Wizards Training Camp 2020-21

[payitforward]

But I thought real plus minus was “bogus”

It's demonstrably bogus, Nat. But that doesn't mean it's wrong.

[dckingsfan]

...Bonga.... turns the ball over at a pretty high rate....

?? I have him at less than 1.85 TOs per 40 minutes. What am I missing?

I have him with a TO% of 17.6... is that wrong?

[payitforward]

No, I just don't understand it.

Hover your mouse over tov% on https://www.basketball-reference.com/players/b/bongais01.html -- you get an explanation that reads "An estimate of turnovers committed per 100 plays."

I don't understand this way of measuring anything. Per 36 minutes last year, Bonga committed 1.7 TOs. Period. He could only have committed 1.7 fewer than that.

[dckingsfan]

No, I just don't understand it.

Hover your mouse over tov% on https://www.basketball-reference.com/players/b/bongais01.html -- you get an explanation that reads "An estimate of turnovers committed per 100 plays."

I don't understand this way of measuring anything. Per 36 minutes last year, Bonga committed 1.7 TOs. Period. He could only have committed 1.7 fewer than that.

Ah, as I understand it, TOV% = (TOV ÷ (FGA + (0.44 x FTA) + TOV)) x 100%

Where:
•“TOV” is the number of turnovers allowed.
•“FGA” is the number of field goals attempted (do not include free throws).
•“FTA” is the number of free throws attempted.

But your point is a good one. Why not just use TOs as the metric. And maybe that is the better metric.

Here is brief description of TOV%.

https://www.pivotanalysis.com/post/what-is-turnover-percentage

[payitforward]

The problem is that formulas of that kind work well when used on large fields of numbers in the fat part of bell-shaped curves. When you get into the area where the numbers are small, the result of the same formula doesn't mean the same thing. Or, rather, the data doesn't give you information with enough significance that you can achieve any knowledge on its basis that would allow you to take action.

The most obvious case is when you work with a small sample size -- e.g. drawing any conclusion from the pattern of data that emerges from the play of Bonga as a rookie -- when he played 120 minutes.

In fact, it can be true in the opposite direction -- at the edge of the numbers where they are unusually large.

For example, look at PER. Guys who are better shooters usually also shoot more. So, to make sure to reward good shooters by raising their PER, the formula rewards almost anyone for taking more shots (shoot above 30%, & the more shots you take the higher your PER).

But, of course, shooting more doesn't actually make you a better shooter! So, PER winds up being fairly inaccurate, or rather unreliable.

Similarly in this case -- Bonga committed a total of 57 turnovers last year, a small enough number that there's tons of particularity & randomness in it. You can't turn it into a percentage & treat the % as if it were a piece of knowledge about the player.

[dckingsfan]

The problem is that formulas of that kind work well when used on large fields of numbers in the fat part of bell-shaped curves. When you get into the area where the numbers are small, the result of the same formula doesn't mean the same thing. Or, rather, the data doesn't give you information with enough significance that you can achieve any knowledge on its basis that would allow you to take action.

The most obvious case is when you work with a small sample size -- e.g. drawing any conclusion from the pattern of data that emerges from the play of Bonga as a rookie -- when he played 120 minutes.

In fact, it can be true in the opposite direction -- at the edge of the numbers where they are unusually large.

For example, look at PER. Guys who are better shooters usually also shoot more. So, to make sure to reward good shooters by raising their PER, the formula rewards almost anyone for taking more shots (shoot above 30%, & the more shots you take the higher your PER).

But, of course, shooting more doesn't actually make you a better shooter! So, PER winds up being fairly inaccurate, or rather unreliable.

Similarly in this case -- Bonga committed a total of 57 turnovers last year, a small enough number that there's tons of particularity & randomness in it. You can't turn it into a percentage & treat the % as if it were a piece of knowledge about the player.

You make a couple of good points. Especially about Bonga's rookie season. And (your underlying point) he is still very young and we still have a very small sample size. And another underlying point (he "seems to be a hard worker and very focused, see what he did with his body before the bubble).

Happy to lean in on your 57 TOs and focus on that. If he can raise his 3PT percentage, take a few more of them and keep the TOs low as a percentage of his utilization - he should be well on his way.

But I guess the best metric on TOs isn't just the number of TOs. I would think utilization would be a factor. Then again, a big running around setting bad screens and creating TOs wouldn't necessarily have a high utilization.

[wall_glizzy]

No, I just don't understand it.

Hover your mouse over tov% on https://www.basketball-reference.com/players/b/bongais01.html -- you get an explanation that reads "An estimate of turnovers committed per 100 plays."

I don't understand this way of measuring anything. Per 36 minutes last year, Bonga committed 1.7 TOs. Period. He could only have committed 1.7 fewer than that.

Ah, as I understand it, TOV% = (TOV ÷ (FGA + (0.44 x FTA) + TOV)) x 100%

Where:
•“TOV” is the number of turnovers allowed.
•“FGA” is the number of field goals attempted (do not include free throws).
•“FTA” is the number of free throws attempted.

But your point is a good one. Why not just use TOs as the metric. And maybe that is the better metric.

Here is brief description of TOV%.

https://www.pivotanalysis.com/post/what-is-turnover-percentage

The reason you use TOV% is the same as the reason you use any rate stat - it normalizes for both pace and usage. Just looking at turnovers, whether raw totals or the amount per some arbitrary number of minutes, tells you nothing about how frequently the ball was in the player's hands, and thus nothing about the relationship between the turnover total and that player's total time in possession of the ball. There are two ways a player can "use" (i.e. use up) a team possession - taking a shot or turning the ball over. TOV% is the percentage of individual possessions by a player which ended in the player turning the ball over.

17.6% is high, for sure - it's actually the number Russell Westbrook posted his rookie year, which he hasn't hit since despite posting very high turnovers from the troglodyte per-minute perspective. However, I don't think it's a total death knell, since we've clearly been kicking the tires on Bonga's play-making abilities - adding that to his already promising 3-and-D profile would be massive. But yeah, 17.6 would be a concerning TOV% for, like, Danny Green or somebody.

On the other hand, it's useful to consider TOV% and USG% together (as we do in the case of Westbrook) - a high TOV% isn't a disaster if you're the engine using some 35-40% of your entire team's offensive possessions (much as a drop in TS% would be expected in the same situation), but given Bonga's very low usage rate of 11% last year, one would hope that he'd be discerning enough to make good use of the possessions he took. Brad Beal, by contrast, used over three times as many possessions (34.4 USG%) while turning it over about 2/3 as often (11.4 TOV%). All that, while assisting his teammates nearly four times as often (29.5 AST% to Bonga's 7.9)!

[SUPERBALLMAN]

Yay we actually get to watch them in action tonight!!

[payitforward]

The reason you use TOV% is the same as the reason you use any rate stat - it normalizes for both pace and usage. Just looking at turnovers, whether raw totals or the amount per some arbitrary number of minutes, tells you nothing about how frequently the ball was in the player's hands, and thus nothing about the relationship between the turnover total and that player's total time in possession of the ball....

Yes, of course. The logic is impeccable. No different from if one were to look at ppg w/o attention to FG%.

The problem is when you go from that well-formed piece of information to make some knowledge claim about a player. If you work with a big sample, you can do that. If you work with small numbers, you can't.

Why doesn't it work? B/c you can't get rid of random factors. I could list a half dozen or more of those, as could you with a moment's thought. That's why there's nothing in particular to learn from Bonga's TO%.

In a way, you make my point for me when you write...

...17.6% is high, for sure - ...17.6 would be a concerning TOV% for, like, Danny Green or somebody....

But, then you compound the previous error with...

..it's useful to consider TOV% and USG% together... - a high TOV% isn't a disaster if you're the engine using some 35-40% of your entire team's offensive possessions (much as a drop in TS% would be expected in the same situation), but given Bonga's very low usage rate of 11% last year, one would hope that he'd be discerning enough to make good use of the possessions he took....

This statement assumes a conclusion that can't be drawn, then reads that conclusion back into the data.

...Brad Beal, by contrast, used over three times as many possessions (34.4 USG%) while turning it over about 2/3 as often (11.4 TOV%). All that, while assisting his teammates nearly four times as often (29.5 AST% to Bonga's 7.9).

Again, worsening an error margin that is inherent in the data by crossing it with another of the same character.

In this case, fortunately, you've provided a case that makes the point obvious.

Consider the circumstances in which Isaac Bonga possesses the ball -- would they be likely to offer a set of opportunities to get an assist comparable to the opportunities offered when Brad possesses the ball?

For a different set of reasons, but in exactly the same way, it would be almost impossible for Brad to turn the ball over on 17.6% of his possessions. OTOH, if I were to pick out all the possessions Brad has that are maximally similar to the occasions the ball is in Bonga's hands, his TOV% on those possessions would be much higher than his overall TOV%.

Without wanting to make a big deal out of it, I need to say that the above was written by someone with years of graduate school study in philosophy. & that I only write it, because you (wall_glizzy) are engaged in interesting & ambitious attempts to use logical & statistical tools to understand basketball players. Above all, I *don't* write it out of any desire to argue with you: I'm a fan of yours! Capisce?

[J-Ves]

5 rotational players are out including 3 starters, but there’s still plenty to look forward to see tonight. TShepp has done well to acquire cheap young talent worth a damn.

[nate33]

5 rotational players are out including 3 starters, but there’s still plenty to look forward to see tonight. TShepp has done well to acquire cheap young talent worth a damn.

Can you be more specific? I saw that Bertans and Lopez are probably out, and Beal and Westbrook might not play. Who is the 5th?

[J-Ves]

5 rotational players are out including 3 starters, but there’s still plenty to look forward to see tonight. TShepp has done well to acquire cheap young talent worth a damn.

Can you be more specific? I saw that Bertans and Lopez are probably out, and Beal and Westbrook might not play. Who is the 5th?

I saw Ish, Beal, Westbrook, Bertans, and Robinson are OUT. Bryant was listed as questionable.

[nate33]

5 rotational players are out including 3 starters, but there’s still plenty to look forward to see tonight. TShepp has done well to acquire cheap young talent worth a damn.

Can you be more specific? I saw that Bertans and Lopez are probably out, and Beal and Westbrook might not play. Who is the 5th?

I saw Ish, Beal, Westbrook, Bertans, and Robinson are OUT. Bryant was listed as questionable.

WTF? Why are Ish, Robinson and Bryant out/questionable?

Eh, I guess the more vets out, the better. Westbrook, Beal, Bertans, Ish and Lopez missing is no big deal. I'd like to see Robinson and Bryant play though.

[DCZards]

Can you be more specific? I saw that Bertans and Lopez are probably out, and Beal and Westbrook might not play. Who is the 5th?

I saw Ish, Beal, Westbrook, Bertans, and Robinson are OUT. Bryant was listed as questionable.

WTF? Why are Ish, Robinson and Bryant out/questionable?

Eh, I guess the more vets out, the better. Westbrook, Beal, Bertans, Ish and Lopez missing is no big deal. I'd like to see Robinson and Bryant play though.

Bryant has a hip issue and Robinson has a stomach ailment, I believe.

[nate33]

Looks like Neto and Winston at PG and Wagner and Pasecniks at the 5!

[NatP4]

Westbrook is apparently not going to play in back to backs this year. How many back to backs are we going to have?

[Rafael122]

Westbrook is apparently not going to play in back to backs this year. How many back to backs are we going to have?

Only the first half schedule has been announced, but we have 7, 3 of which are literal home/home games where it's a back to back but both games are in DC.

[closg00]

Watch on YouTube

[closg00]

Watch on YouTube

[NatP4]

Westbrook, Bertans out, Rui and Mathews questionable

Might get an extended look at Gill and Robinson tonight