Are Metric Stats useful?
Usually when talking about stats, you mention points first, then rebounds or assists, then percentages, blocks, steals, and turnovers. And these are the stats that ESPN and Yahoo displays whenever you want to look at stats. However there is another measure of stats, not used very commonly in Basketball, more of a Baseball thing, called Sabermetrics. It goes beyond the basic percentages and actual raw stats. It looks at things like what percentage of a team's possesions a player uses, or what percentage of availible rebunds a player grabs, or even more specifically offensive rebounding rate. True Shooting percentage is another metric stat. It resembles Field Goal Percentage, but takes Three point and Free Throw shooting into consideration. It's basically an overall measure of how efficently a player scores. Block percentage tells you the percentage of two point field goals a player blocks while that player is on the court. So that means that of every 11 two point field goals attempted, Serge Ibaka(The league leader by almost two percent) will have 1 block. At first it doesn't seem like much but when you really think about it it's quite a bit. Then there is turnover percentage. This represents the amount of turnovers per 100 percent. Just to give you a little perspective, Al Jefferson(the league leader) averages 6 turnovers per 100 possesions, and John Wall averages 25 turnovers per 100 possesions. Steal percentage si the amount of opponent possesions that end with that player earning a steal while he is on the floor. And I assume we are all aware of PER, a measure of per minutes production.
The last couple are offensive rating, and defensive rating. Offensive rating is a mesure of how many points a player scores per 100 possesions, and defensive rating is a measure of how many points a player gives up per 100 possesions. And Offensive Win shares are the amounte a wins contributed on offense, and same for defense.
Now keep in mind that these are all estimates, and that some of these stats pose major flaws and apparently do not take several key factors in to consideration.
Yesterday I was explaining to I believe it was Tuck, that these Sabermetric stats prove that Amare is not nearly as terrible a defender as we assume him to be.
So let's look at Amare's defensive rating year by year, from his rookie season to this season, skipping over the one season where he only played 3 games
Now these numbers obviously mean nothing to you guys, we have no prerequisite, so let's just analyze these number within theselves, we can safely say that there was a three year period wher that number excalated, his last 2 in Phoenix, first in NY, and now let's look at a poven post defender's numbers, Kendrick Perkins.
Now obviously he was still developing those first couple years, and that 97 sticks out, and that was KG's career best year as well, but as you can, baring those three horendous years, his numbers don't look too bad compared to a proven post defender.
Now if you look at Duncan and KG's numbers, they have several years in the 90s, and you'd be surprised how good Duncan's numbers look. Take a look at Shaq's numbers too, wasn't known as the greatest defender but made some All Defensive teams, and clogged the middle.
Now these numbers look flawed, and from the good old eye test, I myself sort of doubt these numbers, and the people that created them also would agree that there are glaring deficicies that these statistics don't take into consideration.
With that being said, it can be somewhat usefull. Looking at rebounding percentages can tell you who really hits the boards hard because raw numbers can also be decieveing, win shares to me seem like a trustworthy statistic. However True shooting percentagedoesn't seem to be the best measure of who the best shooters are, when you look at the league leaders you see a bunch of big guys and Steve Nash, lol.
Not sure whether these numbers will be used more in the future, but I can garuntee you that organizations like the Thunder, Mavs, and Spurs have guys that crunch numbers to give them an estimate of what they have or need, and who does what and how well, and I've sure some of these are factored into teh equation.
Take a look at the league leaders in these catagories, you have to scroll down, it's pretty interesting, even though it may not be the best measure, it must have some purspose of use otherwise it wouldn't have been created or put anywhere online.
Just some fun facts, Melo and Amare each have higher defensive win shares than offensive win shares this season. Some guys like Ron Artest and Derek Fisher have negative(or close to it for Fish) offensive win shares, usually guys never have negative defensive win shares, but Steve Nash did his last year in Dallas, and as you know, Nick Collison is among the league leaders in Plus/Minus once again.
And just curious, what do you guys think of these stats, and do you deem them useful.
Sorry, can't post the link, but you can find it under league leaders on Basketball Reference.
I have friends who've run logistical regressions on the validity of some of these stats as part of his phd in management science to predict future performance and sometimes they're spot on.
Teams are using these kind of stats (synergy sports tech is the leader in this field to my knowledge) all the time now. analytics is HUGE in basketball as teams are trying to one-up each other.
For example, I just got from synergy sports a chart of Anthony Davis's Blocked shots tendencies (LH vs. RH, help vs. primary defender). The detail teams are going into is kinda insane, but you gottta love it.
I'm so confused. I'll just watch games and figure out if they are a good defender from that.
as a math nerd. They have the most use to me in fantasy bball. I would never take Tony Allen over DWade as an example, but they are more useful in the later rounds.
Ultimately, any individual ranking of basketball players is flawed because basketball is a team sport.
They're excellent information and since no one can watch every game of every team, they are a great guard against sample size and recency bias. They aren't the whole picture but are a big part.