Posted by oldave on 2/13/2014 4:30:00 AM (view original):
Seriously. Struggling with the concept of assists as window dressing. Probably mostly because I WANT it not to be window dressing.
So. Do we feel there is no instructive value in assist numbers?
One thing that jumped out at me with my Whittier wooden team this year is the horribly low number of assists we had in our losses.
We have a pretty good team. And are 22-4 with rpi about 10. But the 4 losses were incredibly ugly blowouts. They were all vs good teams. I happen to think I had more talent in each case. But impartial observers would probably say the talent level was pretty even.
Anyways. In those losses I think I averaged about 5 or 6 assisteses obviously. We had fewer made fg. But the rate of asset per made fg was very low also. Coincidence? Or am I confusing cause and effect or chicken and egg?
oldave, thing is, in our time, assists were 100% window dressing, thats based on my observations and the use of the "window dressing" term by old admin. that doesn't mean they won't have some rough correlation with success - as the formula used to figure how often you got an assist did depend on some ratings, which were valuable for success. today, its still the case, but there is probably a closer correlation - because now passing actually has significantly more value. passing in the old engine was only useful for preventing the turnovers *of the player in question*. in motion it wasn't even as important as bh in that regard. that is pretty low importance, when you think of how valuable passing really is. now, passing ups team fg%, so its more important. the formula for assists is largely based on passing, so now that there is more importance of passing, there is more of a correlation between assists and success. so coincidence, not really, but also, you can't dice individual assist figures.
think of it like this. if you are ranked in the top 25, you are probably doing decently well in the projection report. the projection report doesn't include ranking as a component, but it does include some similar components. if you were to dice rankings to try to find small connections between your actions and your NT seeding, could you get somewhere? maybe, but only because it has some correlation with NT seeding. going against the projection report is much better, and often, at small intervals, small increases in ranking actually result in small decreases in projection report - easy wins are way too valuable in the rankings compared to the projection report. so, you would often be lead astray.
think of assists as the top 25, except its much, much more poorly correlated. think of projection report as the value of passing, which we have no measurement for. direct measurement, that is - you could absolutely do FG% studies to get an idea of how valuable it is. i think we are talking multiple % points between high end and low end passing teams, in terms of effect on team mates' fg%, which would make passing in your pg one of the single most important ratings on the whole team. but i haven't ever studied it nearly as closely as i studied things in the olden days, so im really not sure.
anyway, bottom line is, yeah there is some connection between assists and success, and also between assists and the passing impact on fg%. but its more coincidental, just because there are some common elements, with varying degrees of similarity in their weightings. we have far better ways of measuring the value of our players than assists. so im not sure its still 100% window dressing, meaning there is absolutely no value in it to anyone, but in effect, its still window dressing. someone who knew nothing could get somewhere looking at assists, because they have no other knowledge, and it does somehow correlate to success. but its not strong enough for coaches like you and me to do anything with. thats my opinion, anyway - in the old engine, i was 100% confident of this, but now i just don't pay enough attention to details like that, to say with 100% confidence. maybe there is something there? but id be really surprised...