Building an NBA Prop Model — a lockdown project — part 3
Slowness
Firstly thanks to the 22 hardy souls that read my 2nd outpouring of nonsense. As expected interest declined, so I’ve contacted the Pulitzer committee to remove my name from the running. If you have stuck around, cheers I hope it’s interesting — I’m loving it, so will carry on regardless.
A bit of a lag getting this one out. Like my pal Kawhi above I’ve been taking a bit of time for load management and staying away from the computer out of working hours. Going to kick back on again
I’ve also been addicted to Shattered: Hope, Heartbreak and the New York Knicks on The Athletic — it’s riveting and who better to narrate than Chuck D — check it out here…https://theathletic.com/podcast/246-shattered-hope-heartbreak-and-the-new-york-knicks/?episode=1
It’s also on Spofity, but there is an uncomfortable amount of Erectile Dysfunction adverts attached (they aren’t targeted are they?).
I have 5 30-day guest passes for The Athletic I haven’t used yet, happy to share if you want to give me a shout. It’s easily the best sporting journalism resource out there.
Plan today is to kick on through the rest of the ‘information I will capture’
Information — Rest
This is one I am convinced there is value in. In the modern NBA rest is a major issue — one that the players push for, but drives fans mad.
This isn’t news — there is plenty of resource out there focused on this edge
When you combine that with the increase in injuries this season you are getting marquee match-ups that are anything but, the issue is impacting the proposition especially on nationally televised matches.
Imagine taking your 2 kids to see Lakers @ Warriors — Curry, Green, James, Davies and Caruso — cost for tickets alone is ~$750 to be confronted by 40 mins of Kentavious Caldwell-Pope and Kelly Oubre Jr doing his Sandy Lyle impression — You’d be rightfully pissed off
Jordan missed some time in his 2nd season to injury; he then played a near consecutive 8 year stretch before heading off to play rounders. Very few players play 82 game seasons now — player power has moved forward.
Anyway, back to the reason I am looking at this
I believe the amount of rest a player gets between games will impact his productivity; especially in back-to-back matches
Steph Curry is my main subject here and, of course, tells me I’m an idiot. In fact he is more productive in back-to-back matches!
Taking a peak at Lakers players I have captured so far — LeBron doesn’t care how much rest he’s had and AD does show an impact on PTS, RBD and AST
Looking at other players I have captured this is a mixed bag, less impact than I had imagined overall though (below is performance in back-to-back vs. 2 day rest — all have no change to mins played)
KD — PTS down 0.8, RBD down 0.3, AST down 0.4
Kyrie — negligible change
Harden — PTS down 1.6, RBD down 0.2, AST down 0.5
Jimmy Buckets — PTS down 1.6, RBD down 0.6, AST down 0.2
Embiid — PTS up 2.5, RBD up 1.7, AST up 0.4
Finally Mr. Load Management himself — Kawhi Leonard
This looks really bad for Klaw, but should carry a caveat as he almost never plays back-to-backs now and his productivity has monumentally increased over the years — so this is more a representation of old v new Kawhi
Conclusion: Overall rest has a minor impact to player performance, unless it’s Joel Embiid in which case jump on his overs in back-to-back matches!
The real impact of rest is to team performance, something I will need to try and capture if I want to back Handicap/Moneyline/Points
Information — Result
Not going to spend too much time here as the data can get messy. What I need is a ‘blow-out indicator’ — so rather than how a player performs in losses, it’s how many mins does he play in blow-outs
I don’t have this and I don’t think it comes from historical data. I will need to build something in that looks at the current Handicap in the match to calculate expected mins…when I get there!
From the data I have looked at there is a major difference in productivity from players in Wins vs. Losses — so I will include a weighting based on how they perform in those, but need to understand if they are predicted to Win/Lose
Conclusion: This does impact, but need to capture this a different way
Information — Season
Regular Season vs. Playoff Performance
Is Playoff Rondo a myth — is Paul ‘Playoff P’ George more delusional than we think…well in a nutshell ‘Kinda, yeah’
Almost every player I looked at has increased overall productivity in Playoffs vs. Regular — however, I believe this is largely down to one thing…minutes
Below is difference in average mins per match in Playoff vs. Regular
Steph +4 mins
Draymond +8mins
LeBron+4 mins
AD +4 mins
Outside of that there are factors that work against each other. Players are more focused and engaged BUT they are playing better teams that are always trying and both teams have more opportunity to study the opponent — attackers know more about the defender, and defender about attacker.
Conclusion: I will put some weighting on this during Playoff matches, but the increased in expected minutes will be the key
Information — Recent Form
This is the one, the absolute key — we are lucky to be watching greatness right now with Steph Curry on a run that transcends any ‘MVP Race’ and franks him amongst the pantheon of the best to ever do it
The longer a player plays, the less his career statistics matter. Kawhi shows this better than anyone
So we need to know how players are performing in last few matches. I am going to do this in increments to capture immediate form, weekly-ish, fortnightly-ish and monthly-ish — so 1,3,5,10 and 20 games and attribute weighting to that (more weight on consistent form to remove the played out of skin moments)
Conclusion: This is what I really need to concentrate on. I could do this at the expense of all the other data captured but there is some use in it and I also like the pretty bar charts and graphs
My main conclusion from all of this is Form and Minute Projections are the absolute key to this
Next: I’ll need to look at more than Steph Curry so going to grab all of Golden State Warriors and try to put something half-decent together
