reply on 82data
- i agree that the 82games data is to be appreciated and it is very
life in its life. i agree with your statement that support and
suggestions should be the main response from here. i merely was
hoping that more information about sources could fully lay to rest
any questions about them (since i've heard this natural internet
skepticism at this group and in other message boards) and would
actually help the site by increasing confidence and letting use of
site data move forward on a firm foundation. but i meant no offense
and i assume that the data is responsibly offered. the fact that
several people have seen the minute by minute detail relieves my
initial wonder if some of the data might have been based on player
data from sample games or otherwise imputed.
- I'm sure you've already read this, but let me reiterate that the
most appropriate way to validate 82games.com's data, as with
anything else, is to check it against an independent source. Plenty
of this data is available elsewhere, and if I had the time or really
cared about the issue, I'd check it out.
Harvey Pollack's Statistical Yearbook -- at least this year's
edition -- might be a good place to start. Since we know Roland
doesn't calculate +/- quite the way we think of it, there will
certainly be discrepancies there, but we should found similar things
in terms of shot distribution and other such metrics. I'm sure I'm
not the only with Harvey's book.
Here's a link I've posted before where a guy I know wrote a program
to parse data from NBA.com's play-by-plays:
Unfortunately, it's not complete for last season, but I know Ansu
Sesay didn't play any games after it was last updated, so his
numbers could easily be compared to what 82games.com has -- this
tracks many of the same categories, like shooting percentage by
distance, assisted field goals, etc.
The numbers to check Roland's work are out there ... if you're
willing to put in the work, start there. *shurg*