Not sure I buy it, as well as the Warriors played in the first half tonight and Denver in a fourth-quarter comeback.
Try the same exercise with the 2004-05 Nuggets. They went 25-4 after the All-Star break -- a much longer stretch with an equally reasonable explanation, George Karl. I don't think there was anything fluky in that team's point differential. Still, Denver lost in five games to San Antonio in the first round.
One series doesn't necessarily tell us anything, but I can tell you over a larger sample size (11 years worth) that April winning percentage has virtually no correlation with playoff success, once we account for overall ability and seed.
I don't think April winning percentage is what Hollinger's trying to get at. It's more like "full team that will be on the court in the playoffs" winning percentage. Which makes sense - season-long point differential is highly impacted by injuries and trades, and it I don't see how it would hurt to try to control for these somehow. As far as the lack of correlation between April winning percentages and playoff success, I imagine tanking/coasting after clinching could skew that.
JQ, everything you say is perfectly logical, but at some point "trying to control" for injuries and trades dissolves into cherry-picking.
I can understand throwing out teams that had poor or mediocre Aprils after clinching, but the tanking/coasting aspects shouldn't affect teams with surprisingly good April records any more than they did Denver and Golden State down the stretch.
So let's take a look at how some of these teams did. I looked at the difference between winning percentage in April and overall winning percentage.
For these teams, I've got their actual "playoff score" -- 3 points for making the playoffs, +1 for a win, -1 for a loss, +4 for each series win, adjusted to five-game series for seven-game first rounds since 2003 (that's a mouthful) -- and their predicted playoff score, based on a regression combining their record, differential and seed.
Code:
Yr Team Pro. Score
---------------------------
06 Chicago 0.0 2
03 Milwaukee 0.0 2
04 San Antonio 12.3 8
01 Indiana 0.0 1
02 San Antonio 13.0 5
04 Minnesota 13.7 12
06 Memphis 6.0 0
05 New Jersey 0.0 0
02 Boston 7.2 13
02 Toronto 0.0 2
Some of the teams that snuck into the postseason tended to perform better than expected, bout overall there's not a real trend.
In fairness, there's not a lot of major injuries/midseason trades in this group. I notice 2002-03 Milwaukee, which added Gary Payton, and the 2001-02 Celtics, who traded for Rodney Rogers and Joe Johnson. Boston did do significantly better than expected.
I really don't believe any team's chances were as low as four or five percent entering a series. Not even the Bullets. The 59% for the Warriors is a little closer to the historical norm for teams that win Game One on the road, but you would have to figure the disparity between Dallas and Golden State would lower that to some degree.
Really? Even with only two No. 8 seeds winning since it went best-of-five, and none since it went best-of-seven? I can think of several playoff series where, barring injury, the underdog had virtually no chance -- the '96 Bulls-Heat series topping the list, of course.
On a wider topic, this is exactly the debate I was having with myself while I wrote the story -- which is more valid, full season stats or the more recent #s. I normally tend to go with the former, but in these two particularly cases I'm thinking the latter are much more relevant.
Joined: 18 Feb 2005 Posts: 123 Location: Cambridge, MA
Posted: Fri Apr 27, 2007 12:55 pm Post subject:
John Hollinger wrote:
which is more valid, full season stats or the more recent #s. I normally tend to go with the former, but in these two particularly cases I'm thinking the latter are much more relevant.
I have to agree that it's an entirely circumstantial question. The sample size must be just right. It's important to remember that there's such a thing as a sample size being too large.
For instance, how enlightening is it when broadcasters tell the audience what one team's record is against another team over the course of a decade or more?
Really? Even with only two No. 8 seeds winning since it went best-of-five, and none since it went best-of-seven? I can think of several playoff series where, barring injury, the underdog had virtually no chance -- the '96 Bulls-Heat series topping the list, of course.
Okay, that's fair. Using Dean's Playoff Home Court Advantage calculation, the Bulls would have been expected to beat the Heat 99.7% of the time in a seven game series. I would figure that there would be some regression to the mean in there, but that's still impressive.
Using the above calculation, the Warriors chance to win the series when the series started was 2.4%. After winning Game One it jumped to 11.5%, which is way more reasonable. After Game Two it fell to 9.2%.
My 2 cents. Out of curiosity, I've estimated the plus-minus of some of the playoffs starting lineups as the addition of the individual adjusted plus-minus. This somehow corrects the effect of injuries throughout the season. It's far from a perfect metric, as, for instance, I use Iverson data from Philly, or don't take into account players or units getting into rhythm after an injury or trade, plus all the flaws of adjusted plus-minus. The data:
Mavericks
Terry - Harris - Howard - Nowitzki - Dampier
+20.6
Warriors
Davis - Ellis - Richardson - Jackson - Harrington
+11.7
Those are two strong lineups. But Dallas should still be the clear favorite, shouldn't they? Well, not so fast: if I insert Biedrins in place of Harrington, the plus-minus for Golden State goes up to +20.0, and we have a virtual tie. The Mavs can also get a better rating (+21.9) by replacing Dampier with George, as they did in Game 1. Their small lineup with Nowitzki as the only big have been quite successful exploiting matchups the entire season (Net48 of +13.9). I'm sure they are aware of this, which could explain why they tried that approach in Game 1. Given the results, I'm not sure we'll see the same starting lineup again in these series.
Do we have a similar scenario with the Spurs and the Nuggets?
Spurs
Parker - Ginobili - Bowen - Duncan - Elson
+18.9 (+5.0 with Finley instead of Manu)
If we trust these adjusted numbers, the problem for San Antonio is that there's a big drop from Ginobili and Duncan to their backups, while the bench from Denver can keep up with the starters. In fact, their best theoretical unit would have Kleiza instead of Blake (+13.1), which happens to be their 3rd most common lineup.
The last matchup I've studied is Bulls vs Heat
Bulls
Hinrich - Gordon - Deng - Brown - Wallace
+7.3
Heat
Williams - Wade - Jones - Haslem - O'Neal
+14.8
Conventional wisdom agrees here with advanced stats: the champions were the favorites, at least if we ignored Wade health issues and the Bulls superior bench. However, Chicago just needed Wallace to be his usual self, not a -10.4 player, to be above Miami.
Joined: 09 Aug 2006 Posts: 87 Location: where you aren't
Posted: Sat Apr 28, 2007 12:04 pm Post subject:
Ryan J. Parker wrote:
We all know shooting 3s is the best underdog strategy, but over a 7 game series odds are low the Warriors shoot well enough to win.
ironically, the Warriors have been attrocious on threes (27% from beyond the arc for the series) thus far and are still up 2 games to 1. _________________ I'm so sick I be terminally ill
Joined: 09 Aug 2006 Posts: 87 Location: where you aren't
Posted: Sun Apr 29, 2007 12:50 pm Post subject:
Ryan J. Parker wrote:
tHe_pEsTiLeNcE wrote:
Ryan J. Parker wrote:
We all know shooting 3s is the best underdog strategy, but over a 7 game series odds are low the Warriors shoot well enough to win.
ironically, the Warriors have been attrocious on threes (27% from beyond the arc for the series) thus far and are still up 2 games to 1.
When you shoot 73 of 'em (compared to Dallas' 50) you don't have to hit as many.
heh, I don't think hitting 27% on threes is as effective as shooting 46% from the field, seeing as how 27% on threes drags down their effective fg%.
Of course the field goal defense is a lot more responsible for the sucess. The Mavericks, based on experimental probability (reg season) applied to a simple binomial distribution, had something like a .17% chance of shooting 38% in two out of three games. _________________ I'm so sick I be terminally ill
Joined: 09 Aug 2006 Posts: 87 Location: where you aren't
Posted: Mon Apr 30, 2007 2:12 am Post subject:
whether or not the Warriors win this series, I got to get some props for postulating this before the series started and being correct. _________________ I'm so sick I be terminally ill
However, I think their loss to GSW exposed their achilles heel. They aren't able to run and gun the way they used to. In fact, they play at the 4th slowest pace in the entire league. It used to be that the question was "Does Dallas have enough D to win in the playoffs?" (meaning they play at such a quick pace, without too much regard to defense and ball control). Now they do, but the pendulum has swung in the other direction. Other teams are better able to adapt to different paces, but they haven't been able to. Also, they don't force TOs very well, and they send their opponents to the line too much.
So far this series the Warriors have been content to let Dallas grab a truly heroic percentage of offensive rebounds, while (in the three games they have won) they have directly attacked the Mavs' weaknesses that I mentioned: getting to the line early and often, and keeping their TOr down.
It should also be noted that they've been shooting their brains out.
Lastly, over the regular season Dallas played at an average pace of 88.2 possessions per game, while Golden State played at an average of 97.6 possessions per game. Here is are the number of possessions in each game so far this series in order (thanks to cherokee for the awesome Greasemonkey plug-in again): 89, 100, 95, 91.
What's odd is that you'd expect Golden State's Eff to be higher in the faster paced games, and Dallas to have a higher Eff in the slower paced games. However, the opposite is in fact turning out to be the case.
Code:
Game Pace DAL Eff GSW Eff
1 89 94.4 109
2 100 114.3 98
3 95 97.8 112.4
4 91 106.5 117
Besides noting that it's obviously a very small sample size, I'm genuinely flummoxed by this.
Joined: 09 Aug 2006 Posts: 87 Location: where you aren't
Posted: Mon Apr 30, 2007 9:25 am Post subject:
by the way, does anybody have a clue what the hell's with Al Harrington? This has got to be one of the biggest jobs of freezing up in the playoffs I've ever seen; 18.5% from the field (46% reg season), 0% from 3 point land (#8 in the league reg season), 60% fts (70% in regular season, only one which is in the realm of standard deviation)
Quote:
I think a lot of us deserve those props. Of note, I wrote the following in this forum on March 29th:
yeah, I'm more content with the "props" of 17 grand that I'll be getting from betting on the warriors to win the series. _________________ I'm so sick I be terminally ill
All times are GMT - 5 Hours Goto page Previous1, 2, 3, 4Next
Page 3 of 4
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum