December 31st - New Orleans Pelicans @ Oklahoma City Thunder - 2-2

Quote:

---

Originally Posted by Pelicanidae

https://basketball.realgm.com/player.../International

This is what I'm working off. 28 games played for the Sydney Kings between 9/20/2019 and 3/15/2020, in which he shot overall 52/140 from 3 for a total of 37.1%.

The two outlier games I'm talking about were on September 22nd of 2019 against Perth, were he went 8/11 from 3, and again also VS Perth on October 11th when he went for 7/11 from 3. Remove those two games from the 52/140 total, you get 37/118 as his total for the other 26 games, which is 31.3%. Which is pretty bad.

---

I think removing the best games in a small sample size from ANY shooter will make their numbers worse. I don't really get the point of this. Shooters have streaks where they make a bunch in a row and miss a bunch in a row and over enough time it evens out. It's fair to point out he wasn't a particularly dynamic player in the NBL (though he wasn't an unmitigated disaster, like RJ Hampton was), but the numbers are what the numbers are and I don't think removing games where he shot especially well. Unless you have some data that a median shot percentage per game better projects what a shooter's percentage will be than their actual percentage.

I watched a few games (I actually like the NBL) and I thought he had moments where it looked good and others where he was kind of invisible. I think he did a nice job as an on-ball defender and has some potential in that role.

I was very upset again watching the Blazers-Warriors game. There was a very similar play to the Ingram-Roby play from last night where Looney accidently clocked Derrick Jones Jr. in the head on a fast break going for a dunk. Was called a flagrant 1, no ejection, and unlike BI Looney just connected with the guy's head, no ball. Just an absurd call.

Quote:

---

Originally Posted by Biasvasospasm

I think removing the best games in a small sample size from ANY shooter will make their numbers worse. I don't really get the point of this. Shooters have streaks where they make a bunch in a row and miss a bunch in a row and over enough time it evens out. It's fair to point out he wasn't a particularly dynamic player in the NBL (though he wasn't an unmitigated disaster, like RJ Hampton was), but the numbers are what the numbers are and I don't think removing games where he shot especially well. Unless you have some data that a median shot percentage per game better projects what a shooter's percentage will be than their actual percentage.

I watched a few games (I actually like the NBL) and I thought he had moments where it looked good and others where he was kind of invisible. I think he did a nice job as an on-ball defender and has some potential in that role.

---

Sure, but the question is by how much? If you those Didi numbers, you see that removing the two best games turns him from a solidly above average shooter into someone absolutely dreadful.

By comparison, if you take someone like Luke Kennard, who is a very good NBA shooter but not one of those top 5 or top 10 guys like Redick or Curry, and someone who also only played 28 games last year due to injury, you can try the same experiment. He was a 39.9% shooter overall on 6.5 a game, for a total of 73/183 over his 28 games played. If you take his best five shooting games by efficiency from that 28 game span and remove them, 73/183 becomes 55/156, which is 35.2%, which is only a touch under league average. And that's removing the five best games, rather than the two best: by comparison, DiDi's move from solid 37% shooter to dismal 31% shooter by removing only 2 games is a sign that his average shooting performance isn't quite as good as the raw 3pt percentage may indicate. Obviously you are free to disagree with that, and think that players best games indicate their ability, which is fine.

Quote:

---

Originally Posted by Pelicanidae

Obviously you are free to disagree with that, and think that players best games indicate their ability, which is fine.

---

Until someone produces some statistical analysis that removing unusually good or bad shooting performances from prospects creates a more predictive statistic, I'm going to go with a larger sample size being more predictive than a smaller sample size.

dae is absolutely correct. Researching data for 'outliers' is imperative for accurate statistical analysis. It happens all the time....

In the very first game Zion played for the Pelicans, he was four for four (4 of 4) from three point range. For the other 23 games he played he made 2 of 10. So his 3 Point Percentage for 2019-20 was 42.9% I think we can all agree that Zion is nowhere near a 43% shooter from three (if so, stick him out there); so his first game was an 'outlier' in terms of three point shooting and skews the truth about Zion's three point prowess.

Please, I'm not looking for an argument, I'm just stating empirical truth.

Quote:

---

Originally Posted by Biasvasospasm

Until someone produces some statistical analysis that removing unusually good or bad shooting performances from prospects creates a more predictive statistic, I'm going to go with a larger sample size being more predictive than a smaller sample size.

---

There is actually such a thing as luck adjusted shooting regression, but I don't think it's publicly available for the NBL :hihi:

Quote:

---

Originally Posted by As I See It

dae is absolutely correct. Researching data for 'outliers' is imperative for accurate statistical analysis.

Please, I'm not looking for an argument, I'm just stating empirical truth.

---

Thanks for the condescension and the "empirical truth." Man, I need to get off this board.

Quote:

---

Originally Posted by Biasvasospasm

Thanks for the condescension and the "empirical truth." Man, I need to get off this board.

---

You're welcome, sir.

If you throw out the good shooting games, then you should also throw out the bad shooting games.

Quote:

---

Originally Posted by donato

If you throw out the good shooting games, then you should also throw out the bad shooting games.

---

I agree with this. If you throw out the 2 best you should also be throwing out the 2 worst.

Quote:

---

Originally Posted by donato

If you throw out the good shooting games, then you should also throw out the bad shooting games.

---

Assuming there exist a variance as they relate to the rest of the sample...AGREED!!!

Suppose Steph Curry is averaging 30 for the five games (35, 24, 18, 42, 31) of the season. Then in game six he scores zero (0) points in the first three minutes before twisting an ankle (or receiving a double technical, or just taking ill) and sitting out the remainder of the game.

1. Will his Zero Point game drastically effect his per game average after game six?
2. Is the Zero Point game indicative of a typical Steph Curry game?
3. Would you value his sixth game point total in evaluating Steph Curry's scoring prowess?

Therefore the sixth game should be treated as an 'outlier'.

Now, should we throw out the 42 point game because we threw out the Zero (0) point game?

Quote:

---

Originally Posted by P_B_&_G

I agree with this. If you throw out the 2 best you should also be throwing out the 2 worst.

---

I would agree, if those games were outliers. If he had 2 games that were extremely high efficiency and then 2 outlier bad games (where he had shot 0/10 or something) then yeah, you would discount those. However, Didi's less successful shooting nights were his norm; there aren't two notable outlier nights that were just miles worse than the rest. Choosing which ones to discard would therefore be impossible, because you'd just be chucking out two average nights for no real reason.

This has gotten kinda ridiculous given that I showed his regular season numbers which are more representative.

First of all, statistically speaking, you would absolutely NOT include his preseason stats. Preseason and regular season, from a quality standpoint, represent two different "populations." Without getting too far in the weeds, you don't mix those for extremely obvious reasons. Sample size is no excuse. If his regular season sample is too small, then dont run his stats. Just wait.

Secondly, if you simply look at his regular season game log, his stats bear out exactly how he played, which was far from dreadful in any category. It actually showed he had good streaks and bad streaks over far more than just 2 games. He had a combined 6 game stretch of great shooting, and other stretches where he didnt. All in all, about what you would expect from a very young player, in a new league, playing amongst grown men. Didi isn't some Luka or LeBron prospect.

For a 35th pick, what are you looking for him too do? Besides nit pick the hell out his shooting stats?

And as for shooting, we're just going to go back to the BI argument had 2 summers ago, which we know how that turned out. Didi has a very good shooting form. Not only is his form very good but with a quick release to get it off whenever he wants. The absolute LAST thing I'm worried about with Didi is his shooting ability.

He had several moments of looking very good and others where he disappeared. He was anything but "awful." Jesus Christ.

All in all I think the best way to judge his projected shooting is look at the overall season numbers and then look at his free throw percentage. Also, what does his stroke look like?

Throwing out games due to them being outliers is probably a bit of overanalysis.