I thought it might be interesting to gauge our team's prospects this season relative to the competition using ESPN's #NBARank player rankings. The rankings were based on expert's projections on how much the player would contribute in the upcoming season. Other factors which aren't strictly about contribution to winning were no doubt considered in these rankings, but the potential advantage is that they can incorporate a subjective feel for how players are trending in their respective careers. So, I gathered all of these rankings for all teams. The next step for me was to translate these rankings into something resembling win shares. To do this, I took a couple seasons from the recent past (09/10 and 10/11), and I plotted the distribution of win shares for all players in each season, ordered by their win share total (I used the WS metric at basketball-reference.com). It turns out that this distribution of win shares for 09/10 and 10/11 was almost identical, leading me to think that this is fairly stable across seasons. The plot looked like this: So, I used the above curve to translate #NBARank to projected win shares for each player in the upcoming NBA season. The results are summarized in the following table: TEAM CNT Top12 Wins 6 through 12 3 through 5 Top 2 ---- --- ---------- ------------ ----------- ----- MIA 16 56.9 12.6 18.2 26.0 BRK 15 54.5 16.5 20.3 17.8 LAC 14 53.8 13.8 15.6 24.4 GSW 14 52.8 13.0 19.2 20.7 <font style="background:yellow;color:black">HOU 19 52.4 10.3 17.2 25.0</font> CHI 13 52.4 12.8 19.2 20.4 IND 12 51.7 13.6 18.5 19.7 SAS 14 51.7 13.7 17.5 20.5 MEM 15 50.4 14.2 17.3 18.9 OKC 16 48.2 6.3 13.8 28.2 NYK 14 47.6 16.3 13.8 17.6 MIN 15 47.3 14.3 15.6 17.4 DEN 15 47.3 17.7 15.4 14.2 CLE 13 46.8 15.2 14.0 17.6 POR 16 43.8 12.1 13.9 17.8 NOP 15 40.9 7.9 18.1 14.9 DET 14 39.5 9.2 15.5 14.8 BOS 15 39.1 13.9 11.4 13.8 WAS 15 38.2 10.5 12.3 15.4 MIL 16 38.0 14.1 11.9 12.0 SAC 14 37.9 15.8 10.8 11.3 ATL 14 37.8 7.2 14.2 16.4 DAL 16 36.7 10.8 11.8 14.1 TOR 14 36.0 10.7 13.2 12.2 LAL 14 32.4 5.4 10.0 17.0 ORL 15 31.3 10.3 11.7 9.2 CHA 13 30.1 7.1 10.9 12.1 UTA 14 26.3 5.8 9.3 11.3 PHX 16 23.7 4.8 7.4 11.5 PHI 14 18.6 2.7 7.2 8.7 You can see that I've highlighted the numbers for the Rockets in the table. The CNT column says how many players on our roster were ranked by #NBARank. Interestingly, we lead in this category by a wide margin -- 19 of our players were ranked, and no other team has more than 16. Of course, several of these players will be cut before the regular season. Since I'm only interested in players that are likely to play, I tabulated the wins for the top 12 players on each roster, shown in Top12 Wins column. This is roughly a projected win total for the team, based on ESPN's #NBARank player results. I've also ordered each player on every team, and gave the total win shares for players ranked 6 through 12, ranked 3 through 5, and the top 2. This gives an idea of how much a team depends on stars versus depth. These numbers show that only Miami is projected to garner more win shares from their top 5 players than us. To further illustrate this, here's a chart: The final standings predicted by these projections were: Conf TEAM Top12-WS Conf TEAM Top12-WS 1 E MIA 56.9 W LAC 53.8 2 E BRK 54.5 W GSW 52.8 3 E CHI 52.4 <font style="background:yellow;color:black">W HOU 52.4</font> 4 E IND 51.7 W SAS 51.7 5 E NYK 47.6 W MEM 50.4 6 E CLE 46.8 W OKC 48.2 7 E DET 39.5 W MIN 47.3 8 E BOS 39.1 W DEN 47.3 9 E WAS 38.2 W POR 43.8 10 E MIL 38.0 W NOP 40.9 11 E ATL 37.8 W SAC 37.9 12 E TOR 36.0 W DAL 36.7 13 E ORL 31.3 W LAL 32.4 14 E CHA 30.1 W UTA 26.3 15 E PHI 18.6 W PHX 23.7 OKC was surprisingly low, and it would have been even lower (an 8th seed) if I had used Westbrook's updated ranking (downgraded to #23 instead of #5 due to recent injury set-back). You can see from the chart above that this is due to a projected lack of depth for them. There's little contribution after the first 3 players or so. Still, I don't expect OKC to be this bad, but we'll see.
This is cool interesting to me besides okc is how well Boston seems to be ranked compared to a team like the hawks
Is it weighted by playing time? Or is that factored in already? Because, i would weight it for top 5 at 75% of total wins due to their greater impact on the floor. DD
No extra weighting was applied by me. I think that's already taken care of by the player raking combined with the use of the win share distribution.
It appears to be factored in, but not perfectly. WSs are a cumulative stat so it is factored in assuming that the current No. X guy gets as many minutes as guys previously ranked No. X so they accumulate the same WS total. A deep team may get a little extra bump as their first man out of rotation may actually be ranked high enough that previous guys at his ranking were rotation players. The starters may get a little less WSs than predicted, but the bench would get more. However as can be seen with OKC, WS typically are concentrated with the best players so I would doubt this type of error would have a huge impact. Without looking at the data, I would expect a team like Milwaukee may be inflated as they have at least a couple of non-rotation guys almost as good as their starters.
This is interesting. Can we just add player winshares to get team win share, or do we need some tweaks to adjust for system/fit/roles?
In theory, player win shares adds up to total team wins. The assumption made in the OP is that #NBARank player rankings is a close approximation of what the player rankings by total Win Shares will be this season.
Using Durvasa's graph and doing a rough correlation... Rondo predicted to have more WSs than he has had in each of the last 3 years even if you assume he is healthy. Green, it has been two years since he has been that good according to WS..a.k.a when he played in OKC. Bradley, his WSs are practically double his best season. Wallace is about right...may be a tad low actually in WS. Bass appears to be underated by about a WS. Humphries is a little higher than last years performance, but lower than previous years. Sullinger appears a tad low of what he did last year. Olynyk...tough to say what his WS will be. Lee, Crawford, Brooks, and Bogans all appear to have higher WSs than they will get from a sheer minute standpoint. As ESPN seems to be predicting a career year from Bradley, Rondo playing like he did when Pierce and Garnett were tearing it up, and Green "playing" as well as he did in OKC, as well as Lee, Crawford, Brooks, and Bogans not likely playing enough minutes to accumulate the WS needed for their ranking's correlated WS, I think ESPN rankings may be too favorable for the Celtics. I do expect Green to put up a good WS as WS accounts for the things he does well. I expect Rondo will likely be healthier, but without Garnett and Pierce taking attention away from him....his Thomas Robinsonian scoring efficiency may get lower. Bradley...without Garnett making his defense look even better and having to share PG minutes with Rondo, I am not expecting a career year in WS. WS, imo, does a decent job on offense, but is not the best on defense. According to WS, Rondo is about 60% better than Asik defensively on a per minute basis. This does not make sense to me unless Rondo's defense is being severely overrated based on playing with Garnett. Rondo's defensive WSs is the most overrated that I have seen when comparing it to RAPM. P.S. I tend to value Rondo a lot less than most people who don't value adjusted +/- stats. I tend to value Asik a lot more than people who don't value adjusted +/- stats. P.S. Take Two. I think if ESPN had done what Durvasa did, they would have downgraded some of the Celtics Rankings (but not Rondo's as he is a "star") a bit to get them below Atlanta's ranking.
It did..but the Rule of Sixteen, brought it back to 82. I forgot to include the Rule of Sixteen in the original parameters of my formula, but a night of hard drinking provided an epiphany. So we're stuck with only 82 wins.
Very interesting... Thanks Durvasa. Based on player quality alone, this looks very accurate. I love the breakdown of the best W/S's when looking at your 1-2 punch's. Three teams really stand out here in Miami, OKC, and Houston. Also looks like some teams like the Clippers, Houston, and OKC have some major week links in their best 3 through 5 man lineup. The one factor that I still have to consider though is system and coaching that has to have a minor impact in wins at least in the regular season. Chicago and San Antonio seem like teams that could skew this ranking a little bit. I'm also watching Minnesota as well. With Adelman at the helm, factoring in player quality, and most importantly health, they could win 2 or 3 games they shouldn't win in the regular season based on how good of a coach they have and their system punishing teams.
It looks like there is just very little lack of weakness in their best 3 to 5 players that is making them overall jump up. Other than Brooklyn (on paper), no other team has as much quality with all 5 of their starters. Every other contending team has some type of weakness in their starting unit.
This is kinda cool. Am I right that it doesn't account for overlapping roles of players resulting in reduced win share (e.g. Asik/Howard)? Also doesn't account for any synergy between star players (e.g. C/SG combination ought to be a pretty strong plus).
The rankings seem to reflect the general opinion quite a bit. Wolves are a lot higher than I would imagine though.