## Re: streaks

Expand Messages
• ... the ... I just use the BINOMDIST() function in excel to calculate the probabilities of exceedences. It is pretty similar, if not the same to what MikeT
Message 1 of 6 , Aug 17, 2001
• 0 Attachment
--- In APBR_analysis@y..., "John Grasso" <johng@s...> wrote:
> I've recently being doing some research on the 1970-71 NBA season
> (individual high games - not streaks). This was the first year for
the
> Buffalo, Cleveland and Portland franchises. Streaks seem to be quite
> prevalent. Here's a few examples.
>

I just use the BINOMDIST() function in excel to calculate the
probabilities of exceedences. It is pretty similar, if not the same
to what MikeT posted. I just haven't had a chance to carefully check
it. (the 0.0014 number Mike calculated is actually almost the
same as the 0.0017 number I calculated). Using the season records
below as the "true" odds, I find the associated chances of these
streaks occurring...

> Cleveland - 15-67 for the season
> 0-15 start (4.8%)
> 1-0 (18.3%)
> 0-12 (8.9%)
> 1-0 (18.3%)
> 0-7 (24.3%)
> 1-0 (18.3%)
> 0-3 (54.6%)
> 2-0 (3.4%)
> 0-7 (24.3%)
> 6-4 (0.4%)
> 4-19 finish
>

Only the 6-4 finish is that surprising. My sense, too, is that this
kind of streak can definitely happen at the end of a bad season when

>
> Detroit 12-1 start (0.5%)
> 0-6 (0.8%)
> 11-7 (38.7%)
> 11-2 (2.6%)
> 11-21 finish (1.6%)
> 45-37 season
>

OK. These guys were streaky.

> San Diego 23-20 start
> 1-16 (0.02%)
> 8-2 (4.7%)
> 1-4 (20.3%)
> 7-0 finish (all home games) (0.7%)
> 40-42 season
>

2 pretty good streaks here, but, as you mention, we could do this more
methodically and look at home/road numbers. If they were all home
games, that would and the team won 60% of its home games (a pretty
typical NBA figure for a 0.500 team), that moves the odds of a 7-0
finish to 3.1%.

> Portland finished season
> 5-0 (0.6%)
> 2-14 (4.2%)
> 5-0 (0.6%)
> 29-53 season

The 5-0 streaks are weird.

Dean Oliver
• ... The 6-4 wasn t quite at the end of the season - after that hot streak they went 4-19 to finish the season - still they won 10 of their 15 victories in
Message 2 of 6 , Aug 17, 2001
• 0 Attachment
> > Cleveland - 0-15 start (4.8%)
> > 1-0 (18.3%)
> > 0-12 (8.9%)
> > 1-0 (18.3%)
> > 0-7 (24.3%)
> > 1-0 (18.3%)
> > 0-3 (54.6%)
> > 2-0 (3.4%)
> > 0-7 (24.3%)
> > 6-4 (0.4%)
> > 4-19
> > 15-67 season
>
> Only the 6-4 finish is that surprising. My sense, too, is that this
> kind of streak can definitely happen at the end of a bad season when
> no one cares about you.

The 6-4 wasn't quite at the end of the season - after that hot streak they
went 4-19 to finish the season - still they won 10 of their 15 victories in
their last 33 games. After the first 40 games they had only won 3 games and
those were against the other two expansion teams. Their first win against an
established team was over Philadelphia around Christmas.
• ... Something I realized about this analysis is that there are a lot of _possible_ streaks in a season. Though the 7-0 streak at the end had a 0.7% chance of
Message 3 of 6 , Aug 17, 2001
• 0 Attachment
--- In APBR_analysis@y..., "John Grasso" <johng@s...> wrote:
> I've recently being doing some research on the 1970-71 NBA season
>
> San Diego 23-20 start
> 1-16
> 8-2
> 1-4
> 7-0 finish (all home games)
> 40-42 season

_possible_ streaks in a season. Though the 7-0 streak at the end had
a 0.7% chance of happening, there were 75 possible sequences of 7
games for this to happen in the season. Hence, the odds of this kind
of team having 1+ series of 7 consecutive wins in a season are 39%.
No big deal.

On the other hand, the 1-16 streak was pretty strong. A 40-42 team
has a 0.02% chance of going 1-16 in 17 games. There are about 55
possible 17 game streaks in an 82 game season. The odds of any 40-42
team having this kind of streak in an 82 game season are 1.1%. That
is pretty impressive.

Back to the Sting. They have a shorter season, only 32 games. The
odds of having 1+ streak of 1-10 is 3.5% -- pretty unusual. The odds
of having a 17-4 streak or better is 17%, not a big deal. Multiplying
again (gulp), this makes for a 0.60% chance of such a season.

Dean Oliver
• I really need to dig up that paper... :)
Message 4 of 6 , Aug 17, 2001
• 0 Attachment
I really need to dig up that paper... :)

On Fri, 17 Aug 2001, Dean Oliver wrote:

> --- In APBR_analysis@y..., "John Grasso" <johng@s...> wrote:
> > I've recently being doing some research on the 1970-71 NBA season
> >
> > San Diego 23-20 start
> > 1-16
> > 8-2
> > 1-4
> > 7-0 finish (all home games)
> > 40-42 season
>
> _possible_ streaks in a season. Though the 7-0 streak at the end had
> a 0.7% chance of happening, there were 75 possible sequences of 7
> games for this to happen in the season. Hence, the odds of this kind
> of team having 1+ series of 7 consecutive wins in a season are 39%.
> No big deal.
>
> On the other hand, the 1-16 streak was pretty strong. A 40-42 team
> has a 0.02% chance of going 1-16 in 17 games. There are about 55
> possible 17 game streaks in an 82 game season. The odds of any 40-42
> team having this kind of streak in an 82 game season are 1.1%. That
> is pretty impressive.
>
> Back to the Sting. They have a shorter season, only 32 games. The
> odds of having 1+ streak of 1-10 is 3.5% -- pretty unusual. The odds
> of having a 17-4 streak or better is 17%, not a big deal. Multiplying
> again (gulp), this makes for a 0.60% chance of such a season.
>
> Dean Oliver
>
>
> To unsubscribe from this group, send an email to:
> APBR_analysis-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
• ... Given the extra work I did to evaluate the odds of a x% team having a streak of a certain type, I figured I d go back to this post and post the actual odds
Message 5 of 6 , Aug 29, 2001
• 0 Attachment
--- In APBR_analysis@y..., "John Grasso" <johng@s...> wrote:

Given the extra work I did to evaluate the odds of a x% team having a
streak of a certain type, I figured I'd go back to this post and post
the actual odds of such a streak occurring. Low values, less than 5%,
are considered pretty unusual.

>
> Cleveland - 15-67 for the season
> 0-15 start (96.6%)
> 1-0 (100%)
> 0-12 (99.9%)
> 1-0
> 0-7
> 1-0
> 0-3
> 2-0
> 0-7
> 6-4 (25%)
> 4-19 finish
>

Nothing statistically unusual here. Even the 6-4 streak isn't that
unusual for such a team.

>
> Detroit 12-1 start (29%)
> 0-6 (48%)
> 11-7
> 11-2
> 11-21 finish (55%)
> 45-37 season
>

Nothing weird here either.

> San Diego 23-20 start
> 1-16 (1.3%)
> 8-2
> 1-4
> 7-0 finish (all home games) (39%)
> 40-42 season
>

The 1-16 run was weird. Everything else is normal.

> Portland finished season
> 5-0 (35%)
> 2-14 (94%)
> 5-0 (35%)
> 29-53 season

The odds of having 2 5-0 streaks is only 7%, so that may be a little
weird. Otherwise, nothing statistically unusual.

This really points out the old Tversky research that the mind
perceives streaks a lot faster than stats would suggest they are
really happening. Basically, if your mediocre team wins 5 or 7 in a
row, it may not mean anything. It happens a lot to mediocre teams, as
John pointed out here.

I even ran the numbers on the 69-13 Lakers that won 33 straight. The
likelihood of a random 69-13 team winning 33 straight is 15%, not that
unusual for such a good team. It was more unusual to have a team that
good throughout the season. This basically proves that the
Lakers didn't win 69 by fluke. A 60-22 team wins 33 straight only
0.17% of the time. A 65-17 teams wins 33 straight only in 2.3% of
their seasons. What was the Bulls longest streak in their 72 win