APBRmetrics

[jsill]

Hello, everyone. I would like to announce my new website for basketball stats analysis: http://hoopnumbers.com/. So far, I have computed three statistics which I believe to be new or at least not widely known (my apologies in advance if in fact these are well-known measures):

1) Fast Break Trigger Rate (FBTR). This is the probability that the team scores within 10 seconds after a particular player grabs a defensive rebound. It's a rough attempt to get a sense for whether a player is good at outlet passes and spotting fast break opportunities, or whether he holds onto the ball and is content to hand the ball off to the PG after a few seconds for a half-court possession. The 10 second cutoff is admittedly a bit longer than a classic fast break, but I'm trying to detect cases where an advantage was gained by pushing the ball even though a player may not have coming streaking in for a layup right away. I plan to display the results e.g. for 5 second and 7 second cutoffs later.

Some initial results: Eduardo Najera, Lamar Odom, and Carlos Boozer look good. Zach Randolph, Ronny Turiaf, James Posey and Jarvis Hayes look bad.

2) Miss Offensive Rebound Rate (MORR). This is the probability that a player's misses result in an offensive rebound (free throw misses and misses which result in team rebounds for either team are excluded).

Some initial results: Emeka Okafor, Yao Ming, Dirk Nowitzki and Chris Bosh usually have some of the lowest MORRs on their respective teams.

3) Assisted Field Goal Rate (AFGR). This is the probability that an assist was awarded when a player scored a field goal. I think this can be a useful number for summarizing the player's offensive style. Does he create offense with the ball in his hands or does he focus much more on spotting up and moving without the ball?

Some initial results: AFGRs vary wildly from player to player. Last year, Chris Paul had an AFGR of 14% whereas Kyle Korver's AFGR was 92%.

Coming soon:

- Improvements to Adjusted Plus-Minus which boost accuracy
- Which players' assists lead to 3-pointers most often?
- Which players' block are recovered as defensive rebounds?
- Which players' misses trigger fast breaks for the opposing team?

Thanks very much in advance for any feedback.

[Ryan J. Parker]

It would be cool if you had an RSS feed so we know when you post new stuff.
_________________
I am a basketball geek.

[jsill]

Good suggestion, Ryan! I will try to set that up (maybe tomorrow).

[Crow]

Those are very good targets for the provision of new, public data.

Full league top to bottom performance tables (rather than just by team) for each new stat would be very helpful in my mind or at least show top and bottom 25-50. Or in response to your comment about the value of cross-team comparison on Fast Break Trigger Rate maybe have a table of trigger rate over team average and / compared to league position average and perhaps adjusted by both in some fashion.


On Fast Break Trigger Rate, I see what you see, that mostly guards and even more so perimeter players make up the top 2 for this by team. On Toronto perimeter style bigs were the only case where 2 bigs held the top 2 spots on a team. Love's Fast Break Trigger Rate is run of the mill so far. Amir Johnson was among the ten or less bigs in the top 2 spots.


Other possibilities-

Ultimate team efficiency for plays where the first shot is a mid-range shot by say 3 parts of the shot clock

Or compare ultimate team efficiency for plays where the first shot is a 3 pointer vs mid-range vs. inside by part of shot clock. 82games had a sample run for overall "start position" here http://www.82games.com/locations.htm
but seeing it by clock part and team and player would be new ground.

TS% or team offensive efficiency where the offense recovers the blocked shot.

eFG% or TS% or offensive efficiency by player after the team is in the bonus.

[devin3807]

Cool stuff.

Not too surprisingly 3/4 in #2 are the top offensive rebounders on their squad.

#3 already exists.

[battaile]

Excellent, very interested in the MORR and AFGR, thanks!

[IrishHand]

[Ed Küpfer]

Very good stuff. Also seconding the request for an RSS feed.

Some formatting ideas. Obviously these are minor points.



The cells in the column headers maybe should be a different colour. Percentages should be displayed with one or no decimal places. I would use a different font or font size for the row labels and the values. I would also reverse the order of the columns so that the percentages are on the right side (ie numerator|denominator|percentage).

More substansively:

I'd like to see a master table with all players displayed. I'd also like to see a table with these numbers at the team level. Finally, if you displayed your numbers with graphs, I wouldn't exactly complain.
_________________
ed

[Crow]

Miss Offensive Rebound Rate may not tell much about the player who shot, though some may be more selective about shooting when others are in offensive rebound position naturally or by extra effort; but it can speak to their fit with who they play with and perhaps play design by the coach and execution management, if they both design for the shot and design for trying to get the offensive rebound and follow-up to try to improve the potential offensive rebounding. They sure ought to, at least a bit.

[IrishHand]

That assumes that the merits of scheming for increased offensive rebounds outweighs the costs. A certain Mr. Popovich would, I suspect, disagree with the premise of your argument.

(The championship Spurs teams were almost invariably poor OReb teams.)

[Crow]

I am not surprised by this observation. You can shift the 4 factors around and 1 factor usually won't sink you.

Overall they averaged 20th on offensive rebounding rate in the Tim Duncan era. But taking a few minutes to check I see 3 of the titles came with 3 of their 4 best performances on offensive rebounding for the era (13th, 13th and 16th in the league).

What lesson Popovich drew from this at that time or now I don't know.

They did win once when their offensive rebounding rank was 27th, the last time they won. in 2007.

I think it more likely they get another title doing it the way they did (and perhaps emphasizing it more) the first three times rather than the last one but we'll see what rank they pull on this factor and how they finish with it.

Maybe Pop was changed his assessment of offensive rebounding importance to his all-time low. He made it work once. Last season the Lakers were 3rd best on this. The season before the Celtics were 18th. In 2006 the Heat were 16th. You don't have to be good on this whereas you generally have to be good on eFG% and eFG% allowed but it helps not to be terrible at offensive rebounding. Nobody else has won the title and been worse than 18th on offensive rebounding since when surprisingly the championship Rockets won while being last and next to last. I checked 20 more prior years without finding any belng below 18th and while one or two might have fallen in the bottom 40th percentile of the total teams then the main message is that being weak at offensive rebounding and being champion is pretty rare. It happens about 10-15% of the time. Not being "bad" at offensive rebounding and being champ go together 85-90% of the time. I know what I'd try to do.

Of the 3 most recent times it worked the champ was #1 on 3 point makes twice (Houston) and 6th once (last Spurs). So if you are weak on offensive rebounding it looks like you need that offsetting strength (and partial cause of the weakness). Pop appears to have been doing that but how far his unconcern, disregard or just low prioritization of offensive rebounding went I don't know. But if it went far, it may have gone too far to his own detriment.

[jsill]

Irish Hand:

[jsill]

devin3807:

Thanks for pointing out that AFGR exists. I just noticed it now on 82games.com. I will update my site to credit them tomorrow.

Ed:

Thanks for all the suggestions. Certainly, the formatting is pretty pathetic at this point (it's very 1996, isn't it). I will probably do most of what you suggest before too long. Regarding the master table, that is in the works but I decided to do team-by-team first, since in many ways the intra-team numbers are more meaningful. The league-wide rankings will probably have to be normalized to account for overall pace of play or overall offensive rebounding ability of the team in order to be really meaningful.

[AtlBo]

I cooked up an index for 2 of these ifor anyone who would like to compare more deeply these numbers. The next post will be the fast break info. Most of the time I find that there is generally little interest in this type of thing, but these stats seemed useful to me, and I get alot of benefit from these types of cooked up indexes, myself. This index is simply FG*FGA/100. It represents how much of a team's energy is being put into a player's field goals. Sorry about the length. I would love it if there were an attachment feature....

[AtlBo]

I call these the fast break indices. The index is Defensive Rebounts*Fast Break Scores/100 for each player. This kind of thing helps me condense multiple numbers into one number, and I have found it effective in guaging performance. There is a second short list at the bottom which shows the aggregate measure of each team's ability to get out in the fast break.