Nerdnumbers avatar

The Allan Houston Case Study - Updated

Editor's Note (2016-11-08): The Career Table for Allan Houston's Career has been updated to reflect the difference in Allan's performance while on Art's program and off it.


Art Rondeau is a friend of the site that has some great stories about his time being a shooting coach to various Knicks players. Art approached me recently with one of the most fascinating data sets I've ever seen. Art coached Allan Houston on his shooting during the 1999-2000 NBA season. As luck would have it, it turned into a perfect case study to explore the statistical significance of Art's coaching methodology. And that's what makes this remarkable. Allan's performance on Art's program was so much better than off it; I was continually shocked by the data. I'll let you read through the nitty-gritty statistical details below, but I had to impress just how fascinating these results were.


The following is a case study done on Allan Houston’s shooting efficiency and production during the 1999-2000 NBA season. During this season Allan Houston was coached by Art Rondeau before various games, utilizing Art’s “Mental Zone Program” based on neuro-linguistic programming (NLP). This study shows Allan Houston’s performance was dramatically better in games in which he was coached by Art before the game.

I just want to stress that I did this study based on the documentation Art kept, which has been corroborated in legal settings, as well as in articles written about their work together. In short, my stamp of approval rests on the strength of the data Art provided.

Statistics Glossary

  • 2P% - This is the percentage of shots a player makes from two-point range.
  • 3P% - This is the percentage of shots a player makes from three-point range.
  • Effective Field Goal Percentage (EFG%) – This is a player’s shooting efficiency adjusting for shots from three-point range.
  • Free Throw Percentage (FT%) – This is a player’s efficiency from the free throw line.
  • True Shooting Percentage (TS%) – This accounts for the number of points a player scores adjusted for threes and free throws.
  • Net Points (NP) – This is the player’s total points minus their shots and 0.44 shots per free throw. The reason is that not every free throw ends the possession, and 0.44 is the estimate of free throws that end the play. Net Points is essentially a player’s True Shooting but done to also show their volume of shots in the game.
  • Wins Produced (WP) – This is an advanced stat that estimates the number of Wins a player contributes to their team. It has three key components
    • The value of a player’s box score statistics as they relate to wins.
    • Team adjustments for rebounds, assists, and defense.
    • Comparing a player’s performance to players that play the same position.
  • Points over Par (PoP) - This is a variant of the Wins Produced Formula above. The only difference is that instead of being conveyed in wins it is conveyed in point margin. Essentially this is the number of points by which a team of average players plus the player being evaluated would be expected to beat a team of average players on a neutral court. For reference, 0 is average.
  • Points over Par per 48 (PoP48) – This is a players’ Points over Par adjusted for 48 minutes, or what we’d expect from a player if they played one full NBA game.
  • Simple Rating System (SRS) – This is a team metric. This is the strength of schedule margin of victory for a team. A bigger number implies a better team.

The 1998-1999 NBA Season

Art Rondeau was coaching New York Knicks player Chris Dudley on free throws during the 1998-1999 NBA season. Through this relationship, he was introduced to Allan Houston. Allan asked for Art's assistance to help him break out of an eight-game slump in April 1999. Art coached Allan with his "Mental Zone Program" the day before a game against the Charlotte Hornets.

Allan Houston’s Stats on 4/23 versus the Charlotte Hornets

  • 8 for 12 from Two-Point range (66.7% 2P%)
  • 2 for 3 from Three-Point range (66.7% 3P%)
  • 8 for 8 from the line (100% FT%)
  • 81.0% True Shooting
  • +6.5 Points over Par

The Knicks won the game by 5 points, meaning that Allan's performance and his +6.5 Points over Par were critical. The Knicks made the playoffs by one game, beating the Charlotte Hornets out by one game meaning this game had extra significance. One game is a small sample. There are a variety of reasons Allan Houston's performance could have changed so dramatically. However, this game was the reason that Allan Houston decided to hire Art for the 1999-2000 season.

The 1999-2000 NBA Season

Allan Houston hired Art to coach him before games in the 1999-2000 season. For a variety of reasons, Art coached Allan before 30 regular season games and did not coach him in the other 52 regular season games. It turns out for doing a statistical study of the coaching effectiveness this was actually ideal. Below is a breakdown between the two sets of games.

  Games FT% 2P% 3P% TS% NP WP PoP48 Team Record
On Art's Program 30 84.8% 58.6% 54.2% 66.2% 182.2 5.3 +3.5 20-10 (67%)
Off Art's Program 52 83.1% 43.2% 38.1% 50.8% 13.8 1.3


30-22 (58%)

Regarding further validating this dataset, there are some more important factors

  • 25 games off Art’s program were away games
  • 27 games off Art’s program were home games
  • 16 games on Art’s program were away games
  • 14 games on Art’s program were home games
  • Allan Houston was 28 during the 99’-00’ Season
  • Allan Houston was not in a contract year
  • Allan Houston had the same coach (Jeff Van Gundy) as the year before.
  • Many of Allan Houston’s teammates were the same.

The key behind these factors is it provides similar datasets with few outside factors that might impact Allan’s performance jump.

One more pertinent factor is the opponent SRS

Games on Art’s program +0.157 SRS average for opponent faced
Games without Art’s program -0.35 SRS average for opponent faced

To clarify, Allan's games on Art's program were against stronger opponents on average than those off of it. Using a one-tailed unpaired T-Test on and off Art's program provides the following:

  • 0.000145346 in difference for Allan’s Points over Par per Minute
  • 0.000000247518 in difference for Allan’s Net Points per Minute

To clarify, a value close to 0 (and both of these values round to 0 at three decimal places) implies, with almost 100% statistical certainty, that Allan's performance on Art's program was better than off it. When we remove non-shooting factors, which are captured in Points over Par, it gets even stronger. In short, Allan's shooting numbers on Art's program were drastically different and drastically better than off it.

Team by team Breakdown

The Knicks faced 28 NBA teams during the 1999-2000 season. A further breakdown gets us to:

  • 16 opponents with at least one game with Allan on and one game off the program
  • 10 opponents all games off the program
  • 2 opponents all games on the program

In terms of aggregate performance

  • Against teams that he played both on and off Art’s program, Allan Houston had an average PoP48 of +3.2 in the games he was on Art’s program.
  • Against teams that he played both on and off Art’s program, Allan Houston had an average PoP48 of -3.42 in the games he was off Art’s program.
  • Against teams that he never played while on Art’s program, Allan Houston had an average PoP48 of -1.06.
  • Against teams that he only played while on Art’s program, Allan Houston had an average PoP48 of +5.5. Since this sample is only four games and two teams (the Kings and the Mavericks, both two games) I feel compelled to highlight it.

Additionally, of the 16 opponents Allan played both on and off Art’s program, he only showed better performance off the program against three of them:

The Atlanta Hawks

  • +1.03 PoP48 off the program (2 games)
  • +0.25 PoP48 on the program (1 game)

The Miami Heat

  • +1.5 PoP48 off the program (1 game)
  • +0.9 PoP48 on the program (3 games)

The Milwaukee Bucks

  • +3.2 PoP48 off the program (3 games)
  • +0.280 PoP48 on the program (1 game)

I’ll note that 0 PoP48 is “average” production and Allan did still play above average on Art’s program against all of these opponents. Allan predominantly played below average whenever he was off Art’s program.

The All-Star Break

Part of the key to Art's coaching was the impact made to the 2000 All-Star game. Art was hired to get Allan Houston selected to his first NBA All-Star team. Allan was, in fact, selected to the All-Star team that season. What's more, Knicks head coach Jeff Van Gundy was named the Eastern Conference coach of the All-Star game due to the Knicks record, which was 1/2 game better than the Miami Heat's record at selection time. Here is some data about how Allan's performance impacted the Knicks' record before the All-Star game.

99'-00' Regular Season Pre All-Star Break

  Games FT% 2P% 3P% TS% NP WP PoP48 Team Record
On Art's Program 24 84.0% 58.0% 54.0% 65.6% 139.6 4.2 +3.4 17-7 (71%)
Off Art's Program 23 81.4% 40.6% 38.4% 49.2% -6.84 0.57 -2.3 12-11 (52%)

I applied the same one-tailed unpaired T-Test from before, and it has the following values:

  • 0.000355606 in difference for Allan’s Points over Par per Minute
  • 0.000000647886 in difference for Allan’s Net Points per Minute

Once again, the difference is statistically significant, even more so when we remove any of the non-shooting statistics.

One other test I applied was to see how many game results changed. If we replaced Allan Houston's performance with an average performance (replace his Points over Par with 0), we could ask the following question: "How many games would have had different results if Allan had been replaced with an average player?"

Different Result games off Art’s Program before the All-Star Game

Date Opponent Result Result w/ Average Player
12/5/1999 Denver L(-2) W(+6)
12/30/1999 Washington W(+3) L(-1)
1/8/2000 Cleveland L(-3) W(+3)

Different Result games on Art’s Program before the All-Star Game

Date Opponent Result Result w/ Average Player
11/27/1999 Orlando W(+3) L(-4)
12/20/1999 Charlotte W(+3) L(-1)
1/4/2000 Boston W(+8) L(-1)
1/24/2000 Seattle W(+6) L(-4

If Allan had been replaced by an average player in four games on Art's program, we would have expected the Knicks to have lost. Additionally, the Knicks would have netted an additional win with an average player instead of Allan Houston in the games Allan was not Art's program. Given their significance to Jeff Van Gundy being selected as the Eastern Conference All-Star coach, I feel these games are relevant.

Art's Program in the Context of Allan's Career

Allan produced 5.3 Wins and had a Points over Par per 48 of +3.5 under Art's program. This was similar to Anfernee Hardaway on a per-minute basis in 1999-2000 or James Harden in the 2015-2016 season. Here is a breakdown of Allan's Wins and PoP48 for his entire career by season.

Season Games Minutes per Game Wins Produced PoP48
1993-1994 79 19.2 -5.1 -8.1
1994-1995 76 26.3 2.4 -1.3
1995-1996 82 37.5 4.1 -1.1
1996-1997 81 33.1 0.1 -3.0
1997-1998 82 34.7 1.6 -2.2
1998-1999 50 36.3 0.6 -2.6
1999-2000 82 38.6 6.6 0.0
1999-2000 - On Art's Program 30 40.0 5.3 +3.5
1999-2000 - Off Art's Program 52 37.9 1.3 -2.1
2000-2001 78 36.6 4.0 -1.0
2001-2002 77 37.8 0.8 -2.7
2002-2003 82 37.9 3.1 -1.6
2003-2004 50 36.0 0.5 -2.7
2004-2005 20 26.6 -0.2 -3.6

*As noted the table has been updated to reflect Allan's 2000 numbers both on and off Art's program. As additional clarification, the 30 games under Art exceeded the production Allan had in any other NBA season in his career.

In the 30 games on Art's program, Allan Houston produced more wins than he was able to produce in any other full season in his NBA career. Of course, his PoP48 for the other seasons he played never reached his 1999-2000 level. We can note that statistically, Allan Houston was, in fact, a below average player for almost all of his career, except for the 1999-2000 season under Art! I should also stress that Allan Houston's good performance was almost entirely driven by his scoring. Allan's non-scoring stats were below average for a wing both for the 1999-2000 season and for his career.

Another interesting career note is that if we look at Allan Houston's top 100 regular season game, 12 of his top games were those coached by Art during the 1999-2000 season. Of course, his game against Charlotte on 4/23/1999 is also on the list. Thirteen games may seem small, but I'll note that Allan was only coached by Art for 31 games and had a career of over 800 games! I've included asterisks next to which games were in the top 100 below.

Final Note

When Art first approached me about this, I'll be candid. I was expecting to find some mostly noisy data that I'd have to shrug my shoulders at. As you can see the performance difference in the 1999-2000 season alone is absurd. Looking across Allan's entire career makes the contrast even more apparent. I'll reiterate that my endorsement of this report is based on Art's data. That said, I don't know if there's any player in NBA history with a 10+ season career that has thirty games in one season that are so much different than the rest of their career. All I can say to Art is thanks for one of the most interesting case studies I've seen in NBA analytics history.


Art Rondeau Bio and Contact Info

Provided courtesy of Art Rondeau:

"Art Rondeau has developed two programs that have achieved unique results with NBA and other elite athletes. His free throw program quickly helps chronically-bad free throw shooters overcome the physical causes of most misses and significantly improve their shooting percentages. Clients utilizing Art's "mental zone program", based on Neuro-Linguistic Programming (NLP) and the subject of this case study, have, among other things, become an NBA All-Star, won a World Cup skiing Gold Medal, and upset a world-ranked WTA tennis player. You can read Art's blog - Game Time at the Garden of Good and Evil - at and follow him on Twitter (@artrondeau)."

Appendix - Allan Houston Games on Art's Program

Here is the list of dates for the games that Allan was prepared for by Art (games marked with an asterisk denote one of Allan's top 100 regular season career games using Points over Par):

11/2/99 vs CLE, 11/3/99 at CHI*, 11/5/99 at DET*, 11/6/99 at CLE, 11/8/99 vs MIL, 11/11/99 at MIN*, 11/14/99 vs MIA, 11/16/99 at DEN, 11/19/99 at PHO, 11/20/99 at GSW, 11/27/99 vs ORL*, 11/29/99 vs DAL, 12/12/99 vs BOS, 12/14/99 at HOU, 12/16/99 at DAL, 12/20/99 vs CHA , 1/4/00 vs BOS*, 1/7/00 at ORL, 1/12/00 at DET, 1/24/00 vs SEA*, 1/30/00 vs SAC*, 2/1/00 vs ORL, 2/6/00 vs MIA*, 2/7/00 at CHA, 3/17/00 at CHA*, 3/21/00 at IND, 3/24/00 vs ATL, 3/28/00 at SAC*, 4/9/00 at MIA, 4/18/00 vs DET*

geez, if the contrast was that clear why wouldn't every player hire this guy?
Anything about days off leading up to games with or without this guy?

Just curious, it's a remarkable result regardless.
Completely off topic but it looks like Tim Duncan called it quits.
I'm not shocked to find that out given general hiring practices of NBA teams. Art will likely be telling more of his story on his blog in the next few months. It's fascinating stuff.

Good question! Hadn't looked at that one. Given diversity of game types I wouldn't expect a ton, but I'll look into it.

Helluva a player!

Agreed. There's a lot of things like that. I think, in addition to what Dre said, players are resistant to some of those types of things. I think they're so obsessed with their macho workouts they discount a lot of the mental stuff.

Example: most players don't work on meditation. It really helps. Kareem was a pioneer in doing yoga.

How Curry plays looks a bit meditative--his mechanics actually reminds me a lot of Tai Chi (or other inner martial arts--e.g. jiu jitsu, judo, aikido) which I've practiced for a few years. It's truly inexplicable how much Tai Chi has helped my game (and a couple other people I know who've taken it up). I'm particularly surprised no one ever talks about doing Tai Chi in particular with mediation (certain styles--the ones where the general stance is like the defensive stance in basketball, which many of them are, including the kind I study). It not only works on focus and putting emotions/thoughts out of your mind, and also is entirely focused on upping leg/core strength (and back, and other parts of the body), and is all about balance--which comes into play with just about everything in basketball. Sorry to go on a weird tangent--it's just that, having studied Tai Chi and seen the effect it's had on my game, I really wonder how much it might help even a pro player? (And I really don't mean it just because I do Tai Chi. It just REALLY fits with basketball. Other sports too--my teacher is really into golf, and it helps his game a ton. Watching things like tennis, it would definitely help too. It just massively helps with leg strength, agility, body awareness, focus, and energy.)

I know the Warriors have their own weird training systems too--and only Curry and Green use them mostly. A lot of the other players don't use it, and the Warriors doubt (right or wrongly) that many players have the mental discipline to. They involve using these goggles with flashing lights of different colors while you're shooting around that help you learn to automate your game. I know that's a bad explanation--if you're curious, ask me for a link--it'd take me too long to explain the whole thing. Curry does other things like sensory deprivation floats too (you can probably google it--if not I can probably find a link.)

(Sorry to keep using the Warriors as examples. I don't mean it as favoritism--their just my local team so I know them better)

Wow...that went WAY longer than I meant. Sorry, guys. My bad.Editor's note: We'll let it slide this time, I think Art will be happy to get so much response already.
Thanks. Again, really sorry. Realized I went way overboard the second after I posted.

I would think Tai Chi would particularly help with hip checks into the scorers table! ;) I do actually really wonder how much teams (the good ones) actually track this stuff. You'd hope they all do it but as far as I can tell they normally just hire a coach from a team currently doing well as opposed to looking at specific changes a coach might have instituted.
Haha, also true
These increases are pretty high. Fuller disclosure would of course require data not selected by Art but it still compels further investigation.
Why don't more players take advantage? I think Art discussed some of this on his turn on the podcast. Pros have been coached many times by many people by the time they are pros, so skepticism that one person has the way is not unwarranted.
It's a bit weird to publish article like this without confirmation from multiple unbiased sources (despite clear disclaimer in introduction). But it's not half as weird as seeing that comments section morphed into New Age festival.
Randomname, you have turned into king of long meandering and usually unsubstantiated comments.

Whoa, Tai Chi and New Age have nothing to do with each other. My grandmother, dad, and a few uncles started Tai Chi at some point while they were getting older and it definitely helps with balance. I'd suggest that it helps more of the muscles involved with balance equally than just about any other exercise regimen I know of. To the extent that you want to add meditation into it is up to you. As for the article it was pretty explicitly stated that not much work went into it but that it seemed interesting at first glance. I viewed this piece more as a starting off point as opposed to a finished product. More of a "hey, I just read a few chapters of this book and it seems pretty interesting!" as opposed to a "you have to read this trilogy."
I have a couple of comments. Let me start off saying that this is a nice write-up and I fully believe that the data speaks for itself in telling this story, but I have a couple of minor quibbles (because I am super nitpicky about my stats):

1) Why are you using one-tailed T-tests, instead of the more traditional two-tailed ones? It just seems like a rather artificial attempt to inflate the significance of the result. By using a one-tailed test, you are essentially building into your model the assumption that the Points over Par can NEVER be lower under Art's program than off of it, and forcing the test to disregard any instances where that might be the case. Considering how small the p-values are (I assume those are the numbers you are reporting for the T-test, though you don't actually specify), it's not like using the two-tailed test would actually change the interpretation of the results (it will almost certainly still be highly significant), it just seems like bad form to be using a one-tailed test in this type of analysis. Typically, you only really want to use one-tailed tests in the case where the an effect in the other direction is actually biologically implausible, rather than simply not the direction you "want" the effect to go in. Related, you report a lot of average Points over Par under various breakdowns (games on Art's program, games off, and other subgroups), but you don't report any measure of uncertainty to give those numbers context. Yes, the average PoP was +3.2 for games on Art's program against teams he played both on and off the program, but what's the standard deviation of that? What's the confidence interval of that mean? I assume from the T-test results that the standard deviations are fairly small (otherwise you wouldn't get such small p-values, even using the one-tailed test), but I don't see any explicit mentions of the variability.

2) Speaking of variability, I think there is an interesting next-step comparison for this analysis. You write, "That said, I don't know if there's any player in NBA history with a 10+ season career that has thirty games in one season that are so much different than the rest of their career." Now, obviously I don't expect you to dig through every player in NBA history, but what about just looking at other Allan Houston seasons? For example, just the season before or the season after the one 1999-2000 one in question. How variable were his PoP performances in those seasons? How many high PoP games can you find in those seasons, as a means of comparison with the 30 or so from this specific season?
I just noticed the Editor's note to randomname's comment earlier. Apparently long comments are frowned upon here? I honestly didn't realize (I often read the site but rarely visit the comments). My apologies if my previous comment was too long! Editors can prune or delete as they see fit.

Sign in to write a comment.