Back to Back Games -- Are They Really Harder?

by jscot on Jan 23, 2009 3:08 PM PST

Important Warning -- Nerd Alert!!! This is long and geeky. If you read it all, you are brave, crazy, or both.

A regular feature on my Sched Ahead Updates is the table which tells how many back to back games remain for each Western contender.

Current Status

	Total	Road	Cat4	Cat3	Cat2	Cat1
Portland	7	7	4	3	0	0
Dallas	9	5	3	2	1	3
Denver	10	7	4	3	2	1
Houston	7	6	2	4	1	0
L.A.	9	9	5	4	0	0
N.O.	9	7	3	4	2	0
Phoenix	9	8	2	6	0	1
S.A.	8	5	2	3	2	1
Utah	11	9	7	2	0	2

As discussed on those updates, Cat4 is road games against .500 or better teams, Cat3 road games against losing teams, Cat2 home games against winners, and Cat1 home games against losers. Utah has the toughest road ahead as far as back to backs go, both in number and difficulty.

The Question

But are back to back games really any more difficult?

Someone commented here recently that Hollinger doesn't factor in back-to-backs in his strength of schedule. I can't find anything from him on it, so don't know if this is true. Someone suggested any difficulty is because most back to backs are road games. I did some research.

Analysis -- What I Did and Did not Do

I did not look at overnight travel time or distance. My geekiness has limits.

I ignored the first game. My focus is on the second game of the back to back.

I did not factor out three games in four nights or 4 in 5. If a back to back adds difficulty, those add more. Since I treated a 3 in 4 as one back to back, and a 4 in 5 as two distinct back to backs, this analysis may slightly overstate the average impact of a typical back to back.

I did not factor out games where both teams are on a back to back. Would have been nice, but too much trouble. These are included for both teams. This analysis, therefore, may slightly understate the average impact of a typical back to back.

Since those last two factors tend to opposite results, partially cancelling each other, and neither are all that common, I gave a virtual shrug, and said, "Close enough." It's not as I'm being paid for this analysis. My disclaimers tell you what I have and haven't done, and no one can say, "Hey, you dummy, you forgot...." (They will, anyway....)

Further, I did not factor out games where a team is NOT on a back to back and it's opponent is. Ideally, I would compare games with one team on a back to back vs. games where neither is, but.... virtual shrug again. All told, this probably marginally overstates the magnitude of back to back impact, but certainly identifies general trends.

I looked at the last four full seasons. Stat geeks like point differential, but I used win-loss. My reasons:

Ultimately, wins matter.
On a large sample, the two should track fairly closely.
It is theoretically possible that some impact (mental fatigue?) of back to backs could be more significant in close games, so looking at point differential could miss something. A true geek would do both to see if point diff impact mirrors W/L impact.

I loaded all results from the last four seasons. (While pasting into Excel from ESPN, my eye caught something amazing. On 11/15/05, losing 103-79 to the Spurs, Atlanta was led in assists by 8 guys, all tied with one. 8 assists for the team(!), by 8 players. Is that a record? They LED 31-18 after one, then folded 85-48. Probably 5-6 assists in the first quarter, and then.... Incredible. Sorry for digressing.) Following is my extended (what else? This is a jscot post) analysis of those seasons.

All Games

All Games	W	L	%	% diff
BtoB	1048	1346	43.8%
NonBB	3872	3574	52.0%	8.2%

Get used to this table, if you ~~are dumb enough~~ have the courage to read further. It gives win-loss in back to back games, non back to backs, winning percentage for each, and percent difference.

The winning percentage for teams in 2394 back to back games was 43.8%, while on the other 7446 games it was 52.0%. Therefore, teams are 8.2% less likely to win a back to back, right? Well, no.

Road	W	L	%	% diff
BtoB	611	1022	37.4%
NonBB	1357	1930	41.3%	3.9%

Home
BtoB	437	324	57.4%
NonBB	2515	1644	60.5%	3.0%

The preponderance of back to backs on the road, where winning percentages are lower, skews the overall number. However, there is still a difference when broken down by home and away, both home and road games showing an impact, not huge, but also not negligible.

Degree of Difficulty

You will say (if you read this far), "Give me a break. You can't lump all road games together. Back to backs at the Clippers and at the L@kers are not the same thing." I will say, "Oh, you are right! I'll go tweak my spreadsheet." Then, I will come back with:

Cat 4	W	L	%	% diff
BtoB	218	628	25.8%
NonBB	556	1263	30.6%	4.8%

Cat 3
BtoB	393	394	49.9%
NonBB	801	667	54.6%	4.6%

Cat 2
BtoB	187	212	46.9%
NonBB	1120	1146	49.4%	2.6%

Cat 1
BtoB	250	112	69.1%
NonBB	1395	498	73.7%	4.6%

"Look," I will say. "Against both winners and losers, the percentage of both home and road games drops."

Finer Granularity of Opponent Strength

You will then say, because I often nitpick things, so you think I deserve it, "jscot, I've been ignoring it on your Sched Ahead updates, but isn't there a difference between games at Washington and at Philly? Or between home vs. Dallas (6 over .500) and home against the Celtics? Can't you break those down?"

Asking such questions will teach you why one of us will rule the world one day, and one of us won't, for I am ahead of you again. The four categories are not bad for a single season, with a small number of games in each category. However, for this larger sample, I like three levels of difficulty.

The first is teams that win 50 or more. At 50, you are over 60%, a playoff team, a legitimate threat to win at least one playoff series. A team without 50 wins is rarely a threat to win it all (1977 being the exception that matters, no matter what Houston fans think). At 50 wins, you are for real.

The next dividing line is 33 wins. With less than 33, you lost more than 60% of your games. You were bad.

That leaves in-betweeners. There's a big gap between 33 wins and 49 wins. But a team between .400 and .600 is capable of winning or losing almost any game. You can use my data to figure out stats for winning tweeners and losing tweeners, if you want. If you do it yourself, you can feel smart. (I feel smart, but sometimes facts and feelings diverge significantly.)

Opponents: Tough Guys/Middlers/Doormats

Games against "Tough Guys", 50 win teams:

All	W	L	%	% diff
BtoB	156	466	25.1%
NonBB	677	1325	33.8%	8.7%

Road
BtoB	84	344	19.6%
NonBB	215	669	24.3%	4.7%

Home
BtoB	72	122	37.1%
NonBB	462	656	41.3%	4.2%

That is similar to games against winning teams in general, as far as percentage differences. Sample size is smaller, but the difference is still large enough to appear significant. Back to backs, home or away, hurt when playing the best teams.

Games against Middlers (33-49 win teams):

All	W	L	%	% diff
BtoB	540	685	44.1%
NonBB	2052	1807	53.2%	9.1%

Road
BtoB	302	518	36.8%
NonBB	713	1009	41.4%	4.6%

Home
BtoB	238	167	58.8%
NonBB	1339	798	62.7%	3.9%

Home or away, teams do slightly worse on back to back games against middlers.

Games against Doormats (32 or fewer wins):

All	W	L	%	% diff
BtoB	352	195	64.4%
NonBB	1143	442	72.1%	7.8%

Road
BtoB	225	160	58.4%
NonBB	429	252	63.0%	4.6%

Home
BtoB	127	35	78.4%
NonBB	714	190	79.0%	0.6%

Yawn. The same, a 3-5% difference everywhere -- what? Wait! Has nerdism identified a surprise in that bottom number? Do back to backs not matter against doormats at home?

It's hard to say. The sample size is small, so perhaps 0.6% is an aberration. Someone could check ten years of data, but I'll pretend I have a life and pass on that.

A Conclusion

Back to back games, home or away, irrespective of opponent quality, have a noticeable but not huge impact on NBA team's ability to win games, reducing the chance of victory by somewhere in the 3-5% range. The one possible exception is home games against really bad teams, where there may not be any significant general impact.

The 3-5% impact is an absolute, not relative, impact. On road games against top teams, win percentage drops from 24.3 to 19.6. In relative terms, then, the back to back reduces your 24% chance by one fifth, to 19%. Caveat: as stated earlier, this analysis may marginally overstate the magnitude of the impact.

Strength of the Team on the Back to Back.

Obviously, strength of opposition is not the whole story. Enquiring minds want to know, do good teams experience less back to back impact than bad teams, more, or the same? Enquiring minds rejoice, for jscot is here, filling empty minds with ever more useless information. If your eyes have not glazed over yet, there is something seriously wrong. Seek help immediately, or you may read the rest of this.

How do Tough Guys Do?

All	W	L	%	% diff
BtoB	361	241	60.0%
NonBB	1430	592	70.7%	10.8%

Road
BtoB	226	194	53.8%
NonBB	552	340	61.9%	8.1%

Home
BtoB	135	47	74.2%
NonBB	878	252	77.7%	3.5%

Cat 4
BtoB	91	112	44.8%
NonBB	248	244	50.4%	5.6%

Cat 3
BtoB	135	82	62.2%
NonBB	304	96	76.0%	13.8%

Cat 2
BtoB	55	35	61.1%
NonBB	417	187	69.0%	7.9%

Cat 1
BtoB	80	12	87.0%
NonBB	461	65	87.6%	0.7%

This was interesting. Tough Guys suffer worse impact from back to backs than the league as a whole, especially on road games against losing teams. (Granted, the sample sizes are not large, especially on home games.)

Are top teams better concentrators (is that part of why they are better?), and suffer a let-down against weak teams on the road in back to backs, but less so when the opponent is strong? If I were a rich owner, I'd pay somebody smarter than me to dig deeper. If you understand the cause, you might be able to mitigate it....

Cat 1 shows little difference, but it is a small sample. Or perhaps we're seeing a pattern that home games against weak teams aren't much impacted by the average back to back....

Finer Granularity of Opponent Quality for the Tough Guys

Smaller sample sizes here, but interesting anyway.

Tough Guys vs other Tough Guys:

All	W	L	%	% diff
BtoB	59	78	43.1%
NonBB	292	273	51.7%	8.6%

Road
BtoB	35	60	36.8%
NonBB	101	155	39.5%	2.6%

Home
BtoB	24	18	57.1%
NonBB	191	118	61.8%	4.7%

Tough Guys vs Middlers:

All	W	L	%	% diff
BtoB	176	123	58.9%
NonBB	749	256	74.5%	15.7%

Road
BtoB	108	97	52.7%
NonBB	300	146	67.3%	14.6%

Home
BtoB	68	26	72.3%
NonBB	449	110	80.3%	8.0%

There's some big diffs there, despite a small sample size. Why is the impact small for elite teams when playing other elites, but large when playing non-elites? My concentration theory is just a guess. Maybe the Blazer brain trust will analyze this one....

Tough Guys vs Doormats:

All	W	L	%	% diff
BtoB	126	40	75.9%
NonBB	389	63	86.1%	10.2%

Road
BtoB	83	37	69.2%
NonBB	151	39	79.5%	10.3%

Home
BtoB	43	3	93.5%
NonBB	238	24	90.8%	-2.6%

Sample size is so small on home back to backs that the outlier may be irrelevant. But look at road games against Doormats. 50+ win teams lose 30% against Doormats on road back to backs. Don't get depressed if your team struggles in such games.

A Surprising Result

The performance of top teams vs. varying levels of competition is fascinating. I expected greater impact vs. other top teams, but results show otherwise. Perhaps over 10 seasons it would not bear out, but the differences look too large to be just statistical noise.

How do Middlers Do?

All	W	L	%	% diff
BtoB	543	718	43.1%
NonBB	1949	1874	51.0%	7.9%

Road
BtoB	316	544	36.7%
NonBB	649	1033	38.6%	1.8%

Home
BtoB	227	174	56.6%
NonBB	1300	841	60.7%	4.1%

Cat 4
BtoB	99	341	22.5%
NonBB	260	670	28.0%	5.5%

Cat 3
BtoB	217	203	51.7%
NonBB	389	363	51.7%	0.1%

Cat 2
BtoB	105	118	47.1%
NonBB	559	589	48.7%	1.6%

Cat 1
BtoB	122	56	68.5%
NonBB	741	252	74.6%	6.1%

Unlike Tough Guys, Middlers have more back to back impact against winners than against losers on the road. Statistical noise? Maybe, but sample sizes (for road games) aren't that small.

Why (in road back to backs) are "Middlers" impacted more against winners, while "Tough Guys" find the greater impact against losers/middlers? Weird.

Middlers vs Tough Guys:

All	W	L	%	% diff
BtoB	80	258	23.7%
NonBB	299	667	31.0%	7.3%

Road
BtoB	41	189	17.8%
NonBB	95	328	22.5%	4.6%

Home
BtoB	39	69	36.1%
NonBB	204	339	37.6%	1.5%

Middlers vs Other Middlers:

All	W	L	%	% diff
BtoB	280	360	43.8%
NonBB	1057	977	52.0%	8.2%

Road
BtoB	154	275	35.9%
NonBB	338	570	37.2%	1.3%

Home
BtoB	126	85	59.7%
NonBB	719	407	63.9%	4.1%

Middlers vs Doormats:

All	W	L	%	% diff
BtoB	183	100	64.7%
NonBB	593	230	72.1%	7.4%

Road
BtoB	121	80	60.2%
NonBB	216	135	61.5%	1.3%

Home
BtoB	62	20	75.6%
NonBB	377	95	79.9%	4.3%

Not a lot to notice. Middlers see a small impact in all categories, home and away -- small enough that it could be statistical noise, but the statistical noise would all be going in the same direction....

What About Doormats?

I won't' to post Doormat numbers, because the tables make this post so large that the SBN software is screaming. To get them: 1) Subtract TG numbers from Winning numbers (Cat2 & 4), to get Middler Winner (41-49 Ws) stats. 2) Subtract Mid-Win from Middler totals to get Mid-Lose (33-40 Ws) numbers. Subtract Mid-Lose numbers from Loser numbers (Cat1 & 3). You'll have Doormats W/L -- then calculate the percentages. You, too, can be a nerd.

Doormat numbers are all over the place, but Doormats always get walked on. Perhaps it comes down to this -- except against other Doormats (especially home games), Doormats lose. A lot. If they win against anyone else, it is because someone couldn't miss and went 23-25, the other team had the flu, the planets aligned, some guy named Tim was ref, or something. Back to back impacts are unseen, swamped by the sheer badness of Doormats, or overwhelmed by the enormity of some other astounding event which led to a Doormat W.

Final Thoughts

Did you actually read all that, or just skip to the bottom?
A true stat geek would enlarge on this (using 10 years of data) to accurately quantify back to back impact, and factor it into his Strength of Schedule numbers. Otherwise, he's a wimp.
Paul Allen should drop a few bucks hiring a smart consultant to analyze this, examine which teams do better and which don't, and try to assess A) what factors most contribute to this B) what the best teams at this are doing and C) what else we could try. It might gain a win or two a year (each team plays about 20 of these), which might mean home court in a playoff series, which might mean advancing.
Someone someday should compare point differential impact to W/L impact. It won't be me.
Someday, I will rule the world, and my minions will paste stuff into Excel for me. I never want to press Ctrl-C Ctrl-V again. Perhaps Paul Allen can give me a special voice operated version of Windows, or something.

Tweet Comment 23 comments | 6 recs |

Do you like this story?

Comments

All of this must mean that we will make the play offs!

Yeah, I skipped to the bottom

by The Thinker on Jan 23, 2009 3:20 PM PST reply actions

I skipped directly to the end.

I’m a geek of a different kind.

by MiledAnimal on Jan 23, 2009 3:20 PM PST reply actions

Did you

Come up with all this?? Impressive-YES.

We makin the Playoffs.. woot woot.

by Slickrex on Jan 23, 2009 3:21 PM PST reply actions

Nice job, J !

      I stopped reading at “Finer Granularity of Opponent Strength”.
That sounds like a test question from my Human Anatomy
class in college. That was a long time ago, so I’ll apply it to
my strength training regimen.
      Thanks,
                      Walkoff – 41

It's GO time !

by walkoff41 on Jan 23, 2009 3:27 PM PST reply actions

Read through the whole thing

the most fascinating finding is that the tough teams struggle on back-too-backs. I have a theory… could it be that the elite teams are often older, and as such they have a more difficult time recovering in time for the second game?

by Cablinasian on Jan 23, 2009 3:28 PM PST reply actions

That's a reasonable theory, too

Last year, though, Utah won 54 games, and most would have considered them elite, and young.

Last year, they were 0-3 in back to backs on the road against other elites, 7-6 when not on a back to back. Against mid-level teams, they were 0-3 in back to back road games, 4-6 otherwise. This year, they are 1-9 in back to backs, 1-7 on the road with only a win at Memphis, and 0-2 at home with losses to New Jersey and Miami.

They aren’t old, but they are bad at back to backs.

Last year, against other elites, the Spurs went 0-1 on the road, 3-0 at home in back to backs (Texas teams seem to get easier back to back schedules, by the way). In other games, they were 4-12 and 9-5 — not much drop-off, it seems to me.

The Suns didn’t have much drop-off, either, a little bit against Doormats, but other than that, not really. Boston was 3-0 on road back to backs against elites, 0-3 against mid-level teams.

If it is physical fatigue, you would expect it to show up at least as much against elites as mid-levels, wouldn’t you?

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 23, 2009 9:49 PM PST up reply actions

looks more like, um, 'professionalism' or something

born from experience.

also the coaching re elite to elite you mention.

ignacio

by ignacio on Jan 24, 2009 4:42 AM PST up reply actions

So the Spurs are professional

and the Jazz aren’t. Sloan will not thank you. :)

Sample size is so small with one team that it is really hard to say. But it is hardly surprising that experienced elite teams would do better at this than other teams.

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 24, 2009 5:48 AM PST up reply actions

True geeks

read the abstract and conclusion and then decide if the details are worth reading.

FWIW from a physiological standpoint you will not be able to replenish your glycogen stores in less than 48 hours so big minute players have that working against them in back to backs.

Life is exhausting when you are this stupid.

by jonestr on Jan 23, 2009 3:52 PM PST reply actions

So true.

by PoliSam on Jan 23, 2009 5:54 PM PST up reply actions

As posted just above

If physical fatigue is the main factor, you would expect it to show up at least as much against elites as mid-level teams, wouldn’t you? Any idea why it doesn’t?

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 23, 2009 9:51 PM PST up reply actions

Depth?

In this small of a sample, I’d say differences in depth could be a factor. Maybe there’s no systematic difference in middle level teams depth vs. other level teams’ depth, but in one season, there’s likely to be a “fluke” correlation large enough to matter.

by PoliSam on Jan 23, 2009 11:20 PM PST up reply actions

Oops

I didn’t see that he did four years. I still say depth could be a factor… Or coaches ability to manage back-to-backs; coaching ability could be correlated with elite status.

by PoliSam on Jan 23, 2009 11:22 PM PST up reply actions

Except that elites

actually have a worse impact from these than the league as a whole.

OK, here’s a theory. In general, elites have better coaches. Better coaches do a better job of preparing a team for a particular game. This advantage is lessened when on a back to back — there isn’t the practice time to really get a team ready. So elites lose one of their advantages.

But when an elite is playing a back to back, and the second game is against another elite, they perhaps spend a little more time preparing for the second game in advance. If L.A. has a back to back against Washington and Boston, which team will get more prep time from Phil? Obviously Boston. What if it was at Boston and then at Philly the next night? Boston still gets most of the prep time. So this could be part of the picture.

The article you cited was based on two years, but I looked at four.

If depth is a factor, you would expect it to show up reasonably consistently despite the level of the competition. It doesn’t.

Maybe point differential would shed more light on this. It could be that point differential is showing a more consistent impact, but that teams are still losing to other elites because they were going to lose on the road anyway.

Say the back to back impact is 2 ppg (just pulling a number out of the hat). Suppose elites on average lose to other elites on the road by 5 ppg. In that case, a back to back would take you to an average of 7 ppg, which would make a difference in W/L percentage, but perhaps not an enormous difference.

Suppose that same elite team will, on average, beat a mid-level team by 3 ppg on the road, and the back to back impact takes you to an average margin of +1. Now, you’ll see a pretty significant impact on W/L.

Like I said, someone should do the point differential study and compare it to the W/L statistics, it could tell us a lot more.

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 24, 2009 12:37 AM PST up reply actions

Nice work

This topic (or one close to it) has also been studied by statisticians:

The Role of Rest in the NBA Home-Court Advantage

Oliver A. Entine, University of Pennsylvania
Dylan S. Small, University of Pennsylvania
Abstract

To date, the factors which lead to the very large home court advantage characteristic of the NBA have not yet been well isolated. This study analyzes the relationship between that home court advantage and the comparatively fewer days of rest between games that the NBA schedule imposes on visiting teams. A statistical model has been developed and applied to the NBA data for the 2004-2005 and 2005-2006 seasons to estimate the importance of the effect of rest on the magnitude of the home court advantage. The results indicate that lack of rest for the road team, while not a dominant factor, is an important contributor to the home court advantage in the NBA.

http://www.bepress.com/cgi/viewcontent.cgi?article=1106&context=jqas

On the other hand, I think I could do a better job than this article. Of course I could!

by PoliSam on Jan 23, 2009 5:51 PM PST reply actions

I'll try to read that soon

but I doubt two seasons really is enough data to be conclusive, even if their conclusions may be similar to mine. There are so many variables that affect game outcomes.

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 23, 2009 9:55 PM PST up reply actions

Perhaps

There are so many variables that affect game outcomes.

I cannot image approx 1,000 games being significantly better to address that issue than 500. If you just want to know the mean effect of back-to-backs, 500 is enough. If you want to make the more fine-grained comparisons that you did, then 500 might not be enough.

by PoliSam on Jan 23, 2009 11:26 PM PST up reply actions

For general trends, right

To try and identify causes, I think you may need more fine-grained comparisons.

Some things you could look at:
1. Depth. Look at the teams that performed above the average on back to back impact vs. below the average teams. Compare the avg mpg of 7th-9th men on those teams.
2. Veteran-ness. Average age (or years experience, either one) of top three players, or starting unit, or rotation players, or age per minutes played, or whatever.
3. Team stability. How many years of experience on that particular team have the key players had together.
4. Coaching. Rank your coaches (this is subjective, of course), and see how they did.
5. Pace.
6. Defensive efficiency.
7. Offensive efficiency.
8. Rebounding.
9. Clutchness.

Of course, there are other factors that can’t really be measured statistically (and some of the above may be a little dubious). But it would be interesting to see any correlations between relative success on back to backs and some of these factors. Somebody could do it for their thesis, I suppose.

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 24, 2009 12:48 AM PST up reply actions

the first half of this post was incredibly interesting

by pklym on Jan 23, 2009 6:31 PM PST reply actions

You really read half of it?

When I rule the world, everyone will know how to use Excel.

by jscot on Jan 23, 2009 9:55 PM PST up reply actions

i read most of it

there’s a lot to ponder in the realm of assessing a team’s concentration-level. (which might lead us into subjective factors hard from the outside to analyze. girlfriends and so on, ‘private lives.’ how closely do teams monitor these things? how much do they own?)

of course, in the playoffs concentration is assumed, though this may be unwise, and the additional, large factor enters of a a team figuring out or getting a psychic edge on/vs their opponent. which is where coaching, and a given team’s receptivity to coaching, sometimes plays a magnified role (seemingly).

ignacio

by ignacio on Jan 23, 2009 8:26 PM PST reply actions

I could see this being done by the teams

One person could go piece by piece finding out which of them works and which doesn’t …empirecally. It would 20 years or so but then you could start giving advise to the team …right about the time you drop dead from old age

Larry (the new Johnnie Cochran) Miller: "If we get screwed, we're gonna sue"

by 92wastheyear on Jan 23, 2009 8:35 PM PST up reply actions

Devil's Advocate

Looking at the theory that home court advantage is simply a lack of travel, how about this thought. Back to backs are only hard because you’re traveling twice as much. This means that if you have an away game/home game back to back, you’re effectively the road team. In fact, in our home loss to the Clippers we were in Utah the night before and IIRC they were actually in Portland a day before us. Was this a home loss or a road loss then?!?

Basically I’d be interested in seeing Home/Home back to backs compared to Road/Home back to backs. Especially when the other team isn’t on a back to back and gets to be in town for a day. (This is obviously way too much work for any one man, but even the H/H vs R/H would be interesting.)

by Zaig on Feb 3, 2009 9:24 AM PST reply actions