Note: this test is kind of silly, in large part because jump shots include both 3s and 2s, and most teams shoot a pretty similar amount of jump shots (most teams that shoot less 3s, for example, shoot more long 2s). This was mostly to follow through on something I started. I can do the same thing but for 3s and for regular season pace as well, and I think those would be more interesting.
So, there's been a lot of talk about what kinds of styles do well in the playoffs versus what kinds don't. An argument came up in another thread about whether jump shooting teams do worse in the playoffs, since the Raptors look to be one of those right now. I had some free time at work on Friday so kind of got myself started on a bigger job than I anticipated.
If I knew how to upload photos or documents, I would put all my work up here, but since I can't, this is the best I can do:
The specific question I was trying to answer:
Do teams that take a higher percentage of jump shots fare worse in the playoffs than their regular season record would predict?
Data:
I only used results from 08-09 through to 12-13 (so 5 years). I could do more once I get the time if that's a common complaint.
I got shot information from here: http://www.hoopdata.com/teamshotlocs...r=2009&type=pg
I got regular season and playoff results from basketball-reference.
Method:
Step 1: Calculate expected series wins.
I used the method from this article to determine expected playoff series wins for each team in each year tested: http://insider.espn.go.com/nba/story...layoff-coaches
I started with each team's regular season record.
From there I regressed that percentage to the "True win percentage" as detailed here: http://www.basketballprospectus.com/unfiltered/?p=967
Using that percentage, I calculated the odds of each team winning one game against their playoff opponent, using the Log5 formula: http://en.wikipedia.org/wiki/Log5
I then modified that probability for home and away games (using a 60-40 split, which is standard in the NBA).
Using those values, I calculated each team's likelihood to win each series that they played using this: http://www.whowins.com/formulae/probformulae.html
Step 2: Calculate difference in actual vs expected series wins
That gave me the odds for each team in each playoff series to win that series, a number between 0-1. If a team lost, they underperformed by "0-chance to win". If a team won, they overachieved by "1-chance to win".
What that gave me was a number for each team in each year in question, indicating whether they performed better or worse than projected based on their regular season record.
Step 3: Jump shooting-teams
I started by defining a jump shot as any shot taken from further than 10 feet away from the basket. I then calculated the percentage of shots used by each team from that distance, and then subtracted the average of the league from that year in question.
Step 4: Regression
What I had at this point was:
1. Team playoff performance above or below expectations for all teams between 2008-2009 and 2012-2013
2. Team propensity to shoot jump-shots relative to the NBA average for all teams between 2008-2009 and 2012-2013.
Then, I removed non-playoff teams from my sample from each year.
Finally, I ran a regression using excel on these numbers.
Results:
R Square of 0.0000404.
This means, 0.00404% of the variation in playoff performance is determined by a team's propensity to shoot jump-shots.
It's extremely unlikely that there is any correlation between jump-shot tendencies and under or over performance in the playoffs.
Conclusion:
Jump shooting teams are no worse on average in the playoffs than they are in the regular season.
So, there's been a lot of talk about what kinds of styles do well in the playoffs versus what kinds don't. An argument came up in another thread about whether jump shooting teams do worse in the playoffs, since the Raptors look to be one of those right now. I had some free time at work on Friday so kind of got myself started on a bigger job than I anticipated.
If I knew how to upload photos or documents, I would put all my work up here, but since I can't, this is the best I can do:
The specific question I was trying to answer:
Do teams that take a higher percentage of jump shots fare worse in the playoffs than their regular season record would predict?
Data:
I only used results from 08-09 through to 12-13 (so 5 years). I could do more once I get the time if that's a common complaint.
I got shot information from here: http://www.hoopdata.com/teamshotlocs...r=2009&type=pg
I got regular season and playoff results from basketball-reference.
Method:
Step 1: Calculate expected series wins.
I used the method from this article to determine expected playoff series wins for each team in each year tested: http://insider.espn.go.com/nba/story...layoff-coaches
I started with each team's regular season record.
From there I regressed that percentage to the "True win percentage" as detailed here: http://www.basketballprospectus.com/unfiltered/?p=967
Using that percentage, I calculated the odds of each team winning one game against their playoff opponent, using the Log5 formula: http://en.wikipedia.org/wiki/Log5
I then modified that probability for home and away games (using a 60-40 split, which is standard in the NBA).
Using those values, I calculated each team's likelihood to win each series that they played using this: http://www.whowins.com/formulae/probformulae.html
Step 2: Calculate difference in actual vs expected series wins
That gave me the odds for each team in each playoff series to win that series, a number between 0-1. If a team lost, they underperformed by "0-chance to win". If a team won, they overachieved by "1-chance to win".
What that gave me was a number for each team in each year in question, indicating whether they performed better or worse than projected based on their regular season record.
Step 3: Jump shooting-teams
I started by defining a jump shot as any shot taken from further than 10 feet away from the basket. I then calculated the percentage of shots used by each team from that distance, and then subtracted the average of the league from that year in question.
Step 4: Regression
What I had at this point was:
1. Team playoff performance above or below expectations for all teams between 2008-2009 and 2012-2013
2. Team propensity to shoot jump-shots relative to the NBA average for all teams between 2008-2009 and 2012-2013.
Then, I removed non-playoff teams from my sample from each year.
Finally, I ran a regression using excel on these numbers.
Results:
R Square of 0.0000404.
This means, 0.00404% of the variation in playoff performance is determined by a team's propensity to shoot jump-shots.
It's extremely unlikely that there is any correlation between jump-shot tendencies and under or over performance in the playoffs.
Conclusion:
Jump shooting teams are no worse on average in the playoffs than they are in the regular season.
Comment