Reply to thread

Message: [QUOTE="mikekan14, post: 233021, member: 1619"] Reminder, I didn't write the ESPN article but I searched for PASE (Performance Against Seed Expectation) and found this website...[url]http://www.bracketscience.com/articles/_navTop10.asp?i=1[/url] Top Active Coaches (name, PASE) 1. Brad Stevens (only 4 appearances), 1.931 2. Tom Izzo, 0.909 3. Sean Miller, 0.748 4. John Beilein, 0.705 5. Steve Fisher, 0.610 The list stops there for free content but the site uses Bill Self as an example and he's at #33 with a PASE of 0.011 for active coaches with at least 5 Tournament appearances. That apparently is a different list as the one above since Brad Stevens hasn't been removed from the sample data. Reminder, Jim Boeheim's PASE is 0.132. Then using [url]http://statsheet.com/mcb/coaches[/url] for the coaching statistics I compared a few measures. Disclaimer: I don't know if this site has updated with this year's Tournament results nor do I know how accurate it is. However, the same source of error (if it's systematic) should be distributed so I don't foresee the chance that something is that far off that it should be considered a statistical outlier. I selected 9 other coaches to compare based on longevity and name/team recognition. Many other coaches (Howland, Dixon, Gary Williams, Mark Few, etc.) could have been included in this survey but I cut the list off at 10 total. At the crux of the exercise is the acceptance that a coach's win total is directly correlated to how good a coach he is, otherwise the normalization to wins isn't going to support my point. Also, I used the coach's career win total so Calhoun's will include his tenure at Northeastern, Pitino at BU, K at Army, Self at Oral Roberts, etc. I think it should be noted that comparing expectations and resources at their previous mid major schools to their current teams is akin to apples and oranges. In any case, those wins were not excluded. Also, no weighting was considered so I can't over weight Boeheim's wins because that came from the tougher Big East nor can I underweight Calipari's wins in the SEC or CUSA. Perhaps a more statistical analysis would normalize win records to an opponent's cumulative winning percentage. I didn't do that. I also understand there are other measures that exist that others will feel are more appropriate. I looked at 3 measures: 1. Tournament Wins/Overall # Wins (TW/OW): overall coaching wins include tournament record so it is a measure of how many of his total wins are from the tournament. 2. Tournament Wins/Tournament Appearance (TW/TA): measures how "deep" a team goes, on average 3. Final Four Appearances/Tournament Appearance (FF/TA): measures how often a team makes a run to the Final Four. This is probably the least significant measure because of the small sample size. Coach: TW/OW, TW/TA, FF/TA Boeheim: 0.0473, 1.75, 0.125 Calhoun: 0.0583, 2.32, 0.182 Calipari: 0.0595, 2.46, 0.231 Huggins: 0.038, 1.42, 0.105 Izzo: 0.085, 2.50, 0.429 K: 0.085, 2.93, 0.407 Matta: 0.050, 1.78, 0.111 Pitino: 0.061, 2.38, 0.313 Self: 0.059, 2.15, 0.077 Williams, Roy: 0.086, 2.76, 0.333 *If you multiply the 1st and 3rd columns by 100, you'll have the number as a percent. So of Boeheim's 888 wins, 4.7% of them came from the Tournament, where as Tournament wins make up 8.6% of Roy Williams's win total (673). Sorry about the unformatted table It looks like Coach K, Tom Izzo, and Roy Williams have the largest proportion of wins coming from Tournament success of this list. On average, Coach K's teams win just under 3 games in each tournament (i.e. Elite 8). For every ten trips to the dance, Izzo's Spartans and K's Blue Devils will have advanced to the Final Four 4 times. Feel free to check my math. If this could be done for many more coaches, it'd be nice to see the standard deviation and mean. This exercise is only good to see relative differences as opposed to statistical differences. You can't really quantify how much less 1.75 tournament wins per appearance vs 2.15 tournament wins is without the standard deviation (sigma). You could do it for a sample size of 10 but there's so much variation that you really couldn't make a conclusion with 95% confidence. [/QUOTE]

Verification: What is a Syracuse fan's favorite color?