1. Thanks for taking this approach. It helps a great deal vs other alternatives.

2. IMO, it isn't strange at all that Del Harris, as an accomplished veteran coach, wasn't able to ascertain correctly the reasons for Daniels' having this effect in those games. The fact is ... relatively few coaches - let alone "average fans" - have the ability to gauge individual player match-ups with a high degree of accuracy, and it makes little difference what level of basketball we're talking about, e.g. little tykes, elementary school-age, high school-age, college, FIBA or the NBA. The vast majority of coaches at all levels of the game DO NOT have the ability to "coach-to-specific match-ups" when the red light gets turned on [so-to-speak]. In sharp contrast, the majority of coaches would be accurately categorized as being "system-oriented" individuals with the ability to create and see effectively the "bigger picture" for their teams but without the required eye to see the type of detail it takes to hone in with acuity on the proper match-ups in a particular contest.

3. I agree ... Winston seems to have a habit of making very broad statements which are not necessarily supported by a closer examination of the specific circumstances in question.

4. re: sample size validity

This is really the crux of the matter, when you get right down to it.

e.g. whether speaking about Telfair [in general], Glavine-vs-Redmond [specifically], or going back to your prior example of the cancer patient.

IMO, the more significant questions then become these:

i. Which is more relevant and therefore more important to analyze in further detail:

A. The results which would seem to conform with the statistical norms;

or,

B. The results which would seem NOT to conform with the statistical norms?

ii. Why is sample size considered to be a relevant factor, in this equation, in the first?

i.e. By definition, this is a numerically limited environment where authentically representative sample sizes are not to be found.

5. re: Glavine vs Redmond

---------------------------------------------------
But in 51-PA sample, his "true skill" vs, Glavine could literally be anything, and we wouldn't know because they haven't faced each other enough to rule out chance as the driving force behind his 1.075 OPS. The same goes for matchups in basketball... Maybe you could glean some predictive power out of general matchups (how a player does vs. good, average, and bad defenses, for instance), but very specific head-to-head matchups simply don't give you enough sample size for confident predictions. Just like you'd be better off using Redmond's career average to predict his next PA vs. Glavine, you're just better off using a player's season-long NBA stats to predict a playoff performance than some matchup data derived from a miniscule sample.
---------------------------------------------------

IMO, the best predictor of future performance involves neither:

I. Glavine vs Redmond specific stats, exclusively

or,

II. Redmond's career stats, exclusively

... but, quite possibly:

III. An unique blend of the following three:

- Glavine vs Redmond stats +
- Redmond's career stats +
- Highly specific anecdotal observations pertainig to the Glavine vs Redmond match-up, including but not limited to the following, for example:

a. How Redmond reacted [i.e. physically, mentally and emotionally] to the pitches being thrown by Glavine that he was able to contact and put into play?
b. How Redmond reacted to the pitches being thrown by Glavine that he was able to contact but was unable to put into play?
c. How Redmond reacted to the pitches being thrown by Glavine that he was unable to contact because they were deemed to be "balls" [as opposed to being "strikes"]?
d. How Redmond reacted to the pitches being thrown by Glavine that he was unable to contact because they were deemed to be "strikes" [as opposed to being "balls"]?
e. How the ball actually came off Redmond's bat in "a" and "b"?
f. How the fielders were positioned?
g. How the bases were occupied?
h. How the specific game circumstances/strategy may have influenced/effected the resulting outcome[s]?

6. Are you familiar with each of the following:

Principle of Inertia?

The Monty Hall Paradox?

Hot and Heavy: About NBA Shooting and the merits of Bill James Fog article">?

There's a tonne of terrific material being produced today concerning the game of basketball and the proper way in which the role of statistical probability needs to be viewed, relative to the "real-life" strategy, tactics and abilities of the participants involved with the game itself.

When it comes to actually understanding what "the numbers have to say", place me in the group that says, "Those 4 students from Cornell who knew correctly when they were, in fact, 'hot' had the kind of 'accurate awareness' it takes to win a championship," which is where the real magic is to be found in understanding how individual match-ups work in the NBA game.

i.e. How did those four [4] students manage to do what they did? ... as opposed to focusing on, "How everybody else did [and came out with randomness]?"

I apologize, in advance, if none of this is interesting to others and for wasting their time.

Winston found that "for whatever reason", Daniels was killing them against the Spurs. But isn't it strange that Harris, a veteran coach, wasn't even able to identify what exactly it was about Daniels that was killing them? According to Winston, he just said something to the effect of, "I don't know why, but it's in the data, so we'll act on your advice". That strikes me as strange -- if it was a specific matchup, what didn't the guys whose job it is to study film and see matchups know what it was? Also, oddly for a statistician, Winston seems to jump to wild conclusions based on small samples; he considers Sebastian Telfair to be an "outstanding point guard" based on 1 year's worth of adjusted plus/minus data, when A) that data is notoriously noisy for single year samples, and B) nothing about Telfair's past performance in Winston's own stat (much less every other metric out there) suggests he's even an average PG, much less an outstanding one. So my second thought about the Daniels thing was, exactly how much of a sample could there have been for Daniels vs. the Spurs when he was a 3rd-year player? He had only played 197 career minutes vs. SA going into those playoffs, which is nowhere near enough of a sample upon which to judge a player, especially in a volatile stat like APM. And that's if they were using his career numbers; something about Winston's words actually made me think they were using just numbers from that series, which would about as informative as flipping a coin to decide whether he should play or not.

The point is, whenever you get into these very specific matchups, you're going to run into serious sample-size issues. It's like in baseball -- maybe career backup catcher Mike Redmond really does "own" future Hall of Famer Tom Glavine:

http://bbref.com/pi/shareit/wPHjW

But in 51-PA sample, his "true skill" vs, Glavine could literally be anything, and we wouldn't know because they haven't faced each other enough to rule out chance as the driving force behind his 1.075 OPS. The same goes for matchups in basketball... Maybe you could glean some predictive power out of general matchups (how a player does vs. good, average, and bad defenses, for instance), but very specific head-to-head matchups simply don't give you enough sample size for confident predictions. Just like you'd be better off using Redmond's career average to predict his next PA vs. Glavine, you're just better off using a player's season-long NBA stats to predict a playoff performance than some matchup data derived from a miniscule sample.

I've read your responses and I've refrained from participating further in this discussion because it seems that the only thing you want to do is (falsely) accuse others for lacking reading comprehension skills and look your nose down upon this blog for the work that it does. It doesn't exactly endear anyone to your point-of-view, whether there is truth in it or not.

Unless you truly want to engage in a meaningful conversation about the use of statistics in basketball (and not just engage in the activities that I mentioned above), I kindly suggest that you take your agenda elsewhere.

1. I'm not trying to bully anyone. 2. I'm just trying to ensure that my points are not being misconstrued by others. 3. I think your blog is terrific. 4. I didn't say that what you're doing is pseudo-basketball analysis. 5. From my perspective ... it can be impossible/very difficult to engage others in discussions about specific subjects when "the core" of their analysis might be considered to be wrong/invalid and they become "defensive" [annoyed?] to the point of not wanting to talk about it any further.
6. Insight always involves a two [or more] way street. 7. A "yes-man" I am not. 8. You did raise the calibre of the discourse with #18. What I'd like is that the calibre now stays at that level and not be lowered by talking about something which isn't really connected to what we've been discussing here, i.e. what Winston & Harris had to say about the role of stats and individual match-ups when seeking further understanding about the game. 9. This is the part of what you wrote in #18 that I think has the most relevance:

I'm reminded of when people say "Statistics mean nothing to the individual", meaning that across a large sample of many people -- say, people diagnosed with a type of cancer -- some will live, some will die, and you can put a percentage on both groups. But for one person, it's not like that, you can't be 55% dead and 45% alive; you either survive or you don't. So in many ways, probability is right everywhere but wrong in any single place, because while anything can conceivably happen, only one thing actually happens.

and which needs further discussion/clarification before moving on, since I do not believe that "randomness" is the only factor which accounts for the difference that sometimes exists between the separate notions of "probability", "certainty", and "what actually happens" in specific instances.

10. In my travails on-line ... what I find is this:

Despite what many may profess to be the case, there are very few individuals in this world who really truly madly deeply want to discuss how something actually works with another person who might disagree with their perspective, yet, still be polite, fact-based, and interested in hearing what others have to say about the subject. I thought you might be such a person.

1. I'm not nitpicking with [or picking on] Anon.

2. Yes, you mentioned "scouts"; but, Winston and Harris did not. That is a simple fact which I stated. While Anon might well understand what you said, my point was that he did not seem to understand what was said by Winston and Harris.

3. What's being talked about, in this instance, is what Winston and Harris think, re: stats and the importance of individual match-ups in a NBA game; not what I happen to
think about them.

4. Winston and Harris were not talking about stats and the importance of individual match-ups in a situation where "the pre-game numbers predicted a different result" ... i.e. after-the-fact, due to a need for increased "order".

5. While some in the basketball community [you and perhaps others who are like-minded] may well be terrified of randomness, it's simply not accurate to suggest that "everyone" fits into this one category.

-------------------

From my perspective ...

I. Understanding how individual match-ups actually work in the NBA game is fundamental to sound basketball analysis.

II. Unfortunately, much of the pseudo basketball analysis which takes place today is mis-directed, due to an over-reliance on different forms of statistical calculation and averaging.

Please try to not be insulted by what I'm going to say next but ... IMO ... you still do not understand the ramifications of what Winston actually had to say regarding the role of average stats and individual player match-ups ... which, btw, I have yet to read articulated properly anywhere on-line.

e.g. The word "scouts" wasn't mentioned by Winston or Del Harris, yet you chose to include it in your reply for some reason, as a factor in the relationship that exists between "stats" and the accurate evaluation of basketball players.

"Hopefully Anon takes the time to read your answer carefully."

It was a fantastic answer. It was also something I've known and read all along about the importance of stats AND scouts.

1. Hopefully Anon takes the time to read your answer carefully.

2. According to elite level basketball coaches, it's a game based upon individual match-ups ... i.e. between players, 5-man units, main rotations, coaches, and teams.

3. It's a very good thing that you do not want to be surrounded by yes-men; neither do I. :-)