2015-16 Team win projections

Home for all your discussion of basketball statistical analysis.
Kevin Pelton
Site Admin
Posts: 102
Joined: Thu Apr 14, 2011 10:05 pm

Re: 2015-16 Team win projections

Post by Kevin Pelton » Sun Feb 28, 2016 7:10 pm

tarrazu wrote:Yes the vig. If your model projects 56 wins for the Krazy Koalas and the Vegas line is 55.5 (over -130), there is no bet to make. What is the threshold for difference from the Vegas line for your model to suggest a bet on the over/under?
Probably worth noting that one of the things that affects the vig is how much statistical models tend to differ from the opening lines. Wasn't Boston over one of the most expensive totals this year?

kmedved
Posts: 82
Joined: Thu Jul 03, 2014 9:18 pm

Re: 2015-16 Team win projections

Post by kmedved » Wed Mar 02, 2016 3:20 pm

tarrazu wrote:Yes the vig. If your model projects 56 wins for the Krazy Koalas and the Vegas line is 55.5 (over -130), there is no bet to make. What is the threshold for difference from the Vegas line for your model to suggest a bet on the over/under?
Issues like the vig can be dealt with somewhat easily with by the Kelly criterion. However, you can also "convert" between a -130 line and a -110 line if you want, since the errors are mostly normally distributed.
Kevin Pelton wrote:Probably worth noting that one of the things that affects the vig is how much statistical models tend to differ from the opening lines. Wasn't Boston over one of the most expensive totals this year?
Yeah, Boston was around -130 by the season opener. That said, I saw Boston open at 42.5 (-110), so it looks to be mostly a case of the bet intake driving the vig, rather than the difference between the models and the lines. Boston was projected to be an upper 40s win team by most RPM-style models from pretty much the moment they signed Amir Johnson, which was well before Vegas books started taking bets on this. The models didn't change between the bets opening, and the season opening, so it looks like bet volume was what drove the change.

tarrazu
Posts: 72
Joined: Mon Aug 04, 2014 5:02 pm

Re: 2015-16 Team win projections

Post by tarrazu » Wed Mar 02, 2016 5:50 pm

kmedved wrote: Issues like the vig can be dealt with somewhat easily with by the Kelly criterion. However, you can also "convert" between a -130 line and a -110 line if you want, since the errors are mostly normally distributed.
Yes, you can convert the vig. I was just creating a hypothetical explaining that having a higher or lower number then the "vegas line" would not necessarily result in an "over or under bet" with a positive expectation.
kmedved wrote:Yeah, Boston was around -130 by the season opener. That said, I saw Boston open at 42.5 (-110), so it looks to be mostly a case of the bet intake driving the vig, rather than the difference between the models and the lines. Boston was projected to be an upper 40s win team by most RPM-style models from pretty much the moment they signed Amir Johnson, which was well before Vegas books started taking bets on this. The models didn't change between the bets opening, and the season opening, so it looks like bet volume was what drove the change.
'The models' have no bearing on the vegas line. If whoever is setting the lines have similar methodologies, then sure, you could say 'the models' have an influence. Ultimately money is moving the lines. If sportsbooks keep taking money on BOS (say at 42.5 -110) to go over their win total (which could certainly be the result of many smart people projecting BOS much higher then the vegas projection), then they will adjust it upward; first by moving the vig (outside of a new signing or trade that could significantly alter their projection) and eventually to 43 or 43.5.

caliban
Posts: 29
Joined: Mon Oct 26, 2015 8:48 am
Contact:

Re: 2015-16 Team win projections

Post by caliban » Mon Mar 07, 2016 12:14 pm

Had my first kid hence the lack of updates lately.

3 entries running away right now. 5 entries doing better then the aggregated avg projection.

Image

Mike G
Posts: 4429
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2015-16 Team win projections

Post by Mike G » Mon Mar 07, 2016 12:51 pm

Relative to b-r.com's forecast page, about 76% of the season.

Code: Select all

average errors
AJ   5.00      DSM  5.60      yoop 6.62
KF   5.01      Crow 5.82      itca 6.65
km   5.08      rsm  5.82      DrP  6.91
bbs  5.21      fpli 6.02      nr   6.93
DF   5.25      MG   6.37      Dan  6.97
Cal  5.30      snd  6.44      EZ   7.37
tzu  5.34      BD   6.52      taco 7.85

RMSE
Cal  6.20      rsm  6.96      BD   7.94
AJ   6.22      tzu  7.04      itca 8.38
km   6.23      Crow 7.28      nr   8.45
DF   6.47      fpli 7.30      Dan  8.51
bbs  6.62      yoop 7.74      EZ   8.58
KF   6.77      MG   7.80      DrP  9.00
DSM  6.93      snd  7.88      taco 9.29
Cal: Congrats! Name him Steph?

caliban
Posts: 29
Joined: Mon Oct 26, 2015 8:48 am
Contact:

Re: 2015-16 Team win projections

Post by caliban » Mon Mar 07, 2016 2:37 pm

Thanks! Lulz, I think the closest I would get with a gender and language transformation is Steffi tho quite uncommon. Then off course I'd run into the predicament of Steffi Graf who casts Steph Curry in quite a large shadow to begin with.

ampersand5
Posts: 262
Joined: Sun Nov 23, 2014 6:18 pm

Re: 2015-16 Team win projections

Post by ampersand5 » Mon Mar 07, 2016 3:16 pm

Mazel Tov!!!

tarrazu
Posts: 72
Joined: Mon Aug 04, 2014 5:02 pm

Re: 2015-16 Team win projections

Post by tarrazu » Mon Mar 07, 2016 7:40 pm

J.E. wrote:I seems (manually) bumping down the ratings of players with "off-court issues" would have been a good idea this season, and it may be beneficial to do so in the future, as well

Specifically, I'm talking about Lawson (multiple DUIs), Harden (Kardashian) and Markieff Morris (assault charges, twin brother traded)
HSAC post on the "Kardashian effect":
http://harvardsportsanalysis.org/2016/0 ... rformance/

permaximum
Posts: 413
Joined: Tue Nov 27, 2012 7:04 pm

Re: 2015-16 Team win projections

Post by permaximum » Tue Mar 08, 2016 12:59 am

This is ridiculous.

Poeple don't even factor momentum, players' effect on the game of teammates regardless of his own "statistically" captured play (MG called it Iverson effect), rule changes, ball physics changes, basket material changes, parquet changes, style of play changes in different eras, referees, rare but certain match fixes, home-court bias on stats such as assists, blocks, steals, social-media mood, secretly betting friends, any kind of relationship with staff, teammates, opponents, dates with non-popular women, unusual incidents in daily life, arena temperature, humidty, altitude, noise, inactive microorganisms in the body, hypertension, hypotension, causeless mental problems, characteristic resistance to any kind of change, visual distractions in the arena crowd etc. But we're talking about Kardashian effect.

If you want to play Laplace's demon I suggest starting with momentum. Kardashian effect is way down on the list.

Lately, my tone got harsher because I believe instead of looking at the bigger picture, we're trying to answer why this one pixel at the right bottom corner has gotten paler recently. TBH, if you follow that path, you'll find the cause you wanted to find in the first place.

kmedved
Posts: 82
Joined: Thu Jul 03, 2014 9:18 pm

Re: 2015-16 Team win projections

Post by kmedved » Wed Mar 09, 2016 4:01 pm

Image

permaximum
Posts: 413
Joined: Tue Nov 27, 2012 7:04 pm

Re: 2015-16 Team win projections

Post by permaximum » Thu Mar 10, 2016 12:35 pm

Player tracking plus minus is on the run for 2/2 straight after it's introduction. I'm liking it but I wish projected RMSE was lower than 5.

It's also interesting because there have been doubts about SportsVu data's accuracy. However, there are also two more heavily weighed prediction compontents responsible for it's score. Minute distribution and RAPM itself. I guess I want to believe in player tracking data since it offers something new after what - 5-6 years? All those different SPM and RAPM versions in the last years offered marginal improvements.

Mike G
Posts: 4429
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2015-16 Team win projections

Post by Mike G » Thu Mar 10, 2016 4:20 pm

Some years, everyone does better; other years, everyone does worse. This is due to the totally unpredictable factors: injuries, team chemistry, players who just boom or bust unexpectedly.

In a good year for predictors, there are relatively few teams that have these unpredictable events. In a bad year, a bunch of teams are struck.
This is true of simple to complex analytical systems; Vegas with their scouting and insider information; and any blend of resources you may have.

I've been consistently around the 30 to 50 percentile, year in and year out, at this forum. I don't look at any insider stuff, no plus-minus variant, and no "gut feeling" or eyeball input.
Among strictly boxscore inputters, I may be doing as well as anyone.

kmedved
Posts: 82
Joined: Thu Jul 03, 2014 9:18 pm

Re: 2015-16 Team win projections

Post by kmedved » Thu Mar 10, 2016 8:48 pm

Apart from rooting for myself (of course), I do think it would be good for PT-PM to take this down, simply because a new and differentiated source of data is always good. It doesn't undermine anything about RAPM - if anything, it's something that could be used in concert with RAPM.

Mike G
Posts: 4429
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2015-16 Team win projections

Post by Mike G » Fri Mar 11, 2016 9:26 pm

A couple of weeks ago, I endeavored to make a blend of eWins, PER, WS, BPM, and RPM which best retrodicts this year's team success (SRS) based on last year's player rates, in this year's actual minutes.

With no regression to the mean, and no age factor, I get an avg 82-game error of about 5.5
This isn't quite as good as the leaders (5.0-ish) in the competition; it's better than my submitted prediction.

The recipe is 41% based on eWins, the remainder from the other 4 'available' stats.
When the season's over, I may play with regressions to optimize the recipe.

permaximum
Posts: 413
Joined: Tue Nov 27, 2012 7:04 pm

Re: 2015-16 Team win projections

Post by permaximum » Sun Mar 13, 2016 12:13 pm

Actual minute advantage is a big advantage. Just saying.

Post Reply