Trade Blazers trade for Robert Covington!!

[wizenheimer]

RoCo is more like 4/4 and 0/4 any given night. He's significantly more extreme in that respect.

yeah, that's pretty accurate....he's always been a streaky shooter who can get hot for a weeks or two, than go stone cold for a week or two

his biggest drop in efficiency that I see is that he came to Portland with a career FT rate of .215, but is at a career-worst 0.98 this season. I'd estimate that's a result of being parked in corners and outside the three point line. Hard to get to the line without the ball and dribble-drives. He's also shooting a career worst .571 at the rim, and by far, a career worst .167 from 3-10'. Don't know what that's about

 

[handiman]

I've wondered if this is validated by the numbers. Sure feels this way doesn't it?

yeah, that's pretty accurate....he's always been a streaky shooter who can get hot for a weeks or two, than go stone cold for a week or two

Yeah, I'd be surprised if the data didn't bear it out, but I've never seen that sort of info published. Percentages are one thing, but is standard deviation recorded anywhere for shooting? Guys like B. Forbes and Seth Curry seem to be 2/4 or 3/6 every game, so I'd expect them to have a very tight S.D. in comparison.

Even early in his career, I recall RoCo's reputation bouncing between being a shooter and a defender, largely because his shot seemed to come and go by the season.

 

[wizenheimer]

Yeah, I'd be surprised if the data didn't bear it out, but I've never seen that sort of info published. Percentages are one thing, but is standard deviation recorded anywhere for shooting? Guys like B. Forbes and Seth Curry seem to be 2/4 or 3/6 every game, so I'd expect them to have a very tight S.D. in comparison.

Even early in his career, I recall RoCo's reputation bouncing between being a shooter and a defender, largely because his shot seemed to come and go by the season.

maybe look at his monthly splits? that might be too many games though and not cover week to week variance

I just arbitrarily opened his 2017-18 splits:



quite a bit of variance there

2016-17:



2019-20 (was injured in 2018-19):



my guess is that you'd probably see similar variance for any player with career marks of 40.4% on FG's and 35.5% on three's. I know Aminu is probably all over the place in his splits. The curse of role players maybe

for that matter, I think Dame probably has a wide variance, month to month, on three's

 

[bobf]

Yeah, I'd be surprised if the data didn't bear it out, but I've never seen that sort of info published. Percentages are one thing, but is standard deviation recorded anywhere for shooting? Guys like B. Forbes and Seth Curry seem to be 2/4 or 3/6 every game, so I'd expect them to have a very tight S.D. in comparison.

Even early in his career, I recall RoCo's reputation bouncing between being a shooter and a defender, largely because his shot seemed to come and go by the season.

I’ve been thinking about this - how to measure the streakiness of a shooter. Its not so clear how to do it. First thought is, just use standard deviation. Ok. But standard deviation of what? Field Goal % by game? Problem with that is a player who takes more shots per game will have a smaller standard deviation in FG% without being less streaky. For example, in the extreme, a 50% shooter who takes only one shot per game will oscillate between 0% and 100% FG% per game. Lower volume players will appear to be more streaky. Ideally a measure would work across players and not be biased by volume.

So I started thinking of just looking at the stream of shots alone, not grouped by game or week or month...

What would it mean for a 50/50 coin of 100 flips to be perfectly consistent? I would say this:
HTHTHTHTHTHT..... (100 times)
because for any even-length sub-segment of flips the coin is 50% and for any odd sub-segment it’s as close to 50% as possible. How can you be more consistent than that?

And what stream of 100 coin flips is as streaky as possible for a 50/50 coin?

HHHH.... (50 times) TTTTT.... (50 times)

So how to turn those into a number? I just count the number of switches H to T or T to H.

In the first case the coin “switches” 99/99 times. In the seconds case the coin switches 1/99 times. So a measure of 1.0 is perfectly consistent. A measure close to zero is perfectly streaky. A measure close to 0.5 is what a fair coin would do.

That’s as far as my thinking has gone. A problem with my measure is a 60% shooter will tend to be measured as more streaky than a 50% shooter.

Anyway, I have hooked into the NBA stats api and could easily run any streakiness measure against it, just not sure yet what unbiased measure would best capture streakiness.

 

[handiman]

I’ve been thinking about this - how to measure the streakiness of a shooter. Its not so clear how to do it. First thought is, just use standard deviation. Ok. But standard deviation of what? Field Goal % by game? Problem with that is a player who takes more shots per game will have a smaller standard deviation in FG% without being less streaky. For example, in the extreme, a 50% shooter who takes only one shot per game will oscillate between 0% and 100% FG% per game. Lower volume players will appear to be more streaky. Ideally a measure would work across players and not be biased by volume.

Fair point, but I think the consistent shooters (like the Forbes/Curry examples) will still have a low S.D. because they're typically right around 50% even on medium-low volume. Once you've run up and down the court a couple times to get the blood flowing, there's no reason why your first shot should be any less likely to go in than your fifth. So, I look at low-volume shooting variance as a magnifier more than anything, and it's probably somewhat countered by volume shooters having stretches of missing a handful shots in a row that the low-volume guys would never even get to match in a game.

 

[wizenheimer]

I’ve been thinking about this - how to measure the streakiness of a shooter. Its not so clear how to do it. First thought is, just use standard deviation. Ok. But standard deviation of what? Field Goal % by game? Problem with that is a player who takes more shots per game will have a smaller standard deviation in FG% without being less streaky. For example, in the extreme, a 50% shooter who takes only one shot per game will oscillate between 0% and 100% FG% per game. Lower volume players will appear to be more streaky. Ideally a measure would work across players and not be biased by volume.

So I started thinking of just looking at the stream of shots alone, not grouped by game or week or month...

What would it mean for a 50/50 coin of 100 flips to be perfectly consistent? I would say this:
HTHTHTHTHTHT..... (100 times)
because for any even-length sub-segment of flips the coin is 50% and for any odd sub-segment it’s as close to 50% as possible. How can you be more consistent than that?

And what stream of 100 coin flips is as streaky as possible for a 50/50 coin?

HHHH.... (50 times) TTTTT.... (50 times)

So how to turn those into a number? I just count the number of switches H to T or T to H.

In the first case the coin “switches” 99/99 times. In the seconds case the coin switches 1/99 times. So a measure of 1.0 is perfectly consistent. A measure close to zero is perfectly streaky. A measure close to 0.5 is what a fair coin would do.

That’s as far as my thinking has gone. A problem with my measure is a 60% shooter will tend to be measured as more streaky than a 50% shooter.

Anyway, I have hooked into the NBA stats api and could easily run any streakiness measure against it, just not sure yet what unbiased measure would best capture streakiness.

I really want to watch the HCP's head explode when he reads that

 

[bobf]

I really want to watch the HCP's head explode when he reads that

I can guess what his one-word response will be.

 

[kjironman1]

I’ve been thinking about this - how to measure the streakiness of a shooter. Its not so clear how to do it. First thought is, just use standard deviation. Ok. But standard deviation of what? Field Goal % by game? Problem with that is a player who takes more shots per game will have a smaller standard deviation in FG% without being less streaky. For example, in the extreme, a 50% shooter who takes only one shot per game will oscillate between 0% and 100% FG% per game. Lower volume players will appear to be more streaky. Ideally a measure would work across players and not be biased by volume.

So I started thinking of just looking at the stream of shots alone, not grouped by game or week or month...

What would it mean for a 50/50 coin of 100 flips to be perfectly consistent? I would say this:
HTHTHTHTHTHT..... (100 times)
because for any even-length sub-segment of flips the coin is 50% and for any odd sub-segment it’s as close to 50% as possible. How can you be more consistent than that?

And what stream of 100 coin flips is as streaky as possible for a 50/50 coin?

HHHH.... (50 times) TTTTT.... (50 times)

So how to turn those into a number? I just count the number of switches H to T or T to H.

In the first case the coin “switches” 99/99 times. In the seconds case the coin switches 1/99 times. So a measure of 1.0 is perfectly consistent. A measure close to zero is perfectly streaky. A measure close to 0.5 is what a fair coin would do.

That’s as far as my thinking has gone. A problem with my measure is a 60% shooter will tend to be measured as more streaky than a 50% shooter.

Anyway, I have hooked into the NBA stats api and could easily run any streakiness measure against it, just not sure yet what unbiased measure would best capture streakiness.

Not sure you can track shooting with a coin toss without bringing in other variable.
Your welcome.

 

[bobf]

Not sure you can track shooting with a coin toss without bringing in other variable.
Your welcome.

If I play the game 12 times I can expect each coin 4 times.
1. Two headed coin 4 times 4 heads
2, 75% coin 4 times 3 heads
3. Fair coin 4 times 2 heads

9 times I will get heads. 4 times it was the two headed coin. So I get 4/9.

But I’m not using coins to track shots. I’m using a coin to show that my measure of streakiness / consistency is reasonable. I’m open to suggestions for a better measure. I’m looking for a measure that is not biased by shot volume.

 

[bobf]

Fair point, but I think the consistent shooters (like the Forbes/Curry examples) will still have a low S.D. because they're typically right around 50% even on medium-low volume. Once you've run up and down the court a couple times to get the blood flowing, there's no reason why your first shot should be any less likely to go in than your fifth. So, I look at low-volume shooting variance as a magnifier more than anything, and it's probably somewhat countered by volume shooters having stretches of missing a handful shots in a row that the low-volume guys would never even get to match in a game.

I think if you take one shooter and compare his std dev of fg% on 10 shot nights vs 15 shot nights his 10 shot nights will have a higher std dev. Not because he’s less consistent. It’s not a good measure if I can only compare shooters of equal volume. Maybe there’s a way to scale it..

 

[kjironman1]

If I play the game 12 times I can expect each coin 4 times.
1. Two headed coin 4 times 4 heads
2, 75% coin 4 times 3 heads
3. Fair coin 4 times 2 heads

9 times I will get heads. 4 times it was the two headed coin. So I get 4/9.

But I’m not using coins to track shots. I’m using a coin to show that my measure of streakiness / consistency is reasonable. I’m open to suggestions for a better measure. I’m looking for a measure that is not biased by shot volume.

I'm talking about other human variables like life issues and just overall random acts that we all deal with.

 

[bobf]

I'm talking about other human variables like life issues and just overall random acts that we all deal with.

I don't understand what you are getting at. Life issues etc. are possible causes of streakiness but I'm not looking at causes, I'm just trying to find a way to measure "is this guy streaky" or "is this guy consistent" whatever the cause. Just like we have 3pt FG% to measure "is this guy a good 3 point shooter". It's not a perfect measure. It doesn't account lots of things, but it's a useful starting point - all else being equal a higher 3pt% is better. Then we factor in what we know by observation that is not "equal" such as Dame's 3's are tougher, contested more, he can get them off almost at will and in the clutch.

I'm trying to find a similar way to measure streakiness. Standard deviation seems like the obvious way, but standard deviation of FG% game by game is biased badly by volume. So I'm looking for something else. My whole coin flip thing was to put forward a measure that seems to match my intuition of streakiness and is not biased by volume. HTHTHTHTHTHTHTH is consistent HHHHHHHHTTTTTTTTT is steaky. I could have said make/miss instead of HT but they both start with "M"!

That's it.

 

[Rastapopoulos]

 

[illmatic99]

i think part of that has to do with the guy replacing Roco being so bad defensively. he does alternate with Enes/Melo in the rotation, who are consistently two of our worst defenders.

but then again, roco is a legitimately good defender even in a vacuum. good find!

 

[B-Roy]

So what’s everyone’s thoughts on Covington halfway into the season?

His counting stats are pretty bad, and he went through a terrible shooting slump early, but damn if he isn’t one of my favorite role player in years. I compare him to Wes in terms of his effort every game.

Also think his game will translate well into his 30s. Hell Ariza is 36 and still the same player. Think RoCo could at least do that.

 

[Scalma]

Keep forever

 

[wizenheimer]

So what’s everyone’s thoughts on Covington halfway into the season?

His counting stats are pretty bad, and he went through a terrible shooting slump early, but damn if he isn’t one of my favorite role player in years. I compare him to Wes in terms of his effort every game.

Also think his game will translate well into his 30s. Hell Ariza is 36 and still the same player. Think RoCo could at least do that.

he has settled into the Blazers systems, such as they are, and is contributing on both ends of the floor. And Portland asks him to do a lot on defense, yet he still pitches in on offense. He's also rebounding fairly well for his size

one thing I don't think many realized is that RoCo has always had pretty high variance in his 3 point shooting. He's streaky, which really isn't that unusual for a player like him. He'll never shoot like Kyle Korver, but he will hit some big shots every once in a while, like tonight. I don't think he's as good a PF as Aminu, but he's a better wing than either Aminu or Harkless. He's the best Blazer forward since Aldridge, but that bar has been set pretty low

 

[SIeepwalker]

I hope he retires a Blazers

 

[Rastapopoulos]

So what’s everyone’s thoughts on Covington halfway into the season?

His counting stats are pretty bad, and he went through a terrible shooting slump early, but damn if he isn’t one of my favorite role player in years. I compare him to Wes in terms of his effort every game.

Also think his game will translate well into his 30s. Hell Ariza is 36 and still the same player. Think RoCo could at least do that.

He's been our second most important player for about the past month. He's even better than Aminu (who was a lot better than most people would admit). He leads the team in +/-, so that'll automatically turn the usual suspects off him. He may even be good enough for a while that people won't moan that we could've taken Isiah Stewart/Precious Achiuwa/Saddiq Bey/Tyrese Maxey/Immanuel Quickley/Desmond Bane/Xavier Tillman this draft and whomever next draft if we hadn't traded for him.

 

[wizenheimer]

I'll add: I'd be willing to trade Jones for Aaron Gordon, depending on who/what else was included. I would not trade RoCo for Gordon

 

[Rastapopoulos]

We traded 2 FIRSTS WTF?

You were right to protest, it turns out.