But FWIW Pelton had us at 49 last year which means we outperformed his projections for four consecutive years. 43 seems low to me unless we get injured or Ibaka really shits the bed.
3 current Raptor PG's in top 10 DRPM-only.
1 former one at #66 - CoJo. One spot ahead of Kyrie.
1 former one at #66 - CoJo. One spot ahead of Kyrie.
CoJo had a hell of a down year. Part of why I don't think we'll miss him much, at least relative to last season. We already missed the CoJo that played the year before (and made a lovely, if too brief, appearance in that end of season run while Lowry was hurt).
The good teams (win both at home and on the road) are Raps, Cleveland, Boston, Washington, and maybe the Bucks.
The average teams (win at home but lose lots on the road)are comprised of Miami, Detroit, Charlotte, maybe Philly if they get some health luck.
The crappy teams (lose at home and on the road) are looking like Atlanta, Chicago, Nets, Knicks, Orlando and Indy.
If the Raps
A. play 500 ball against the good teams in conference plus one they can get 9 wins ..
B. Play 600 ball against the average on any given Sunday teams they should rack up 10 wins..
C. Play 700 ball against the east's lowest of the low that's 17 wins.
D. Play 500 plus 1 against the West conf that's 16 wins.
That's 52...
maybe they don't do as well against the west and it flips to 14/16 ..which would be 50 wins.
First off, while I've never been much of a Blake Griffin fan (I think he's overrated), I think Clipper Nation are going to be pleasantly surprised by Patrick Beverley's ability to run a team and play big minutes. Obviously he's no Chris Paul but he's a real time player.
I read that article by Kevin Pelton and felt very underwhelmed. Just in his write-ups alone there are so many outliers that all he proves is that RPM is no good for predicting wins, at least not alone. So many times he would say "this team won XX games, well below/above what RPM they finished with". I think Boston was 5 wins above, and Minnesota 7 games below? That's a horrible margin for error.. I'm not saying they should have a crystal ball, but with all the resources ESPN guys have to work with, all I can say is damn amazing work by @DanH.
First off, while I've never been much of a Blake Griffin fan (I think he's overrated), I think Clipper Nation are going to be pleasantly surprised by Patrick Beverley's ability to run a team and play big minutes. Obviously he's no Chris Paul but he's a real time player.
I read that article by Kevin Pelton and felt very underwhelmed. Just in his write-ups alone there are so many outliers that all he proves is that RPM is no good for predicting wins, at least not alone. So many times he would say "this team won XX games, well below/above what RPM they finished with". I think Boston was 5 wins above, and Minnesota 7 games below? That's a horrible margin for error.. I'm not saying they should have a crystal ball, but with all the resources ESPN guys have to work with, all I can say is damn amazing work by @DanH.
Thanks!
All I'll say to this is that ESPN does a lot of complicated stuff to try to predict player impact - they use a special formula using multiple years of player's minutes patterns to guess at their upcoming ones, they use aging curves to project improvements and declines in players at various stages of their careers, they use projection metrics for incoming rookies based on historical data - and all of it, if you look at the statistical reliability, is essentially like throwing crap against the wall and seeing what sticks. It's not an issue with RPM per se - my model using RPM has the Raptors a near-50 win team. It's all the other stuff they do to try to predict what will change from last year to this one. My approach is simpler in that I only fix a team's total minutes based on added or removed players - otherwise, I assume no change at all from the prior season. It's inherently wrong. The intentions of the ESPN projection system are more honest - they are trying to account for actual changes that, on average, tend to be born out. But in specific cases (like Lowry of late, the main reason they keep underpredicting the Raptors - he's on the wrong end of the age curve but just keeps getting better), their adjustments go in entirely the wrong direction compared to reality and it throws the whole system off.
While in my case, I just assume all those adjustments will wash out for each team - which is inherently dishonest, that won't happen, but it's also more stable, and the variance from a simple assumption like that to reality might be a little greater in the aggregate, but there will be fewer outlier cases where the projections are wildly off. Obviously there are cases of players taking unpredictable leaps, or having injuries you wouldn't expect, or just a team suddenly gelling (or not) and going way off track, or trades, that no approach can account for. But by trying to account for as many trees as possible, it seems like some of these projections like ESPN's attempts, tend to miss out on the forest.
I think another thing that often underestimates the Raptors is DeRozan. I don't know if it necessarily factors into RPM specifically, but DeRozan continues to defy most advanced metrics (and to be honest, myself personally) by playing in a such way that shouldn't work but does.
Or, perhaps making adjustments for aging etc is correct, Kevin's estimates are actually superior and the results over a tiny sample are meaningless.
I have the Raptors most likely win total continuing to slide downward with the confidence relatively high. Key players are past their peaks, younger talent is very marginal and really bad off-season moves
The Raptors have been a good team for a few years, but I don't think management has a clue why. They're committed for two or three years, but then it's time for a reboot from top down
Let's not underrate regular season. I'm starting to see this more and more. People acting like regular season ain't shit and 30 win teams are out there going to finals or some shit. Going into playoffs with home court advantage has been vital for this team.
Comment