Old Bear
Postdoc
Posts: 3994
Reg: 11-23-04
|
11-02-15 09:59 PM - Post#194219
I just watched PTI. Tony and Mike decried "Ivy League Analytics". I doubt they were alluding to Mike James. I certainly respect Mike and his views. However, I also agree with Cameron McCormick. If I may quote him "Stats are like bikinis, they show a lot, but they do not show everything."
|
IvyHoopsFan01
Freshman
Posts: 68
Age: 63
Reg: 10-28-14
|
11-03-15 04:50 PM - Post#194239
In response to Old Bear
Statistics are just like mini-skirts, they give you good ideas but hide the most important thing
|
IvyHoopsFan01
Freshman
Posts: 68
Age: 63
Reg: 10-28-14
|
11-03-15 05:09 PM - Post#194240
In response to IvyHoopsFan01
I agree with MJ that stats truly do rule ...
Statistics show that every two minutes in England a woman gives birth. This woman must be stopped.
Statistics show that at the age of 80, there are 5 women to every man. What a lousy time for a man to finally get these odds.
|
Tiger69
Postdoc
Posts: 2814
Reg: 11-23-04
|
11-04-15 11:02 AM - Post#194266
In response to IvyHoopsFan01
The odds are good. But, the goods are odd.
|
IvyHoopsFan01
Freshman
Posts: 68
Age: 63
Reg: 10-28-14
|
11-04-15 01:12 PM - Post#194275
In response to Tiger69
A statistician's wife has twins. He was delighted, and he called to tell his minister the good news.
"Excellent!", said the minister. "Bring them to church on Sunday and we'll baptize them."
"No," replied the statistician. "Let's just baptize one. We'll keep the other as control."
|
TheLine
Professor
Posts: 5597
Age: 60
Reg: 07-07-09
|
11-04-15 02:10 PM - Post#194277
In response to IvyHoopsFan01
In all seriousness, analytics are better than the alternative and that's all that matters. Teams use more sophisticated models than the publicly available ones.
Also Tony and Mike are idiots. Maybe a step above Screamin A. Smith and Skip Bayless but still idiots.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
11-04-15 02:25 PM - Post#194281
In response to TheLine
All "analytics" provides is a toolbox to test whether hypotheses that you generate by watching or participating in something are some percentage likely to be true or not. You'd be terrible at creating hypotheses about something if you don't observe it frequently. You'd be terrible at testing your hypotheses if you don't have a deep understanding of how to generate probabilistic outputs from objective information.
I think what people are arguing when they speak out against analytics is this unrealistic strawman where numbers somehow flow into a computer, the computer returns a result and people blindly follow whatever it says. That, of course, would be silly and destructive. But to decry the general use of real, analytically-driven processes is to make yourself look foolish, because all you're loudly proclaiming is that you are completely shut off to one source of (often valuable) information.
|
TheLine
Professor
Posts: 5597
Age: 60
Reg: 07-07-09
|
11-04-15 02:43 PM - Post#194284
In response to mrjames
The only people who think analytics are used blindly are the same idiots who reject analytics and prefer 'gut feel'.
Those people get a couple things wrong. One is that analytics is a nuanced field where models are constantly improved upon. Another is that they neglect that people with good 'gut feel' typically use a form of analytics - for example the football coach who is good at breaking down game tape or the baseball manager with index cards of past outcomes.
|
SRP
Postdoc
Posts: 4911
Reg: 02-04-06
|
11-04-15 04:09 PM - Post#194286
In response to TheLine
Systematic analysis of things that were long done by intuition has often yielded immediate gains in insight (and competitive performance), as in WWII operations research, which for example vastly improved submarine hunting, or in Thorp's study of blackjack. It's just silly to not use analytical methods or to stick to inferior analytics (which is what many "old school" types do).
The above examples, though, were based on a thorough causal understanding of the strategies and outcomes. Sometimes it seems to me that we have inductive regularities without causal justification trumpeted as reliable guides when they are not.
|
IvyHoopsFan01
Freshman
Posts: 68
Age: 63
Reg: 10-28-14
|
11-04-15 05:31 PM - Post#194289
In response to SRP
While I enjoy finding humor in analytics, it clearly is playing an ever increasing role in sports, business, healthcare and wide range of other fields. When I went to business school, our stats class was not considered one of the more important classes and was not often applied early in our careers. Today, for all of the top 10 MBA programs, the students have big data and data analytics incorporated into numerous classes. MBA students can’t get enough of big data. Neither can business recruiters and even venture capital firms. Intuition, leadership and communication skills are still critical, but for future business and sports management professionals, it would be foolish to not become highly proficient in data analytics.
Glad MJ is a regular contributor to this site.
|
HARVARDDADGRAD
Postdoc
Posts: 2691
Loc: New Jersey
Reg: 01-21-14
|
12-06-15 04:54 PM - Post#196801
In response to IvyHoopsFan01
Looking at the season thus far, strength of schedules are interesting.
Let's start with Penn at an SOS of 330. Irrational exuberance?
How about 324 (Dartmouth), 322 (Princeton) and 308 (Brown)?
I can certainly understand Cornell (240) and even Columbia (161) and Yale (147), who have scheduled some good games.
Harvard's schedule (56) was obviously locked in well before Siyani's injury. It would be fun to see next year's Crimson squad play this schedule. Other than Holy Cross, losses to nationally recognized teams (Kansas, Providence, BC, UMass, Northeastern). Hope opponents like these are still willing to risk playing Harvard next year.
|
palestra38
Professor
Posts: 32811
Reg: 11-21-04
|
12-06-15 06:25 PM - Post#196809
In response to HARVARDDADGRAD
Aren't you showing irrational exuberance about freshmen who have not played a game and a PG coming off an ACL?
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-06-15 06:42 PM - Post#196810
In response to palestra38
More on this on the next episode of the 14GT podcast!!!
|
HARVARDDADGRAD
Postdoc
Posts: 2691
Loc: New Jersey
Reg: 01-21-14
|
Ivy League Analyitics
12-06-15 08:15 PM - Post#196817
In response to palestra38
I don't think so. I was shocked by the result in Kansas and the play of the youngsters. However, that doesn't make me think Harvard is competitive for the title this year. They're still really 1-6. What we did learn yesterday is that they can get hot and scare ANY team.
I'm just pleasantly surprised to see that it can be fun when you have low expectations! There's a chance these freshmen are merely playing for a spot on next year's roster. Sad, but true.
On the other hand, Edosomwan, Okolie, Cummins, Egi and Miller are solid. They've just been overlooked by most in the shadow of the 8 players who are not here but were last year.
Edited by HARVARDDADGRAD on 12-06-15 08:17 PM. Reason for edit: No reason given.
|
jadwin
Sophomore
Posts: 191
Age: 74
Reg: 01-14-15
|
Re: Ivy League Analyitics
12-09-15 11:54 AM - Post#196966
In response to HARVARDDADGRAD
Kenpom.com has the following rankings for Ivy Teams as of today:
Yale #76
Princeton #85
Columbia #111
Harvard #156
Dartmouth #228
Brown #242
Penn #264
Cornell #274
The current ranking is similar to the pre-season Ivy League poll. Time will tell how it plays out.
|
HARVARDDADGRAD
Postdoc
Posts: 2691
Loc: New Jersey
Reg: 01-21-14
|
12-09-15 01:21 PM - Post#196977
In response to palestra38
Sorry, I misunderstood your post. Yes, I am concerned that the 2016-17 will match an older and experienced Princeton squad (possibly looking to repeat) against a freshmen laden Harvard team led by a point guard who relies on quickness but is coming off a knee injury. It could be an amazing matchup and - frankly - a cinderella story for a Princeton basketball team taking on the 'evil empire'/Kentucky of the Ivy League. It should be fun and interesting. Of course, this completely discounts the returning Harvard players (Edosomwan, Egi, McCarthy, Johnson), and their story is only starting to be written. Will Harvard compete this year? Will they be a spoiler?
I am wondering what other teams may be in the hunt next year. Yale, Columbia, Brown and Penn stand to lose their key contributors. Harvard and Princeton much much less so.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-09-15 03:15 PM - Post#196981
In response to HARVARDDADGRAD
Harvard's 2016 class is projected for 57 win shares over its four years in Cambridge.
The class with the highest accumulated win shares in its four years is Cornell 2006 at 62.5 WS.
The most the model overpredicted any team was 2011 Penn (24.4 predicted WS, but only 3.2 produced). So, even if the 2016 Harvard class matches that Penn class as the most overrated of all time, it would still produce 36 win shares, which would make it the sixth most productive class of all time (back to 2002). Here are the top 5 classes in terms of actual WS produced:
Cornell 2006 62.5 WS
Penn 2003 56 WS
Harvard 2011 55 WS
Princeton 2009 48 WS
Harvard 2009 43 WS
And here are the Top 10 predicted win share classes with actual WS produced:
1. Harvard 2016 57 Pred/(TBD Actual)
2. Harvard 2011 40 Pred/55 Actual
3. Harvard 2009 35 Pred/43 Actual
4. Harvard 2012 33 Pred/20 Actual (TBD final)
5. Penn 2003 29 Pred/56 Actual
6. Princeton 2003 28 Pred/9 Actual
7. Penn 2005 27 Pred/17 Actual
8. Brown 2004 26 Pred/31 Actual
9. Columbia 2004 25 Pred/32 Actual
10. Penn 2011 24 Pred/3 Actual
More on this on the podcast to come later this week.
|
whitakk
Masters Student
Posts: 523
Age: 32
Reg: 11-11-14
|
12-09-15 03:21 PM - Post#196982
In response to HARVARDDADGRAD
Yale should finish top half, because, duh, it's Yale, and because they've recruited well to build around Mason. Seems a stretch to compete with H/P though.
Hard to separate the rest of the league at this point. Penn will keep improving even without DNH. Brown will have tons of perimeter talent but no proven bigs. Dartmouth might have two stars and no depth. Columbia will be a wild card.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-09-15 03:27 PM - Post#196983
In response to mrjames
Also, in terms of the model, here's how it would rank the last five Ivy classes in terms of expected four-year WS:
2016:
1) Harvard 57 WS
2) Yale 18
3) Penn 10
4) Columbia 6
T5) Brown 3
T5) Cornell 3
T5) Dartmouth 3
8) Princeton 0
2015:
1) Cornell 23
2) Harvard 17
T3) Columbia 16
T3) Penn 16
5) Yale 13
6) Dartmouth 10
T7) Princeton 10
T7) Brown 10
2014:
1) Princeton 17
2) Penn 16
3) Harvard 14
T4) Dartmouth 13
T4) Brown 13
T6) Yale 10
T6) Cornell 10
8) Columbia 6
2013:
1) Penn 17
T2) Princeton 16
T2) Brown 16
4) Harvard 14
T5) Columbia 13
T5) Cornell 13
T5) Dartmouth 13
8) Yale 10
2012:
1) Harvard 33
2) Dartmouth 16
3) Penn 14
T4) Yale 10
T4) Columbia 10
T4) Cornell 10
7) Princeton 6
8) Brown 3
|
jadwin
Sophomore
Posts: 191
Age: 74
Reg: 01-14-15
|
12-09-15 04:11 PM - Post#196987
In response to mrjames
Some fairly significant discrepancies for a number of years as to Actual vs. Projected. Science vs. Art??
|
