Last I saw he was projected around 30ish, right around where he was in B150. Are you looking at projections in alphabetical order? You have to translate the #s into z scores to turn them into approx rankings. He's around 30, don't worry.

Thanks for replying! I just checked again "Premium: Rest-of-Season Total Value, 9-cat". Paul Millssap has a 9cat value of "-0.221" which is ranked at 114.

Yes the projections are messed up. They've become current averages. I posted about it in website help forum and here, and tweeted them. Haven't heard back yet.

Hey guys, we had to clear out our database due to the issue from three weeks ago and that reset some of our formulas. Panda is pushing out a fix later tonight and you'll see some changes.

Also, just since we're chatting on this, projections are an interesting task in-season. When I create the B150 it takes a full three months to get in there and vet every number. In-season projections are impossible to handle that way, obviously, so you do have to take a two-pronged approach using algorithms and hand-crafted approaches. We recognize the inherent difficulty in this and our goal is to have the best projections, but getting to a point where I'm confident as I am with the B150 is a major piece of research for us (that is actually underway).

We're going to actually write about these projections from the perspective that they are somewhat flawed, or even seriously flawed if a situation is zig-zagging all over the place. So you'll see that all unfold in the coming days and weeks.

Interesting. I sort of do some of this stuff myself so I know the juggling act, but would rather see projections weighted more heavily to the fwd projections you put all the work into, rather than current averages. Certainly least this early in the season.

Of course. Without delving into specifics, they're heavily weighted toward forward-looking projections from the B150.

Sounds great

Still finding them unusable today. The percentages are either just this season's %s, or very close to it. I don't see the use of having projections where Durant is only an 80% FT shooter, Giannis shoots 62% from the field, Porter shoots 55% from the field, Westbrook 66% from the line, etc etc.

I know you're working on these, and I apologise for just sounding like a whiner, but I really think they need to be steered more towards projections / career averages / last season averages etc versus averages from the first 6 games of the season, if they're to be of use to us.

Thanks

I talked with Panda and he pushed out the changes about two hours after you wrote this. Thanks for the patience while we had the changeover. AB