APBRmetrics

[Mike G]

We've discussed home assist bias in several places, how to adjust our stats for them, etc; and somewhere in there Mutombo (re: shotblocking) was mentioned. This makes me wonder if home bias in other counted stats could be systematically messing with boxscore-based player ratings.

The 2008-09 home/away splits for teams are a snapshot. I've just divided home totals by away totals. The ratio 1.10 means a team got 10% more of whatever in home games. Home and away minutes played are within 1.6% for all teams. Ranked by Wins (home/away), just in case there's some correlation with 'playing better'.

[Mike G]

So how many supposed blocked shots might Mutombo have gotten as a 'bonus' from home scorekeepers? Assuming half his minutes are played at home, a simplified adjustment factor is put to his annual Blk% (lifted from b-r.com).

Home and away blocks are listed; the H/A is the ratio of Home and Away blocks per minute.

[Mike G]

[Mike G]

This past season, Chris Andersen had a stupendous Block% of 9.3 . Of players with at least 1400 minutes (since 1974), only 3 players have surpassed this : Manute (4 times), McIlvaine, and Bradley.

Denver was 2nd in the league in the % increase of blocks at home (vs road). Here are the league's top 5 in Blk%, along with their home/away totals; ratio of player home/away Blk/Min (also for team); and a couple of possible 'adjusted Blk%' based on individual and team H/A ratios.

[erivera7]

Interesting that Dwight doesn't have much of a home bias, with regards to his block totals.

It's fair to say that Howard's block totals are more of a true representation than Andersen's block totals.
_________________
@erivera7
I cover the Orlando Magic - Magic Basketball

[Mike G]

Dwight had a negative home bias, getting just .97 as many Blk/36 as he got on the road. On the chance he's just a slacker in Orlando, I also calculated an Adjusted Blk% based on the team's 1.07 home benefit.

Dwight had more than half of the team's blocks (231/439), so the rest of the team must have enjoyed about 19% more blocks while at home.
_________________
`
36% of all statistics are wrong

[DJE09]

Mike,
Didn't the last thread about home court bias also have some comment about "star" treatment on the road - ie. the guys from the sonics was happy to give Stockton anything that looked remotely like an assist, but not to Utah generally. This would mean we might expect Dwight to get the bonus blocks on the road as well as at home?

I noticed that for some teams that there was possibly a relationship between Blk / Stl and fouls - eg. MIA has a STL ratio of 1.13 and Blk of 1.05 vs foul of 0.91 so they got more blocks and steals at home whilst called for less fouls. These things seem related in my mind, and Indicate a bias due to referee's call rather than score keepers (ie. ref let player away with a 'steal' at home that they would call a foul on the road).

I am not suggesting this as a reason for all the Block difference, but think that it could account for some of it; since presumably the score keepers can't be responsible for the difference in PF.

[Mike G]

All very good points. In theory, any turnover could be credited a 'steal', if nobody's looking (or nobody cares); any missed FG might have been a 'block'.

I put in H/A Wins and PF just for the reason you suggest -- as indicators that a team really may play much better and/or get favorable reffing at home. Scorekeeper favoritism (on the road) for select players almost certainly does happen. In the case of DHoward, neither his personal H/A ratio nor the team H/A, taken alone, will make that correction.

Or maybe we could speculate that his .97 (H/A), being .10 short of the team's 1.07 (for Blk) represents that, on top of his getting 7% 'extra' blocks when at home, he's getting 17% extra on the road. This requires an assumption; but one might ferret out these 'road favorites' in just this way.

There's also the possibility that Blk/Stl/TO that should have gone to Player A are systematically being shunted to/away from Player B. Or that a given player is both a home favorite and a road pariah (Chris Andersen?), to account for up to twice as many (blocks per36) at home.
_________________
`
36% of all statistics are wrong

[HoopStudies]

The danger with this analysis is that teams and individuals do play better at home. So how much of what you're calling "bias" is actually bias vs playing better? It's not right to label it as "bias." They are "differences", with how much being associated with scorekeeping vs playing different not known.
_________________
Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers.

[Mike G]

We don't know how much is simply 'scorekeeper bias', and that's why I've made multiple suggestions for adjustment. Still, it's possible that there are a wide range of adjustments that would be preferable to none

The opening table was ranked by 'home improvement' -- Home Wins/Away Wins. Condensing that table into 3 equal parts, top to bottom, we get the following averages within the tiers:

[Ryan J. Parker]

Just because we know scorekeepers score things differently doesn't mean we can attribute these differences to them. It is important we note the differences and try to further understand what may be the cause of them, or rather what environment they are associated with (such as play differences, etc).

I know you have to have a pretty rigid experiment to make causality statements, although I'm certainly no expert in this area. My sense is that simply adjusting stats because of some inherent difference between scorekeepers has a lot of assumptions tied do it, such as the assumption that scorekeepers are the only reason why the difference exists. I don't think it makes sense that we make this assumption without digging a lot deeper.
_________________
I am a basketball geek.

[Mike G]

Check out the first few years of Mutombo's career. Accompanying what seem to be years of high home bias (for blocks), his Blk% jumps (later dips) enormously. After correcting for this, his Blk% follows a smooth trajectory.

Both he and this year's Chris Andersen example were in Denver. Maybe someone knows: Did either of these players, for a year, become 60-90% more prolific shotblockers just in home games?

It's reasonable to suppose there's an interaction with the refs as well -- In Denver, it's a block; on the road, it's a foul -- some of the time.
_________________
`
36% of all statistics are wrong

[DJE09]

Ryan,
We regularly break things down based on Home / Road analysis. You seem comfortable with the idea that the statistics provide evidence that players "play" better at Home.
But, a reasonable alternative hypothesis is that there exists bias in the behaviour of score keepers / referees in favour of the home players. We do not have to suppose that this is intentional, merely that it occurs.

I would be extremely comfortable with the idea that LeBron, Paul Pierce, DHow, etc get special treatment when they are on 5 fouls - if there is another player the foul can get called on, then that will probably occur, or even no foul get called. I have seen too many assists attributed when the receiving player takes 1 or 2 dribbles and lays up a contested shot (ie defender there) to think that all assists are "real".

As a rational man, especially when provided with evidence (anecdotal or otherwise), it is reasonable to explore the extant data for what we can deduce. One might expect scorer bias to occur in those box-score stats that they attribute (assists/blocks), as opposed to the stats that the referees attribute (scoring / fouls). An investigation into the data looking for this is warrented.

I am not sure Mike was asserting that his Home/Away ratios reflected the actual degree of scorer bias, but more as an indication of it's presence. I am also wondering why this sort of adjustment is not acceptable, but Adjusted +/- is. Especially since Adjusted +/- doesn't seem to adjust for fact that a team is at home or on the road. (WHY????)

Lastly, Mike, I wonder if you are wanting to engage Dean on this in a theoretical sense if Denver was the right choice of team to pick on

[HoopStudies]

[Ryan J. Parker]