Probability distributions for predicting basketball players' points scored?

Question

I have had some success creating predictive models using a fairly simple Poisson distribution model for individual players' rebound and assist numbers in basketball games. These figures match the requirements of the poisson distribution; namely, they are infrequent and occur 1 at a time.

I was working on trying to create similar models for points scored by a player in a basketball game. However, point scoring occurs differently: players can score a 2 point basket, a 3 point basket, and a 1 point free throw. If I had numbers for all of these different types of scoring, any suggestions on how to put them together and model a player's total scoring output?

I'm mainly using this for two different purposes:
1. Predicting how many points the individual player scores (ie. "Probability that Player X scores more than 15 points)
2. Predicting players' scores in relation to each other (ie. "Probability that Player X scores more than Player Y).

Thanks in advance!

Answer 1

Rather than attempting to model scoring by the sum of three independently varying random variables, have you considered simply modeling it as a single random variable and ignoring the underlying 1, 2, and 3-point contributions?

That would be much easier, and I suspect if you've gotten reasonable results on rebounds and assists, you will for total points as well.

Answer 2

Would you suppose that a normal distribution for total point output might work? I know normal distributions are not discrete, so that you might get predictions like 15.2874 as an answer to some question.

It would probably be easier not to worry about where the points come from (1,2, or 3 points), but just look at overall point production. There is an average and a stan dev, and I could easily see how from there it would have a normal-like distribution. Very high scores would be highly unlikely, though not impossible.

I suppose you would have to make some adjustments, since negative scores are not possible. There must be some model for normal-like distributions in which negative values are not possible, such a population distributions or something like that.

I hope this helps.

Answer 3

a million. Magic will win 60 video games. promote 2. Kevin Durant will time-honored 30 ppg this season. purchase 3. Tim Duncan will time-honored 20-10. promote 4. Kobe may have a 50-factor interest. purchase 5. Wade will lead the warm temperature in scoring. promote 6. Bulls will attain convention Finals. promote 7. Tyreke Evans will be seen for MVP. promote 8. DeMarcus Cousins will be Rookie of the three hundred and sixty 5 days. promote 9. Blake Griffin will time-honored a double double. purchase 10. Wizards will make the Playoffs. promote 11. A participant no individual expects to be an All-well known individual will be one this season. promote 12. Spurs will win 50 video games back. purchase thirteen. Kevin Durant will lead the league in scoring. purchase 14. Yao Ming will play all 80 2 video games. promote 15. Joakim Noah will be an All-well known individual. promote 16. warmth will beat Lakers by technique of a minimum of 20 at Christmas. promote 17. Kobe will time-honored below 25 ppg. promote 18. Orlando will make the Finals. promote 19. Rondo will strengthen a leap shot. promote 20. The 2 hundred+ will be lengthy surpassed by technique of the playoffs. promote .

Answer 4

Xscores Basketball

Answer 5

Is basically to get at activity betting with the Zcodes System that you will discover here https://tr.im/nqiEa .
Zcodes System is a program which will provides you with completely computerized sports choices based on a powerful forecast model. After that it bottles you the information each and every day of the season with easy instructions on the best way to position bets and constantly win.
Zcodes System covers all the key events in the sporting schedule when one sports time ends, yet another is getting started.
Zcodes System includes activities like: football, football, hockey, soccer, basketball, horse racing and a whole load more.

Answer 6

1

Answer 7

You have already solved your own problem (the way you analyzed and framed the problem already have the answer in it).

It is 3 poisson processes. 1 point process, 2 point process, 3 point process; each has its own lambda. (count by basket)

Total point is just the sum of 3 processes (1x 1pt basket + 2x 2pt basket + 3x 3pt basket). Good job, good application.

Answer 8

Sports betting systems are sets of events that when combined for a particular game for a particular sport represents a profitable betting scenario