We'll have to see if it continues, but right now, definitely '17. Last year was great, no doubt, but we were a little overrated. Our margin of victory was only 2.6, and our "expected" pythagorean wins were only 48 compared to 53 actual wins, indicating we were statistically lucky to win 5 extra games, making us look better than we actually were. SRS (simple rating, taking into account margin of victory and strength of schedule) was 2.25, or 8th place.
This year, our pythagorean expected wins are 9 and we have 9 wins, so we are not abnormally lucky. Our margin of victory is up to 9, and SRS is up to 7, for 3rd in the league.
So, thus far, better this year.
Out-performing the pythagorean W-L isn't necessarily about 'luck'. Teams that have very good half-court execution tend to do well in games that come down to a few possessions. I.E., 'crunch time' favors half-court execution. And the Celtics last year were extremely good in half-court / crunch-time execution because of their devastating pick & roll game (fueled by a certain little guy).
I totally agree that this year's team is significantly better than last year's. This year's roster is just so much more balanced. Last year's lineups were way too small-guard-centric. This year we are depending on and getting way more out of our bigs and our swings.
But I disagree that last year's was 'overrated' or 'lucky'. They went to 5 games in the ECF and if not for devastating injury, might have taken that series further. Their record and seeding were legit.
I've been calling for this years team to do better all along. Both before and after the IT/KI trade, I was calling for 55+ wins, threatening 60 if things all went well. I didn't see any reason to change that prediction after the trade. I admit that I hedged and revised my projection to ~54 wins after Hayward was lost, but given how fast Jaylen and Jason have developed to take up the slack, I'm back on my original 55+ projection. Barring injury, of course.
Eh, you'd expect it to be more consistent year to year if that were the case. This year we match evenly (small sample). Last year +5. 2 years ago -2.
Golden state: This year matches, last year even, 2 years ago +8.
OKC: -3/+4/-4
Memphis: 0/+1/+7
San Antonio: 0/+1/+1
Cleveland: 0/+2/0.
Generally, teams are right on, or pretty close. When there are big divergences (either positive or negative) they are not typically sustained year to year, even without major coaching or player changes.
You'd think that all teams would thus cluster around .500 in 'close games' (games decided by just a couple of possessions) if close-game out-comes were truly random.
But they don't. Last year the Celtics were 4th best in 'close' games with a .643 W/L percentage. The teams at the top of the rankings were all 'good teams': WA, HOU, UTA, BOS and the teams at the bottom were crappy teams: MIN, DEN, CHA, LAL, BKN. All those teams had terrible half-court efficiencies on both ends of the court.
https://www.teamrankings.com/nba/stat/win-pct-close-games?date=2017-06-13
The Pythagorean W/L projection does have fairly strong correlation overall, but it can be skewed if a team has just a handful of anomalous games. The Cs last year had several notable 'blow-out losses', especially early in the season (Anyone remember the DEN, WA & GSW games?). Even late in the year, they had a 16 point loss to the freaking Kings as well as the 23 point blow-out to the Cavs.
Certainly, random luck IS a factor for deviations from Pythagorean W-L. But it is not the only factor.
It's not quite that close games are 50/50, but the deviation from pythagorean wins IS mostly random. As easily demonstrated that MOST teams cluster very close in terms of actual wins vs pythagorean wins, and, as I showed you a few examples, when teams DO stray from their pythagorean win expectations, either positively nor negatively, they then tend to return to zero or even the opposite by the next year. If deviating from pythagorean wins was a skill, or depended on certain skills, then it would be repeatable year to year, but it's not.
What it shows is that if we kept the same roster as last year and everyone played the same, we probably would have won 49 or so games without actually being a "worse" team.
I'm not disagreeing that random luck is a huge (possibly primary) factor for deviations from pythagorean wins.
But if you examine year after year of team win % in "close" games, the fact is, there is (a) a wide dispersal with the top teams winning 60-80% of those games and the bottom teams winning well under 40% and (b) the top teams in winning close games tend correlate fairly tightly with the better quality teams, who tend to be able to execute better when the games slow down.
It's not a 100% correlation. But it is pretty strong. Just toggle through the years at the link I gave.
Follow the fate of a team known to be strong for period of years, like San Antonio from 2011 - 2017 and they tend to ranked near the top of the 'close game' winning % every year.
I haven't run the actual numbers to get a coefficient, but if you simply view the overall ("All Games") winning percentage ranking each year next to the "close game" winning percentage and slide through the years and you will visually see that the standings of each correlate fairly strongly.
I think we are talking about different things.
I looked back at my first post, and I never said anything specifically about close games. And yes, good teams win more close games and bad teams lose more close games than average.
But that doesn't explain the variation in expected wins vs actual wins. The Celtics "Should" have won 48, but won 53. Washington, the "best" clutch team, "should" have won 46 and won 49, so a 3 game over performance. Houston, #2 most clutch, "should" have won 55 and did win 55. Utah, #3 most clutch, Should have won 52 but only won 51.
So I don't know how closely correlated those are, but if they were closely correlated and based on a definable skill, you would expect:
- Year after year, the same team (or primary players) being higher in the "Clutch winning percentage" rankings, AND
- Year after year, those same teams outperforming their expected wins.
The fact that good teams tend to stay on the upper end of winning close games but are really inconsistent in terms of outperforming expected wins kind of tells me that the win % in close games does not correlate all that well in terms of explaining why a team outperforms the expected wins.
I guess, said another way, if we could hypothetically play last season over again with the same rosters, I would be more surprised if they won 53 than if they won 48 games. Doesn't change anything about how fun last year was, but is valuable info in terms of whether they should have brought back the same roster or tried to change it up like they did.
Topic: Who's better...'16 or '17 Celtics? (Read 6250 times)