Sports Statistics

I am trying to come up with a new way to measure how good any given basketball player is. My name is Tommy, so I'll call it the Tommy Index. My thinking is that the best players are the ones who give their team the best chance of winning the game. The team that scores more points wins the game 100% of the time, so each point that a player scores as an average would count as 1.0 (1x100%) toward the Tommy Index. I'm not sure what the statistics are on how often a team with more rebounds wins the game, but let's say the team that has more rebounds wins the game 70% of the time. If that is the case, a player's rebounds would count as 0.40 (1x40%) toward the Tommy Index. The multiplier is 40% because 70% is 40% of the way from 50% to 100%. Harmful statistics could be subtracted from the Tommy Index. Let's say the team with fewer turnovers wins the game 80% of the time. This means that the team with more turnovers wins 20% of the time. So, turnovers would count as -.60 (1x60%). Here, 60% was the multiplier because 20% is 60% of the way there from 50% to 0%.

Does anyone have thoughts on this. And does anyone know how you'd actually calculate something like this. The key is getting statistics on how often the team with more rebounds, etc. wins the game.

They already have statistics that speak to players value. Things like WAR and even simple +/-. I would stick with those, Tommy. Also, totally unrelated but where are you from? Please tell me it's the east coast. Please tell me you're Italian.

Also, basketball teams tend to run the same rotations. So you could just cap the team. Your system would be a lot of work for a small advantage. Basically it would benefit in times where a player is injured or sitting out a game. And in such cases oddsmaker cook such things into the line anyway. It would be more beneficial for bench players and betting quarters but you wouks have to really study a team and know their rotation. Statistics like WAR (sometimes WARP) are more valuable in baseball where lineups change daily.

Yeah, what STU said it was came to mind w me as well... Namely , the +/-  statistic.  

That tells you pretty clearly how well a player is working both ends of the floor, and their "value."  

Yes, from the east coast, but not Italian.

I had this idea just as a general way to measure who the best player is.  Wasn't intending to use it for sports betting or fantasy.  As a measure of who the best players are, I think +/- has a lot of limitations.  If Lebron plays all 48 minutes, but his team loses the game, he will have a negative +/- for that game (even if he had a great game).  +/- seems to be more useful to measure players that play limited minutes or come off the bench, but even then, you run into problems.  If I tend to get subbed into the game with a great player, I will have a better +/- even if I suck.  Similarly, if I tend to get subbed in the game with bad players, I might have a bad +/- even if I'm pretty good.

Food for thought. When Dennis Rodman was with the Bulls someone did a study and found that with all his rebounds the Bulls were able to maintain possession of the ball on average FIVE minutes more PER GAME. I wished I had saved the article, I honestly can't remember if they gave a breakdown on scoring, There has to be a web page out there that can give you this information. Fan graphs might be worth a look or Basketball Reference? Good luck! 

Not sure how detailed you want or if you are looking for something to get you started.  

But if you are interested I can send you a couple of links to get you started.  

For example, I can post a link on Rodman’s rebounding, like the guy mentioned.  It analytic in great detail.  But the link will lead you to many others like it.  Very good if you are a numbers guy and like stats, etc.  Maybe you can use that to extrapolate it to get what you are interested in. 

I think the part I'm looking for help with is that I can't figure out how to compile stats to figure out how often the team with more of stat x wins the game.  For example, the box score includes how many rebounds each team had.  The box score also includes how many points each team had.  Is there a way to compile the box scores from every game this season to figure out how often the team with more rebounds won the game?

Yes they are compiled for you.  You will just have to search for exactly what you want and extract it.  

Here are some links and notes that may lead you to look around some:

https://hoopsu.com/the-truth-behind-the-boards-and-winning-the-rebounding-war/

https://skepticalsports.com/the-case-for-dennis-rodman-part-14-a-rodman-v-jordan/

Even at NCAA level:

1. The team that recorded the high field goal percentage was the winner in 75 percent of the basketball games.
2. The team that scored the most free throws was the eventual winner in 70 percent of the games.
3. The team that recovered the higher number of rebounds was the winner 65 percent of the time.
4. The team that committed the higher number of personal fouls was the eventual winner in oniy 25 percent of the games.
5. The team that held the point advantage at halftime was the winner 74 percent of the time.

With respect to predictive power of the model, field goal conversion percentage explained 61.4% of the variance in winning percentages, whereas average three-point conversion percentage of the opposing teams explained 18.9%. These two variables combined to explain 80.3% of the total variance in winning percentage (adjusted R2 = 78.8%). In the study of human behavior, this percentage is extremely large, suggesting that an NBA team’s success can be predicted with an excellent degree of accuracy.

That's really helpful.  Thanks

It’s a start.  Depending on how detailed you want — there are many others.  Analytics are very big in sports now.