PaKii94 wrote:dougthonus wrote:...
When you break the sets into smaller subgroups instead of using the larger data set, the sample size becomes less relevant and meaningful.
Agreed. But that's why you start with the largest dataset and THEN chunk it down to build up a story of how it was formed. You don't start with the smaller set.
:dontknow:
I'm not convinced that timeseries data, in this instance, is more useful than the averages. I agree that generally you weight more recent data, but in this case, the data is consistent across a three year period.
I, as a data scientist by profession, am trying to tell you that analyzing timeseries data as a timeseries IS more useful than the averages. The data IS consistent with an injury trend over a three year period (and for other players too).
The variations in what he has done are not what I would expect to be out of bounds for normal variations of guys getting on hot and cold streaks.
But your expectations are flawed then! I literally generated graphs for you showing you what the difference between normal variations and statistically significant trends are! I asked you for an example of a player that you thought was similar (Niko) and then provided data to show it's a different situation (and even that player had an injury trend) but you still don't believe it. I truly don't know how to explain it otherways.
How have you determined his injury dates to see what data falls into what branch? It would appear that you started with the assumption he has healthy and unhealthy data sets, and lumped the data into those sets based on performance rather than coming up with dates he was healthy and unhealthy and then creating the data sets based on that. This means by default, your data is absolutely destined to support your opinion, because it was segmented to back your opinion not based on knowledge of when he was actually hurt.
I am slightly offended that you think I would approach it like this. I determined the cutoff by the dates reported of the injury/when it was seen during the game. If a player misses some games that they say are due to a physical ailment, that's probably an injury right? If he is attempting to play through something, his game is expected to dip before recovery. Go back and my read post it has the injury dates:
viewtopic.php?f=10&t=1947141&p=82217073#p82217073
?s=20
^ That's the oblique injury.
^It literally took 1 month of recovery, of Lauri playing passive/soft/taking it easy/playing the perimeter role, as not to reinjure it and lines up with him giving a f*ck the next month and what do you see? 21ppg @ 64TS%.
He was
trending up during the hot month, not down (which is what you would expect to see with normal hot streaks as players cool off). He was at 25 ppg @ 65TS%.
When does his play fall off again? Right when he rolls his ankle pretty severely (compounded with WCJ going down and him having to play C) here:
?s=20
See how he's speculating he might miss a few weeks? Lauri ended up continuing to play (but passively/soft/recovery mode Lauri).
^All of the above lines up with the eye test also. I am rewatching the games again and just finished Nov 3rd Pacers game. The difference in play/energy/competitiveness flipped like a switch. It wasn't any "cold" streak. It was a deliberate move. The other players weren't looking for him and he wasn't looking for the ball.
A player doesn't go from one month of 45TS% to one month of 65TS% as a cold/hot streak.
-----
I think what it says is to me actually is that big men shooters (at least last year) weren't very valuable or were pretty rare. I think teams have largely transitioned to just going smaller.
True but then you have to remember those smaller players are then classified as PF/C also and are also included in the dataset. Also, if you expand that list to include SF-PF hybrids, it's still a very small list.
I think this is a different discussion. We were starting with the premise of whether he was an elite shooter. An elite shooter is going to take more unassisted threes and more threes with less space. It's valuable, absolutely, in looking at the player's overall offensive abilities of course, just separate from their ability as a shooter IMO.
Sure I agree with that. but like I said, there are maybe a handful right now who can do that (and maybe 1 big). I classify those as generational/HOF shooters. I consider Korver/Ray allen/Klay to be elite shooters but they wouldn't be elite by your definition.
Could Lauri create a few more 3s himself? Yes. I don't really expect that from him before he improves all the other aspects of his game. Also he does need the opportunity/volume/usage to do it. He takes smart shots. If he's only taking 10, he's not going to take contested pull up 3s like lavine.
Finally, unassisted 3s ARE a small subset. Lauri would only need to make 1 unassisted 3 more every 10 games to be on a competitive level compared to other F/Cs
Why expect only the part that was bad to get better, but the part that was good to not decline rather than both parts to regress towards the mean?
These numbers can't regress to the means because they
are the overall means! I expect his numbers to regress towards his healthy numbers. This year was pretty even distribution of healthy vs unhealthy games (roughly 50-50).
Then you can break it down further and compare healthy Lauri vs injured Lauri trends year to year.
Healthy Lauri went roughly 36 - 38 - 40 -> a consistent trend upwards including his troubles with the wing
Injured Lauri was consistent...consistently bad... roughly 30-34%. That's also a pretty clear trend.
I am not sure I trust the way you determined healthy vs unhealthy. It feels like you let the data decide when he was healthy instead of finding when he was healthy and looking at the data. This means the data will automatically match your conclusion, because it was constructed to determine his health based on the results rather than determining his health based on his health. At least, it appears that way to me, but maybe I am incorrect on this point, and you have better ways of tracking when he was fully healthy vs not than I understand.
I covered that above. These subsets are from the reported dates of injury. But again, I am slightly offended. I wouldn't gerrymander the data to fit my narrative. My hope is to see Bulls win. I don't prop up/bash players depending if I like them or not. I am a Lauri fan because I know he is a good player. I don't think he's a good player just because I am Lauri fan
. None of my IRL friends are invested enough in the bulls to actually hold a discussion with outside of the media clickbait talking points. They just watch the lavine highlights on youtube and tell me "Bulls need to get Lavine some help!!!" 
