Yes. My criticism of the Olynyk pick is that they should have been looking for the moonshot, but opted not to. Once you factor in the likelihood that 18 year old Greek Freak who no one knew much about could have been a bust, Olynyk might actually have rated more highly.
Sorry in advance if I am not understanding you correctly here, but how are these keys any different than John Madden saying pregame that the Patriots would do well to score more points than they allow? Or by keys do you mean these are your criteria for evaluating the various decision makers in their ability to draft players?
Sorry, I meant those to be pretty obvious, but to provide several different categories for evaluating Ainge. I don't think by any of those categories he's been particularly successful. The main argument in his favor is basically that where he's been picking, expectations are pretty low, which is what the analysis was meant to quantify and evaluate. And even based on the relatively low expectations of his picks, he's graded out below average.
My point is that the distinction is somewhat meaningless. Evaluating young talent is the important dimension of the job that we should be talking about.
Yeah, unless you are constantly missing in the top half of the lottery which Ainge isn't doing. If Jaylen and the 2 Nets pick flop, it changes the narrative a lot.
I think you need to run the analysis for each GM to determine if Ainge grades out "below average". Your methodology seems to overweight the misses at the ass end of the first round: JR Giddens (30th pick) and JaJuan Johnson (pick #27). Those late picks are always crap shoots, and I don't see why those misses should be given so much weight relative to, say, the drafting of Marcus Smart.
If Ainge drafts Markelle Fultz and DeAndre Ayton and they both flop, I am not sure that changes the narrative because every single GM would have drafted those two
Yeah, for sure. In fact, he grades out below the mean value for what you'd expect to get by drafting at those spots. But it's quite possible/likely that only a couple GMs are above the mean, but are above it by a lot, so that most GMs are a bit below.
Powe was 2006, but yeah, those were big hits. Davis is +11.2 and Powe is +9.7 by this method, which would be his top two gets. In fact, including them pushes his overall effort to +8.3 WS above expected.
Right, so the only problem with doing this is that to get an accurate estimate of the median, you need to combine data across a lot of years, which is a problem because you either end up including really old data, or including new data where players haven't had as much time to accumulate WS. So I just tried this, using 2006-2014, and scaling each year to have the same total WS. This adds some noise to recent years since you're essentially multiplying noisy data by a big factor (e.g., for 2014, it's 5.7*WS, 2013 is 2.5*WS, and after that it gets closer to 1), but it's probably close-ish. Then you can get a median value per pick, and after the first few you can even smooth a bit (i.e., median value for pick 7 based on picks 6-8).
This is more of a reason to use a regression of some kind as a model, not a reason to use either the median or the mean. Eight wins is "wrong", as it's overly influenced by a couple big hits who happened to be drafted 43rd, but 0 wins is probably also wrong, cause sometimes there is talent there.
I'll bet the curve's shape is similar for picks in the 20s, with the average pick producing significant value but the median pick yielding next to nothing.Any expected value for draft picks should use median values and distributions as opposed to mean values.
Using the 43rd pick as an example (Michael Redd, Trevor Ariza, etc.)
Mean expected value = 8 win shares
Median expected value = nothing
You could approach this problem all sorts of ways. I would argue that just compiling data using multiple drafts are problematic because the talent pool is so variable. But for a back of the envelope calculation the median gives you a whole lot of information. For almost no extra work you could throw in the upper and bottom quartiles of the range to get a box-and-whisker plot.
I think this is true. The NBA has also implemented a number of rule and/or CBA changes to ensure that this trend continues.
I've seen this criticism pop up regularly over the years, so I think it bears addressing. At the time of that draft, I shared your view that the Celtics should be aiming for moonshots rather than solid value picks that lacked super star upside. (Although, I've always liked Olynyk, so I was perfectly happy with this particular pick, but that's not really relevant to my commentary here.) But, events since then should push us to reevaluate whether we were correct. Because, I can't speak for you, but my belief that we needed a moonshot type pick was premised on the belief that the Celtics best case for the next several years was being a mediocrity, fighting to get into the back of the playoffs. As such, good solid players weren't what we needed, we needed a new star or two to lead the team.
Disagree, because - as good as they are right now - the Celtics aren't a legit title contender.
Giannis is big enough and mobil enough to play the 4 next to Horford, he'd go a long way towards making life (even more) miserable for Boston's opponents.
