Wins, Losses, and Seeds, oh my!

by joelor on Sep 19, 2008 3:01 PM PDT

Someone posted something interesting in the last week or so about the total number of wins in the regular season by playoff teams from the Western conference. It turns out this number was 434 last season, much higher than the 418 from 2006-2007 or the 406 from 2005-2006. This got me to thinking about a historical analysis of wins among the 1-8 seeds in the Western conference.

Since we've been speculating a lot about what the Blazers playoffs hopes are like, whether we can make the playoffs, and what it will take, I thought I'd look at a pure-numbers analysis of playoff team history in the WC.

Disclaimer: This is based purely on numbers and trends from previous years, and does not take into account qualitative circumstances surrounding the conference.

So I looked at # of wins by seed in every NBA season since 2000-2001. Here is a quick chart:

Seed	2000-2001	2001-2002	2002-2003	2003-2004	2004-2005	2005-2006	2006-2007	2007-2008
1	58	61	60	58	62	63	67	57
2	56	58	59	56	59	54	61	56
3	55	58	60	57	52	44	58	56
4	53	57	51	55	58	60	51	54
5	53	50	50	52	51	49	52	55
6	51	49	50	50	50	47	45	55
7	50	45	47	45	49	45	42	51
8	47	44	44	43	45	44	42	50

I would have put averages at the end of that, but there is limited width to the post so here:

Seed										Avg Wins
1										60.75
2										57.375
3										55
4										54.875
5										51.5
6										49.625
7										46.75
8										44.875

Ok, wow the formatting is getting crazy here, but I'm going to try to insert a graph, we'll see how this works:

Ok, that did not work at all, and now the formatting is even more crazy than before because my blinky cursor is no longer showing up.

Anyway, the point of these numbers is this. Statistically, last season in the Western Conference was an ANOMALY. Evidence:

1) The 8th seed in 07-08 won 50 games, yet only once in the previous 7 years has the 8th seed won more than 45 games.

2)The top seed won fewer games than the top seed has won in any of the seasons shown here.

3)The #1and #8 seeds were divided by a mere 7 games. The average number of wins separating the #1 and #8 seeds since 2000 is almost 16.

This quantitative evidence suggests that there is no way that it will take 50 wins to get into the playoffs again, on the contrary it suggests that, with a season similar to last year, the Blazers could make the playoffs with the 8th seed. After listening to many posters talk predictions, I think a moderate average estimate for number of wins is 49. In a given year, 49 wins will net the 6-7th seed, more likely the 6th (because the 07-08 win total of 55 wins for the 6th seed skews the average). In another currently popular post, usmcr3049 predicts 55 wins this coming season. This estimate will give the Blazers the third seed in the West in an average year.

Ultimately, my point is this: I think predictions of 55 wins is toooooo high. I think predictions of the 8th seed in the West is tooooo low.

My prediction: 50 wins, 6th seed.

Word.

PS if anyone has any formatting ideas about how to import some data/graphs/charts better I could do some much more interesting analysis (not to mention much purtier). Maybe even dust off the old E-views, do a little econometrics? Some actual p-values and confidence intervals would make this all much more defendable.

Tweet Comment 29 comments | 6 recs |

Do you like this story?

Comments

I have nothing to add, except I'm first

"Screw the chalupa – it’s the victory I want." --timg56

by MiledAnimal on Sep 19, 2008 2:49 PM PDT reply actions

I'm not sure where you got your win totals from. Just a quick look at your 8 seed win totals, and I see they are inaccurate.

I posted the total wins for the 8th seed in the Western Conference going back as far as you did recently in the post you were referencing. You numbers are not correct.

The Dude abides

by noaher on Sep 19, 2008 2:55 PM PDT reply actions

Here's the link you were referencing.

Here#

The Dude abides

by noaher on Sep 19, 2008 2:58 PM PDT up reply actions

Good catch

For some reason it was screwing up the 8th seed when I was posting it from ESPN to excel…Got it fixed and though it removed a little emphasis, my points remain the same.

"Jerryd is straight ahead at you. Rudy dips around. Jerryd is a rock. Rudy is the wind. Jerryd loves the ball in his hands. Rudy moves without it. Jerryd defends by getting up in your grill. Rudy plays the spaces in between. Jerryd has focused vision. Rudy sees all around him. Jerryd likes to score off of the dribble. Rudy can catch and shoot. Jerryd is aggressive. Rudy is sneaky. Jerryd will hit you hard. Rudy will annoy you until you hit him." -Dave

Word.

by joelor on Sep 19, 2008 3:02 PM PDT up reply actions

I neglected to note that I happen to agree with your point 100%

I was in the 48 W camp (is there a 48 win camp?) a month ago. Rudy’s play at the Olympics and the Blazer press’ drooling over Bayless and Oden has me willing to concede that we could just as well win 50.

The Dude abides

by noaher on Sep 19, 2008 3:12 PM PDT up reply actions

I tend to agree with your idea

The top teams seem to have gotten better, (lakers got Bynum back, Nornets pick up Posey), while the bottom end playoff teams seem to have taken possible steps back, (denver, Phx, Dallas) without upgrading much over the summer. So your idea that last season was more of an fluke, and that this year it will return to the norm is possible. When I was researching my post, I realistically thought only 2 teams in the west might approach 60 wins, the Lakers and the Hornets, but there are a bunch of team who could reach 50, (pds, utah, spurs, houston, even Phx and Dallas). But there is no way all of them will reach 50, so someone is going to have to fall for portland to do it. Obviously by my point i believe Denver, Dallas and Phx will all fall short of their hopes this year, and give portland the opportunity to get 55 wins.

by usmcr3049 on Sep 19, 2008 2:55 PM PDT reply actions

I got my fingers crossed!

Word.

by joelor on Sep 19, 2008 3:03 PM PDT up reply actions

Third and fourth seeds, and the new seeding rules.

For most of that period, the third seed went to the division champ with the worst record,
and the fourth seed went to the winningest wild card team.
That changed for the past two year, (look at that 60-win Spurs team in 2006 for the obvious trigger),
and now the top four seeds are in order of record.

The averages show the residual effect of the now-outdated “best wild card = 4th seed” policy;
see how close that 54.875 average for the 4-seed is to the 3-seed’s 55-win average?
That reflected what it used to take to get those respective seeds, which no longer applies.

So! – here are the averages re-done, showing what seeds the given win records would’ve gotten
in ALL of those years, had the current rules been in place for the whole sample,
which should give a more accurate win-seed relationship for the middle of the pack.
(Interestingly, the 2-seed is affected more than the 4-seed; look at 02-02 and 03-04,
and you see that the 2 & 3 would’ve flipped places under the current rules.)

Averages:

1-seed: 60.75
2-seed: 58.375 (a full win higher than the old-rules average)
3-seed: 56.75 (1.75, or nearly 2, wins higher!)<
4-seed: 52.125 (over 2 1/2 wins lower!) (!)
5-seed: 51.5
6-seed: 49.625
7-seed: 46.75
8-seed: 44.875

The bottom four seeds, plus the top seed, are unaffected, of course.

"Mister Oden is a very, very big human being." - Jerryd Bayless

by QualityPie on Sep 19, 2008 3:36 PM PDT reply actions

Cool, good to know

but technically this doesn’t affect my analysis, as I don’t think anyone expects the blazers in the 2-4 seed range. I actually didn’t know this though, so thanks.

Word.

by joelor on Sep 19, 2008 3:49 PM PDT up reply actions

Well, see the follow-ups below.

2-4 is not a “seed range”.
Your analysis (which just plain rocks) applies to the question of that 50-wins-ish range of the 4 thru 6 seeds.

"Mister Oden is a very, very big human being." - Jerryd Bayless

by QualityPie on Sep 19, 2008 4:18 PM PDT up reply actions

Hmmm. Just looking at it, the seeds tier up in blocks.

Here’s how the gaps break down now, where the unbracketed numbers are the win totals,
and a bracketed number is the difference of the win totals it’s between:

60.75 (2.375) 58.375 (1.625) 56.75 (4.625) 52.125 (0.625) 51.5 (1.875) 49.625 (2.875) 46.75 (1.875) 44.875

There’s a HUGE gap between 3 & 4, and the tiniest between 4 & 5; really, a 4-5 matchup is an even contest now.
The big deal in the league now is to be a top-3 seed.

And after that, the way that 6-7 difference is a full win wider than the 5-6 or 7-8 spreads tells me these are the tiers:

Top 3
4-6
7&8 (and quite possibly a team or two who just missed the playoffs altogether)

So regarding the Blazers, I guess the assumption is that we should be in that 7-8 area,
and the question is, can we crack that 4-6 seed range?
If so, any of those 3 seeds should be in reach; they’re bunched that closely.

It’s that nearly 3-win gap between 6th and 7th that is most daunting, even though those two unequal seeds
get fairly equal opponents in the first round (the 2- and 3-seeds, just over a game and a half apart).
The 6 seed tends to be much better than the 7, but doesn’t get rewarded for it.

"Mister Oden is a very, very big human being." - Jerryd Bayless

by QualityPie on Sep 19, 2008 3:59 PM PDT up reply actions

Aaaaaaand one last summary detail: The win range of each tier.

Tier 1 (seeds 1 – 3)
Win range: 56.75 – 60.75 (the high fifties)

Tier 2 (seeds 4 – 6)
Win range: 49.625 – 52.125 (clustered around 50 wins)

Tier 3 (seeds 7 & 8, and maybe a just-missed-it kind of team)
Win range: 44.875 – 46.75 (a winning record, but not threatening 50 wins)

There ya go.
Playoff teams sort out into those 3 categories.
- Winning record, but not threatening 50 wins.
- Roughly 50 wins
- Well over 50 wins.

Q.E.D.

"Mister Oden is a very, very big human being." - Jerryd Bayless

by QualityPie on Sep 19, 2008 4:15 PM PDT up reply actions

+1 for use of QED

+5 for anyone who knows the latin words that QED is an acronym for and
+5 for the translation into English

WITHOUT looking it up, not that I’d be able to prove you did or didn’t

Devil's Advocate or just argumentative?

by Magnum on Sep 20, 2008 4:24 PM PDT up reply actions

I forget if it was "Quo", "Quid", "Qua", or some such for the "Q" . . .

. . . but otherwise, it’s “Q. Erat Demonstrata”, or “It is demonstrated”:
Some certain state has been shown true.

"Mister Oden is a very, very big human being." - Jerryd Bayless

by QualityPie on Sep 20, 2008 4:37 PM PDT up reply actions

"Quid pro quo" translates as This for That

Process of elimination indicates that Qau = It

or I could be comepletely wrong

"I figured out how to get the canoe down the mountain, but I will need a snow blower and all your butter"-Michael Kelso

by 92wastheyear on Sep 20, 2008 4:45 PM PDT up reply actions

Of course I was wrong

link

"I figured out how to get the canoe down the mountain, but I will need a snow blower and all your butter"-Michael Kelso

by 92wastheyear on Sep 20, 2008 4:58 PM PDT up reply actions

good grief

you can email each other pictures of your slide rules, but do you have to tussle on here?

by Bust a Bucket on Sep 21, 2008 6:40 PM PDT up reply actions

In school (math) my teacher always said "Quod erat demonstrandum"

= “What could be proven/demonstrated”. Don’t know how great he was in Latin though.

St. Bayno's 120 Haiku Prospects: Nicolas Batum in the paint—prairie grass, blown by wind.

by Norsktroll on Sep 20, 2008 4:51 PM PDT up reply actions

more realistic sorting

instead of sorting by seed, I sorted by place in conference (# of wins). From this point we get a better view of where a team will land given the mean number of wins and the standard deviation for each place.
Place ave wins Std dev
1 60.75 3.34
2 58.125 1.90
3 57 1.89
4 53 2.77
5 51 2.14
6 49.375 3.44
7 46.625 3.08
8 44.875 2.57

The last column show the typical variation in the number of wins for that given placing. To summarize the general trend any place typical moves up or down by 2-3 wins in any given year. Therfore 49 wins would be likeliest to get you the 6th place but it is not atypical that 49 wins gets you either the 5th or 7th place as that is within on standard deviation (ie happens roughy 1/3 of the time).

by NWfan on Sep 19, 2008 3:51 PM PDT reply actions

sorry about the format

guess I should use the preview button
Place……….ave wins……..Std dev
1……………….60.75………..3.34
2……………….58.125 ………1.90
3……………….57…………….1.89
4……………….53…………….2.77
5……………… 51…………….2.14
6……………….49.375……….3.44
7……………….46.625……….3.08
8……………….44.875……….2.57

by NWfan on Sep 19, 2008 3:58 PM PDT up reply actions

I really wanted to add them up for the eastern conference also, for comparison, but I am too lazy. I don’t suppose you got that far?

by lukeyhere on Sep 19, 2008 3:53 PM PDT reply actions

Eastern conference

Seed….00-01…..01-02……02-03……03-04…..04-05…..05-06…..06-07….07-08…………Avg Wins
1…………56…………52…….50……….61………59…….64………..53………66 …………….57.85714286
2…………52…………50…….49……….47………54…….52………..50………59 …………….51.57142857
3…………50…………49…….48……….54………45…….49………..47………52 …………….49.14285714
4…………48 …………44…….48……….42………47…….50………..44………45 …………….45.71428571
5…………47…………44…….47……….41………45…….42………..49………43 …………….44.42857143
6…………46…………43…….44……….41………44…….41………..41………41 …………….42.14285714
7…………43…………42…….42……….39………43…….41………..41………40 …………….41.14285714
8…………41…………42…….42……….36………42…….40………..40………37 …………….39.85714286

And, just for NW fan,

Seed…….Standard Deviation
1………5.829

2……….3.662

3……….2.816

4……….2.673

5……….2.765

6……….1.923

7………..1.408

8………..2.330

Word.

by joelor on Sep 19, 2008 4:13 PM PDT up reply actions

It could be an aberration, it could be the start of a trend ;-)

Wait a second while I look for my Six Sigma black belt, ah, here it is, confidence interval, box chart, mumble, mumble, control chart, ucl, now where is my Principles of Economics, here we go, Stats 101, 82games, cross-check with PER and Wages of Wins, calculate calculate, so here it is: 45-37, 8th seed, +/- 1-2 games. Same result as by looking at the schedule through the crystal ball. But your historical analysis was better. Now I can believe it: 48 to 50 wins. Booya!

(via Canis Hoopus)

St. Bayno's 120 Haiku Prospects: Nicolas Batum in the paint—prairie grass, blown by wind.

by Norsktroll on Sep 19, 2008 4:08 PM PDT reply actions 1 recs

BTW, that 67 wins highlight? Dallas. Lost in the first round. So much for that.

The analysis in this post is solid and interesting and I didn’t want to belittle it. But it only indicates how much wins a team would approximately need in a given average year to reach a certain playoff spot based on a pretty small sample size. It gives no basis as to how good in this particular year we will be, or the division, or the conference. Still nice, so thanks for it.

St. Bayno's 120 Haiku Prospects: Nicolas Batum in the paint—prairie grass, blown by wind.

by Norsktroll on Sep 19, 2008 4:23 PM PDT reply actions