:18 AM - Post#228915    
    In response to hoopsfan

Does box score analytics diminish the value of Mike Smith because his teammates missed shots and deprived him of assists? Or does box score analytics diminish Bryce Aiken because his teammates scored points and Siyani Chambers had assists each of which would otherwise have gone to Bryce?

 
hoopsfan 
Masters Student
Posts: 516

Reg: 12-26-04
04-25-17 05:39 PM - Post#228919    
    In response to Stuart Suss

Well good point. I'm not sure and I am not sure how one would know, although I would not be shocked if there is a way to look at this analytically.

My impression is that playing on worse teams hurts most of your numbers and playing on better teams helps them. Again, I could be wrong and I am not a student of metrics much beyond the box score. But take Maodo Lo for example: his numbers in the year Mullins and Rosenberg were out are very close to his numbers when they returned - except his assists were up from 65 to 98 as he had better players to pass to. The point is that having better players around him did not hurt his numbers and even helped his assists.



 
SomeGuy 
Postdoc
Posts: 4224

Reg: 11-22-04
04-25-17 07:04 PM - Post#228921    
    In response to hoopsfan

Lo was a weird case -- his ORAT and PER dropped senior year despite better players around him. Intuitively, those types of stats should improve with stronger options (better shots for Lo, doesn't have to create everything himself, easier matchups), and counting stats should drop (other guys taking away possessions).

 
Chet Forte 
PhD Student
Posts: 1858

Reg: 03-02-08
04-25-17 07:15 PM - Post#228922    
    In response to SomeGuy

Lo had several games as a junior which were off the charts, Princeton where he went for almost 40, and the second Harvard game, when he embarrassed Wes Saunders, the best defender in the IL that year, for 30 plus. I think his less dominant senior year was the result of having better talent around him. I think it was in his personality to defer to his teammates when they were as good as Mullins, Rosie, etc. And in fact, I remember being frustrated by his play during his senior year, when he should have taken games over in crunch time and didn't seem to want to do so. Nonetheless he was still one of only two unanimous first team All Ivy selections.


 
 Page 4 of 4 « First<1234
Icon Legend Permissions Topic Options
Report Post

Quote Post

Quick Reply

Print Topic

Email Topic

5519 Views




Copyright © 2004-2012 Basketball U. Terms of Use for our Site and Privacy Policy are applicable to you. All rights reserved.
Basketball U. and its subsidiaries are not affiliated in any way with any NCAA athletic conference or member institution.
FusionBB™ Version 2.1 | ©2003-2007 InteractivePHP, Inc.
Execution time: 0.142 seconds.   Total Queries: 16   Zlib Compression is on.
All times are (GMT -0500) Eastern. Current time is 01:21 PM
Top