Wednesday, February 1, 2017

On Streaks and Breaks

3 articles in two days... What's gotten into me.

So after remembering the Botwinnik's quote, and after publishing the stats how the teams actually play after different breaks, a new idea came to me - check whether the teams on streaks are affected positively or negatively by breaks.

For the sake of the analysis, I assumed the following:
  • A break is a period of three days at least between games.
  • A streak is a sequence of at least three wins in a row, or at least seven points in four games.
So we check for the last thirty years (as far as NHL.com would let us in) if the streaking team was able to keep the streak alive, or whether the streak was broken:

SEASON ALIVE BROKEN
1987/1988 5 11
1988/1989 12 7
1989/1990 8 14
1990/1991 13 11
1991/1992 17 13
1992/1993 20 16
1993/1994 19 20
1994/1995 2 7
1995/1996 15 11
1996/1997 15 11
1997/1998 12 20
1998/1999 12 9
1999/2000 18 12
2000/2001 21 11
2001/2002 17 6
2002/2003 13 10
2003/2004 12 14
2005/2006 31 15
2006/2007 16 16
2007/2008 23 24
2008/2009 15 20
2009/2010 14 17
2010/2011 19 11
2011/2012 22 11
2012/2013 6 3
2013/2014 15 15
2014/2015 16 16
2015/2016 16 14
2016/2017 8 11
TOTAL 432 376

Actually, it looks like the streaks weren't affected by the break either way. 53.4% of the times the streak continued, 46.6% of the time it went dead. There is a very large discrepancy between the seasons, although I'd attribute it to lesser parity between the teams overall in these years. For the last 5 years, the probability for the streak to stay alive has been 50.8% (61 cases of extended streaks out of 120).

Now, what would change, if we define a break a little bit longer, by a single day:

SEASON ALIVE BROKEN
1987/1988
2
2
1988/1989
4
1
1989/1990
3
1
1990/1991
4
4
1991/1992
7
8
1992/1993
8
2
1993/1994
7
7
1994/1995
1
1
1995/1996
6
7
1996/1997
6
2
1997/1998
5
5
1998/1999
6
2
1999/2000
9
3
2000/2001
7
4
2001/2002
6
4
2002/2003
5
1
2003/2004
3
6
2005/2006
16
4
2006/2007
8
5
2007/2008
10
6
2008/2009
8
8
2009/2010
6
6
2010/2011
9
3
2011/2012
8
4
2012/2013
2
1
2013/2014
3
9
2014/2015
5
9
2015/2016
7
6
2016/2017
3
8
TOTAL
174
129

The changes are rather interesting. Now, overall, the chances of streak to continue are up to 57.4%, and only in 42.6% of the cases it came to a stop. But in the last five years - since the last lockout - and with the schedule changes so that there are at least two games between every team (increasing travel), the ratio drops from 50.8% to the humble 37.7% (20 out of 53!)

Extending the breaks to five days provides too little data to draw any conclusions.

So I am inclined to agree with Dr. Botwinnik, that extended breaks of more than three days throw teams off their pace and should be reduced to minimum. Three days are borderline alright.

No comments:

Post a Comment