Jayson Tatum

[CptCrunch]

Tatum projects as a lot better 3pt shooter than Warren to me

This. Tatum is going to be an excellent 3 point shooter in the NBA. Warren on the other hand will never be.

The spacing this creates is huge in today's game.

Are you guys basing this solely off of his FT%

That seems to be the new trend. Not saying it is wrong but to pencil Tatum in as a 3pt threat in the NBA seems a bit premature IMO and could wind up being his downfall

Ask yourself this: what happens to Tatum if he ISNT a good 3pt shooter?

You are straight up hating. Tatum is probably the most likely prospect to develop an elite 3pt shot among the those without a good or elite shot already. His mid range is elite. His free throw rate is beyond elite.

Statistical projection can't be perfect, but free throw % is by far the best stat we have for 3 pt shot projection: http://nyloncalculus.com/2015/05/25/projecting-draft-prospects-three-point-percentages/

I'm not worried about Tatum's shooting. My concern is if he lacks the defensive basketball IQ like Andrew Wiggins or the lateral quickness like Tobias Harris to be an plus defender.

[darealjuice]

This. Tatum is going to be an excellent 3 point shooter in the NBA. Warren on the other hand will never be.

The spacing this creates is huge in today's game.

Are you guys basing this solely off of his FT%

That seems to be the new trend. Not saying it is wrong but to pencil Tatum in as a 3pt threat in the NBA seems a bit premature IMO and could wind up being his downfall

Ask yourself this: what happens to Tatum if he ISNT a good 3pt shooter?

You are straight up hating. Tatum is probably the most likely prospect to develop an elite 3pt shot among the those without a good or elite shot already. His mid range is elite. His free throw rate is beyond elite.

Statistical projection can't be perfect, but free throw % is by far the best stat we have for 3 pt shot projection: http://nyloncalculus.com/2015/05/25/projecting-draft-prospects-three-point-percentages/

I don't think you understand your source. He's not saying that looking at FT% alone is better than looking at 3PT% alone, he's saying his model looking at FT%, 3PTA frequency, and 3PT% together is significantly better than looking at 3PT% alone. FT% and 3PT% on their own are nearly equally (in)effective at predicting NBA 3PT% when used alone with correlation coefficients of something around .44.

http://www.espn.com/nba/draft2015/insider/story/_/id/13003659/nba-draft-michael-frazier-dangelo-russell-top-shooters-draft

So if we take the full sample of players with at least 250 career 3-point and free throw attempts in the Sports-Reference.com NCAA database and at least 500 career 3s attempted in the NBA, college free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429). Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually, explaining about a quarter of the variation in what players shoot.

[Arsenal]

Tatum projects as a lot better 3pt shooter than Warren to me

This. Tatum is going to be an excellent 3 point shooter in the NBA. Warren on the other hand will never be.

The spacing this creates is huge in today's game.

Are you guys basing this solely off of his FT%

That seems to be the new trend. Not saying it is wrong but to pencil Tatum in as a 3pt threat in the NBA seems a bit premature IMO and could wind up being his downfall

Ask yourself this: what happens to Tatum if he ISNT a good 3pt shooter?

Look at his FT% and look at his shooting form. I'd bet A LOT of money that he'll be a good 3PT shooter. His FT% as an 18 year old freshman is insanely good.

[shawn_hemp]

This. Tatum is going to be an excellent 3 point shooter in the NBA. Warren on the other hand will never be.

The spacing this creates is huge in today's game.

Are you guys basing this solely off of his FT%

That seems to be the new trend. Not saying it is wrong but to pencil Tatum in as a 3pt threat in the NBA seems a bit premature IMO and could wind up being his downfall

Ask yourself this: what happens to Tatum if he ISNT a good 3pt shooter?

You are straight up hating. Tatum is probably the most likely prospect to develop an elite 3pt shot among the those without a good or elite shot already. His mid range is elite. His free throw rate is beyond elite.

Statistical projection can't be perfect, but free throw % is by far the best stat we have for 3 pt shot projection: http://nyloncalculus.com/2015/05/25/projecting-draft-prospects-three-point-percentages/

I'm not worried about Tatum's shooting. My concern is if he lacks the defensive basketball IQ like Andrew Wiggins or the lateral quickness like Tobias Harris to be an plus defender.

Feel like I've been pretty objective about Tatum throughout the year, seems like a pretty valid question...

If I was like "Tatum will never develop a 3pt shot because he sucks and Duke sucks" then yeah, I'd be hating

Ricky Rubio is a pretty good FT shooter fwiw...

[CptCrunch]

Are you guys basing this solely off of his FT%

That seems to be the new trend. Not saying it is wrong but to pencil Tatum in as a 3pt threat in the NBA seems a bit premature IMO and could wind up being his downfall

Ask yourself this: what happens to Tatum if he ISNT a good 3pt shooter?

You are straight up hating. Tatum is probably the most likely prospect to develop an elite 3pt shot among the those without a good or elite shot already. His mid range is elite. His free throw rate is beyond elite.

Statistical projection can't be perfect, but free throw % is by far the best stat we have for 3 pt shot projection: http://nyloncalculus.com/2015/05/25/projecting-draft-prospects-three-point-percentages/

I don't think you understand your source. He's not saying that looking at FT% alone is better than looking at 3PT% alone, he's saying his model looking at FT%, 3PTA frequency, and 3PT% together is significantly better than looking at 3PT% alone. FT% and 3PT% on their own are nearly equally (in)effective at predicting NBA 3PT% when used alone with correlation coefficients of something around .44.

R^2 is called coefficient of determination. It is a fairly garbage measure of model performance but for the purpose of this exercise, we'll entertain the idea that R^2 can be used to rank order model performance.

Of course 3PT % is the (presumably) best measure, but on a univariate level, FT% and 3PTA are the secondary predictors of future 3 pt shooting success. As a result, they are the secondary covariates thrown into a 3pt shot projection model. If you were to fit univariate models, they would come up at the top of the list.

Anyways, I'll scrape some data and work on a model or work on replicating some the publicly published results within the next few weeks.

[darealjuice]

You are straight up hating. Tatum is probably the most likely prospect to develop an elite 3pt shot among the those without a good or elite shot already. His mid range is elite. His free throw rate is beyond elite.

Statistical projection can't be perfect, but free throw % is by far the best stat we have for 3 pt shot projection: http://nyloncalculus.com/2015/05/25/projecting-draft-prospects-three-point-percentages/

I don't think you understand your source. He's not saying that looking at FT% alone is better than looking at 3PT% alone, he's saying his model looking at FT%, 3PTA frequency, and 3PT% together is significantly better than looking at 3PT% alone. FT% and 3PT% on their own are nearly equally (in)effective at predicting NBA 3PT% when used alone with correlation coefficients of something around .44.

R^2 is called coefficient of determination. It is a fairly garbage measure of model performance but for the purpose of this exercise, we'll entertain the idea that R^2 can be used to rank order model performance.

Of course 3PT % is the (presumably) best measure, but on a univariate level, FT% and 3PTA are the secondary predictors of future 3 pt shooting success. As a result, they are the secondary covariates thrown into a 3pt shot projection model. If you were to fit univariate models, they would come up at the top of the list.

http://www.espn.com/nba/draft2015/insider/story/_/id/13003659/nba-draft-michael-frazier-dangelo-russell-top-shooters-draft

So if we take the full sample of players with at least 250 career 3-point and free throw attempts in the Sports-Reference.com NCAA database and at least 500 career 3s attempted in the NBA, college free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429). Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually, explaining about a quarter of the variation in what players shoot.

You're right it's the correlation of determination not correlation coefficient, my mistake. Regardless, they're all not good predictors individually, and it doesn't matter which one is "best" when none of them are good lol that's why people build models to consider them all. I don't have a horse in this race and think Tatum should be a good shooter, but saying FT% is "by far the best stat we have for 3 point shot projection" is just wrong, and that's all my point is.

[CptCrunch]

I don't think you understand your source. He's not saying that looking at FT% alone is better than looking at 3PT% alone, he's saying his model looking at FT%, 3PTA frequency, and 3PT% together is significantly better than looking at 3PT% alone. FT% and 3PT% on their own are nearly equally (in)effective at predicting NBA 3PT% when used alone with correlation coefficients of something around .44.

R^2 is called coefficient of determination. It is a fairly garbage measure of model performance but for the purpose of this exercise, we'll entertain the idea that R^2 can be used to rank order model performance.

Of course 3PT % is the (presumably) best measure, but on a univariate level, FT% and 3PTA are the secondary predictors of future 3 pt shooting success. As a result, they are the secondary covariates thrown into a 3pt shot projection model. If you were to fit univariate models, they would come up at the top of the list.

http://www.espn.com/nba/draft2015/insider/story/_/id/13003659/nba-draft-michael-frazier-dangelo-russell-top-shooters-draft

So if we take the full sample of players with at least 250 career 3-point and free throw attempts in the Sports-Reference.com NCAA database and at least 500 career 3s attempted in the NBA, college free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429). Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually, explaining about a quarter of the variation in what players shoot.

You're right it's the correlation of determination not correlation coefficient, my mistake. Regardless, they're all not good predictors individually, and it doesn't matter which one is "best" when none of them are good lol that's why people build models to consider them all. I don't have a horse in this race and think Tatum should be a good shooter, but saying FT% is "by far the best stat we have for 3 point shot projection" is just wrong, and that's all my point is.

It is the best predictor among a sea of crappy ones. Most of them are crappy because there isn't a data series out there that can project NBA shooting success. That is literally what the article says: "College free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429)".

fyi, pearson correlation squared is equal to R^2 in univariate regressions, there is a trivial derivation here:
http://www.win-vector.com/blog/2011/11/correlation-and-r-squared/

There is no artificial constraint on model size in this case, but if you asked a consulting statistician to produce a univariate model to predict 3pt shooting success, a FT % model would be the model.

[darealjuice]

R^2 is called coefficient of determination. It is a fairly garbage measure of model performance but for the purpose of this exercise, we'll entertain the idea that R^2 can be used to rank order model performance.

Of course 3PT % is the (presumably) best measure, but on a univariate level, FT% and 3PTA are the secondary predictors of future 3 pt shooting success. As a result, they are the secondary covariates thrown into a 3pt shot projection model. If you were to fit univariate models, they would come up at the top of the list.

http://www.espn.com/nba/draft2015/insider/story/_/id/13003659/nba-draft-michael-frazier-dangelo-russell-top-shooters-draft

So if we take the full sample of players with at least 250 career 3-point and free throw attempts in the Sports-Reference.com NCAA database and at least 500 career 3s attempted in the NBA, college free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429). Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually, explaining about a quarter of the variation in what players shoot.

You're right it's the correlation of determination not correlation coefficient, my mistake. Regardless, they're all not good predictors individually, and it doesn't matter which one is "best" when none of them are good lol that's why people build models to consider them all. I don't have a horse in this race and think Tatum should be a good shooter, but saying FT% is "by far the best stat we have for 3 point shot projection" is just wrong, and that's all my point is.

It is the best predictor among a sea of crappy ones. Most of them are crappy because there isn't a data series out there that can project NBA shooting success. That is literally what the article says: "College free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429)".

fyi, pearson correlation squared is equal to R^2 in univariate regressions, there is a trivial derivation here:
http://www.win-vector.com/blog/2011/11/correlation-and-r-squared/

There is no artificial constraint on model size in this case, but if you asked a consulting statistician to produce a univariate model to predict 3pt shooting success, a FT % model would be the model.

Like I said, it doesn't matter which one is best when they're all crappy. You cut out the next sentence: "Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually."

No one would ask for a univariate model, so who cares if FT% what would be used when it's only slightly less bad than a 3PT% model? The point of a modeling data is to get the best correlation possible using the useful data you have available, not unnecessarily constrain yourself to a single variable. If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

[CptCrunch]

http://www.espn.com/nba/draft2015/insider/story/_/id/13003659/nba-draft-michael-frazier-dangelo-russell-top-shooters-draft



You're right it's the correlation of determination not correlation coefficient, my mistake. Regardless, they're all not good predictors individually, and it doesn't matter which one is "best" when none of them are good lol that's why people build models to consider them all. I don't have a horse in this race and think Tatum should be a good shooter, but saying FT% is "by far the best stat we have for 3 point shot projection" is just wrong, and that's all my point is.

It is the best predictor among a sea of crappy ones. Most of them are crappy because there isn't a data series out there that can project NBA shooting success. That is literally what the article says: "College free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429)".

fyi, pearson correlation squared is equal to R^2 in univariate regressions, there is a trivial derivation here:
http://www.win-vector.com/blog/2011/11/correlation-and-r-squared/

There is no artificial constraint on model size in this case, but if you asked a consulting statistician to produce a univariate model to predict 3pt shooting success, a FT % model would be the model.

Like I said, it doesn't matter which one is best when they're all crappy. You cut out the next sentence: "Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually."

No one would ask for a univariate model, so who cares if FT% what would be used when it's only slightly less bad than a 3PT% model? The point of a modeling data is to get the best correlation possible using the useful data you have available, not unnecessarily constrain yourself to a single variable. If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

> If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

Case closed, I was right. You initially quoted my disputing my claim, so there's that.

[darealjuice]

It is the best predictor among a sea of crappy ones. Most of them are crappy because there isn't a data series out there that can project NBA shooting success. That is literally what the article says: "College free throw percentage actually correlates slightly better (.443) to NBA 3-point percentage than college 3-point percentage (.429)".

fyi, pearson correlation squared is equal to R^2 in univariate regressions, there is a trivial derivation here:
http://www.win-vector.com/blog/2011/11/correlation-and-r-squared/

There is no artificial constraint on model size in this case, but if you asked a consulting statistician to produce a univariate model to predict 3pt shooting success, a FT % model would be the model.

Like I said, it doesn't matter which one is best when they're all crappy. You cut out the next sentence: "Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually."

No one would ask for a univariate model, so who cares if FT% what would be used when it's only slightly less bad than a 3PT% model? The point of a modeling data is to get the best correlation possible using the useful data you have available, not unnecessarily constrain yourself to a single variable. If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

> If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

Case closed, I was right. You initially quoted my disputing my claim, so there's that.

Uh you're not right lol. You said FT% is by far the best statistic for predicting NBA 3PT% when it's better by an insignificant margin and is equally useless as 3PT% alone. That's my whole point that you seem to keep whiffing on. You can't ignore bad 3 point shooting just because of good FT shooting just like you can't ignore bad FT shooting just because they have a good 3 point percentage.

[CptCrunch]

Like I said, it doesn't matter which one is best when they're all crappy. You cut out the next sentence: "Neither is a particularly good predictor of 3-point shooting in the NBA, but together they're stronger than either stat individually."

No one would ask for a univariate model, so who cares if FT% what would be used when it's only slightly less bad than a 3PT% model? The point of a modeling data is to get the best correlation possible using the useful data you have available, not unnecessarily constrain yourself to a single variable. If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

> If it makes you feel better than FT% is statistically better than 3PT%, but in practice they're about equally bad, and together they're only kind of good.

Case closed, I was right. You initially quoted my disputing my claim, so there's that.

Uh you're not right lol. You said FT% is by far the best statistic for predicting NBA 3PT% when it's better by an insignificant margin and is equally useless as 3PT% alone. That's the whole point that you seem to keep whiffing on.

I'm not 'whiffing' on the point. You just don't fully understand the link between univariate and multivariate regression. FT % is the best single factor at predicting future 3pt sucess, I don't understand why you are even disputing this fact. I'm done digging through this rabbit hole with you when you have already admitted that my initial claim is correct.

Regardless, I'm as Bullish about Tatum shooting well in the league as any prospect this year.

And for your reference, I work as a statistician as my day job.

[shawn_hemp]

If only they played the games on calculators right?

[Duke4life831]

Its definitely not a guarantee that Tatum is going to be a good 3pt shooter. But if you were to put money on a guy that can add that part to his game Tatum has to be up there as one of the safest bets. Phenomenal FT shooter, great form on his shot, great from mid range and has already shown potential with his 3 with shooting 37% from 3 in conference play. Is he going to be money from 3 from day 1? No but in a few years when he hits 22-23 I think he will be just fine.

If he doesnt add a 3 point shot I think you can look a DeRozan as a comparison on how he could still be a great scorer. DeRozan didnt have the mid range game that Tatum has when DeRozan was at USC, I could easily see Tatum being a great scorer just from his mid range and being great at the line.

[Mulhollanddrive]

Out of all the flaws you can have, being a shooter who shoots 33% from 3 at college, is far from the worst.

DeRozan was a 16% three point shooter in college.

[kalenclayton]

Jayson Tatum kinda looks like a better shooting, slightly less athletic Rudy Gay. He also looks like he is more willing to pass than Rudy. If that is the player he turns in to, then we will have a problem on our hands. Does anyone else see this?

[Nick Sigler]

Jayson Tatum kinda looks like a better shooting, slightly less athletic Rudy Gay. He also looks like he is more willing to pass than Rudy. If that is the player he turns in to, then we will have a problem on our hands. Does anyone else see this?

That's exactly what I see and I think he'll follow a pretty similar career trajectory to Rudy - putting up good scoring numbers on a bunch of bad teams, because he can get buckets, but he's not a player you can ever build around or really feature if you're going to be a playoff team.

[kalenclayton]

Jayson Tatum kinda looks like a better shooting, slightly less athletic Rudy Gay. He also looks like he is more willing to pass than Rudy. If that is the player he turns in to, then we will have a problem on our hands. Does anyone else see this?

That's exactly what I see and I think he'll follow a pretty similar career trajectory to Rudy - putting up good scoring numbers on a bunch of bad teams, because he can get buckets, but he's not a player you can ever build around or really feature if you're going to be a playoff team.

I'm glad I'm not alone. I am a Kings fan and have watched Rudy countless nights so I mostly agree with your statement. One thing to note is that Tatum is super young and if he is a more willing passer, I think he can definitely be something great. The Grizzlies molded Rudy into a 1st option, iso type of player when he should've been coached out of that. Unfortunately, they needed to rely on him and he could never shake that mentality. He's a great baller, but not the best team player (he's a fantastic guy to have on the bench though; great attitude). If Tatum can be coached to be a team guy and not be the "star/I need to get my shots" guy, then he can definitely be on a winning team (possibly even THE guy). The question is "who is the right coach".

[doordoor123]

Jayson Tatum kinda looks like a better shooting, slightly less athletic Rudy Gay. He also looks like he is more willing to pass than Rudy. If that is the player he turns in to, then we will have a problem on our hands. Does anyone else see this?

That's exactly what I see and I think he'll follow a pretty similar career trajectory to Rudy - putting up good scoring numbers on a bunch of bad teams, because he can get buckets, but he's not a player you can ever build around or really feature if you're going to be a playoff team.

I can understand what you mean, but I disagree. Rudy gets a lot of his value from defense and rebounding. He also lives in the mid-post. Gay slows the game down. Tatum likes to drive in the open court, dish in transition, he likes to abuse mismatches by driving from the post and he likes the area right before the restriction line. He also has a better feel for the game in general. Gay plays by himself and Tatum is more of a team player IMO.

[Nick Sigler]

Jayson Tatum kinda looks like a better shooting, slightly less athletic Rudy Gay. He also looks like he is more willing to pass than Rudy. If that is the player he turns in to, then we will have a problem on our hands. Does anyone else see this?

That's exactly what I see and I think he'll follow a pretty similar career trajectory to Rudy - putting up good scoring numbers on a bunch of bad teams, because he can get buckets, but he's not a player you can ever build around or really feature if you're going to be a playoff team.

I can understand what you mean, but I disagree. Rudy gets a lot of his value from defense and rebounding. He also lives in the mid-post. Gay slows the game down. Tatum likes to drive in the open court, dish in transition, he likes to abuse mismatches by driving from the post and he likes the area right before the restriction line. He also has a better feel for the game in general. Gay plays by himself and Tatum is more of a team player IMO.

Good points, but I get the feeling the majority of Tatum's scoring against NBA level defenders is going to have to come from that mid-post area that Gay works in, slowing the game down just like Gay does.

[doordoor123]

That's exactly what I see and I think he'll follow a pretty similar career trajectory to Rudy - putting up good scoring numbers on a bunch of bad teams, because he can get buckets, but he's not a player you can ever build around or really feature if you're going to be a playoff team.

I can understand what you mean, but I disagree. Rudy gets a lot of his value from defense and rebounding. He also lives in the mid-post. Gay slows the game down. Tatum likes to drive in the open court, dish in transition, he likes to abuse mismatches by driving from the post and he likes the area right before the restriction line. He also has a better feel for the game in general. Gay plays by himself and Tatum is more of a team player IMO.

Good points, but I get the feeling the majority of Tatum's scoring against NBA level defenders is going to have to come from that mid-post area that Gay works in, slowing the game down just like Gay does.

See, that's what I get. It's possible because of his shorter wingspan, but I think he's quicker than Gay and his passing will help give him open lanes to the baskets.