Romeo Langford(merged threads)

[Somebody]

I also have some ideas of how to improve the WS/48 and MPG statistics to be more meaningful but will share them when I do some work on that

I look forward to seeing what you do. I’m very interested to know what you mean by ‘more meaningful’, for instance. That sounds worth hearing about.

I meant to measure the quality of a draft choice or how would a player ranks against comparable draft picks.
WS/48 (like you just logged in minutes on a winning team and your draft peers didn’t) and MPG (you play for a bad team and the minutes were not earned by you but given to you unlike your peers) ranking may have a lot of noice.
I was thinking of some sort of Gini coefficient that controls for that noise.... like multiply the percentile ranking of two statistics and see how close A player is to 100%... for instance :
If you rank a 15th pick in the range 13-20 draft picks from the last 20 drafts Giannis and Kawhi will be close to 100% in both MPG and WS/48 hence their Gini will also be close to 100%
Players that get less minutes will be all over the 0-100% spectrum - with the average or I should say median at around 25%

And for the hard core data scientists this is not exactly Gini what I’m describing but more of a index^2

Obviously the top picks 1-5 are not easy to measure as a lot of drafting depends on consensus picks, game theory and strategy (trading down)... having a star at the position that the best available player plays etc..

Once you go past pick 10 statistics/rankings can be used and or justified as a measure as consensus means less ,drafting mistakes are less costly and there are a lot of trade downs and up to the point that at the 2019 draft the commentators couldn’t keep up with all the action.

I think further nitpicking of box based stats (unless you adjust for the quality of them, eg. the value of an assist/bucket/block/steal) won't help much in your quest of evaluating young players. Incorporating impact metrics over a 3-5 year stretch (preferably closer to 5) while accounting for the players' roles on the team (eg. end of bench player, rotation player, starter, star), team/opponent strength, injuries, backup/starter quality (depending on whether the prospect in question is a starter or bench player), etc will give your model a more holistic evaluation of young players.

[NKY fan]

I also have some ideas of how to improve the WS/48 and MPG statistics to be more meaningful but will share them when I do some work on that

I look forward to seeing what you do. I’m very interested to know what you mean by ‘more meaningful’, for instance. That sounds worth hearing about.

I meant to measure the quality of a draft choice or how would a player ranks against comparable draft picks.
WS/48 (like you just logged in minutes on a winning team and your draft peers didn’t) and MPG (you play for a bad team and the minutes were not earned by you but given to you unlike your peers) ranking may have a lot of noice.
I was thinking of some sort of Gini coefficient that controls for that noise.... like multiply the percentile ranking of two statistics and see how close A player is to 100%... for instance :
If you rank a 15th pick in the range 13-20 draft picks from the last 20 drafts Giannis and Kawhi will be close to 100% in both MPG and WS/48 hence their Gini will also be close to 100%
Players that get less minutes will be all over the 0-100% spectrum - with the average or I should say median at around 25%

And for the hard core data scientists this is not exactly Gini what I’m describing but more of a index^2

Obviously the top picks 1-5 are not easy to measure as a lot of drafting depends on consensus picks, game theory and strategy (trading down)... having a star at the position that the best available player plays etc..

Once you go past pick 10 statistics/rankings can be used and or justified as a measure as consensus means less ,drafting mistakes are less costly and there are a lot of trade downs and up to the point that at the 2019 draft the commentators couldn’t keep up with all the action.

I think further nitpicking of box based stats (unless you adjust for the quality of them, eg. the value of an assist/bucket/block/steal) won't help much in your quest of evaluating young players. Incorporating impact metrics over a 3-5 year stretch (preferably closer to 5) while accounting for the players' roles on the team (eg. end of bench player, rotation player, starter, star), team/opponent strength, injuries, backup/starter quality (depending on whether the prospect in question is a starter or bench player), etc will give your model a more holistic evaluation of young players.

I agree with what you’re saying. I was looking at career stats for the cohorts of players. This methodology has a ton ways to improve ... however bad picks will rank low no matter what because their career is usually over by year 2 or 3.

[Somebody]

I also have some ideas of how to improve the WS/48 and MPG statistics to be more meaningful but will share them when I do some work on that

I look forward to seeing what you do. I’m very interested to know what you mean by ‘more meaningful’, for instance. That sounds worth hearing about.

I meant to measure the quality of a draft choice or how would a player ranks against comparable draft picks.
WS/48 (like you just logged in minutes on a winning team and your draft peers didn’t) and MPG (you play for a bad team and the minutes were not earned by you but given to you unlike your peers) ranking may have a lot of noice.
I was thinking of some sort of Gini coefficient that controls for that noise.... like multiply the percentile ranking of two statistics and see how close A player is to 100%... for instance :
If you rank a 15th pick in the range 13-20 draft picks from the last 20 drafts Giannis and Kawhi will be close to 100% in both MPG and WS/48 hence their Gini will also be close to 100%
Players that get less minutes will be all over the 0-100% spectrum - with the average or I should say median at around 25%

And for the hard core data scientists this is not exactly Gini what I’m describing but more of a index^2

Obviously the top picks 1-5 are not easy to measure as a lot of drafting depends on consensus picks, game theory and strategy (trading down)... having a star at the position that the best available player plays etc..

Once you go past pick 10 statistics/rankings can be used and or justified as a measure as consensus means less ,drafting mistakes are less costly and there are a lot of trade downs and up to the point that at the 2019 draft the commentators couldn’t keep up with all the action.

I think further nitpicking of box based stats (unless you adjust for the quality of them, eg. the value of an assist/bucket/block/steal) won't help much in your quest of evaluating young players. Incorporating impact metrics over a 3-5 year stretch (preferably closer to 5) while accounting for the players' roles on the team (eg. end of bench player, rotation player, starter, star), team/opponent strength, injuries, backup/starter quality (depending on whether the prospect in question is a starter or bench player), etc will give your model a more holistic evaluation of young players.

I agree with what you’re saying. I was looking at career stats for the cohorts of players. This methodology has a ton ways to improve ... however bad picks will rank low no matter what because their career is usually over by year 2 or 3.

Absolutely, bad picks usually get shipped out of the league in a couple of seasons. What I'm suggesting is to build a better statistical model that can offer a more in-depth (and accurate) analysis of prospects in a draft class (or even multiple) by incorporating impact metrics alongside adjusted box stats (OBPM by bball reference is pretty good, haven't seen a box stat that quantifies defence well though, maybe defensive RAPTOR due to its play-tracking/plus-minus/box score hybrid methodology).

[ScoobyDoo]

How about instead of Ainge trying to get super cute in every draft, he just looks at Tankathon's player analysis before making a draft pick:

#14, Romeo Langford:
http://www.tankathon.com/players/romeo-langford     A slew of red (negative) flags.

Take Brandon Clarke instead:
http://www.tankathon.com/players/brandon-clarke     A massive list of green(positive) flags.

At #20, just take Matisse Thybulle   Again, a massive slate of green flags and "historically" disruptive and great defensively at the college level
* Smart and Thybulle had potential to be one of the best defensive back courts of all time, as back ups - Smart can play the point. Carsen Edwards had a load of red flags as well, not green.

At #22, just take Bruno Fernando   May take some time, but has a loaded set of green flags, Versus Grant Williams, who had a load of red flags.
* Get us a big man and since you already took Clarke at #14, you can pass on Grant Williams.

Then, instead of getting super cute and throwing $2.5 million on another French stiff in Poirier just go use that money to sign a player (Noel, O'Quinn, Cauley-Stein, etc., that were still available and who have proven they can actually play and contribute at some level in the NBA). All of those guys signed for about $500K to $1 MM less than we paid for Poirier. 


I don't hate Langford, Williams or Edwards at all. I think Langford has a shot at being special and that both Williams and Edwards can carve out some roles, though both are undersized for today's game (another specialty of Ainge, take guys who are too short for their position and then compoound it by a lack of eleite athleticism and short arms. That's right in Ainge's strike zone.

I like Ainge overall but boy does he seem to make if far more difficult than it needs to be when it comes to the draft.  I'm obviously not saying you just look at Tankathon and pick, but a lot of these picks are kind of smack you in the head obvious, then Ainge usually does the opposite of that. His drafts typically make me nuts.     
 

[ScoobyDoo]

For reference:

http://www.tankathon.com/players/matisse-thybulle

http://www.tankathon.com/players/ty-jerome

http://www.tankathon.com/players/carsen-edwards

http://www.tankathon.com/players/bruno-fernando

http://www.tankathon.com/players/grant-williams

[Fafnir]

Ainge does not care about consensus draft boards when it comes to who he picks. Never has and never will. Reporting is consistent on this, just how he is.

[ScoobyDoo]

Understood. Perhaps he should.  He's good on trades, so so on the draft. If you're great with X's and O's on offensive but not a great defensive mind, hire a great defensive mind. Ego gets in the way sometimes, which is understandable, if unpalatable.

   

[NKY fan]

Understood. Perhaps he should.  He's good on trades, so so on the draft. If you're great with X's and O's on offensive but not a great defensive mind, hire a great defensive mind. Ego gets in the way sometimes, which is understandable, if unpalatable.

   

I have been preaching for Danny to make more trades and less drafting.

[ETNCeltics]

Thybulle is better than all of the C’s picks. Not sure what DA was thinking.

probably what many nba GMs were thinking on draft night - in college thybulle was a bad shooter from outside.

please, if you would, provide us with your opinions on him that you had on draft night. thanks.

What do his opinions on draft night matter? It's not his job to be right about players, it's Danny Ainge's job.

I wouldn't trade Ainge, I think he does a fantastic job overall. That doesn't put him above criticism.

A lot of GMs might've been thinking that about Thybule, but I'd wager they were also thinking there's no way they'd take someone as undersized at Grant Williams with that high of a pick. IIRC, he measured at 6' 5.75 and isn't much of a leaper. Simply can't understand the thought process that led to that pick.

[Fafnir]

No one knows anything about Thybulle's shooting ability yet at this point, 3 point shooting sample size for him is tiny. Other than perhaps a spot up shooter he does look hopeless on offense otherwise. (which when you're that elite on D isn't an issue if you can hit spot up 3s)

I am guessing his limited offensive touch weighed in on Danny and a lot of other teams evaluation of him.

A lot of GMs might've been thinking that about Thybule, but I'd wager they were also thinking there's no way they'd take someone as undersized at Grant Williams with that high of a pick. IIRC, he measured at 6' 5.75 and isn't much of a leaper. Simply can't understand the thought process that led to that pick.

FWIW there were a couple of draft analysts who had Grant pretty high on their board. (Sam Vecenine is one I remember clearly)

His advanced numbers and production were really really good in college, he just doesn't have the physical attributes (height and leaping ability) that normally go along with such production.

[W8ting2McHale]

No one knows anything about Thybulle's shooting ability yet at this point, 3 point shooting sample size for him is tiny. Other than perhaps a spot up shooter he does look hopeless on offense otherwise. (which when you're that elite on D isn't an issue if you can hit spot up 3s)

I am guessing his limited offensive touch weighed in on Danny and a lot of other teams evaluation of him.

A lot of GMs might've been thinking that about Thybule, but I'd wager they were also thinking there's no way they'd take someone as undersized at Grant Williams with that high of a pick. IIRC, he measured at 6' 5.75 and isn't much of a leaper. Simply can't understand the thought process that led to that pick.

FWIW there were a couple of draft analysts who had Grant pretty high on their board. (Sam Vecenine is one I remember clearly)

His advanced numbers and production were really really good in college, he just doesn't have the physical attributes (height and leaping ability) that normally go along with such production.

Exactly.

Ainge knew that he would have a young bench and he needs couple of players that maybe didn’t have a high ceiling, but knew how to play the game and contribute. Grant Williams is that kind of player. We need a cheap bench, and in the NBA, that is either young guys or old guys chasing a ring or last paycheck. Since Ainge had the draft capital he went with someone cost controlled instead of a one year rental or cast off. He took Langford as a player with upside. He wanted a shooter like Herro, but had to settle for Carsen instead.

[Fafnir]

Yeah the Celtics plan A was to draft Herro, who would have been awesome for this team.

[NKY fan]

Yeah the Celtics plan A was to draft Herro, who would have been awesome for this team.

why didn't they trade up for him? It is not like they lacked assets...

[Fafnir]

Yeah the Celtics plan A was to draft Herro, who would have been awesome for this team.

why didn't they trade up for him? It is not like they lacked assets...

With whom? I don't think many of the teams above were looking to move down all the way to 14.

Minnesota already moved down a bunch, Atlanta doesn't need more low round picks, Heat wanted Herro, Charlotte maybe?. Then you have to assess the price and what demands teams were making. Minnesota and Atlanta paid a lot to move up.

Also reports were the C's thought they were getting him, I wonder if they got snookered that the Heat were taking someone else.

[footey]

Yeah the Celtics plan A was to draft Herro, who would have been awesome for this team.

why didn't they trade up for him? It is not like they lacked assets...

With whom? I don't think many of the teams above were looking to move down all the way to 14.

Minnesota already moved down a bunch, Atlanta doesn't need more low round picks, Heat wanted Herro, Charlotte maybe?. Then you have to assess the price and what demands teams were making. Minnesota and Atlanta paid a lot to move up.

Also reports were the C's thought they were getting him, I wonder if they got snookered that the Heat were taking someone else.

Most mock drafts had Herro falling to Celtics at earliest; Heat swooped him up one pick before us.