Predicting playoffs from days of rest records in regular season

Home for all your discussion of basketball statistical analysis.
Post Reply
Mike G
Posts: 4303
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Predicting playoffs from days of rest records in regular season

Post by Mike G » Wed Apr 10, 2019 7:40 pm

Over here, Crow did an excellent summary of Western playoff teams' MOV in various states of rest between games.
viewtopic.php?f=2&t=9480&start=390
Zero days rest = back-to-back games. One day off between games is most common in RS: about 50 of 82 games, on avg.

As of yesterday, playoff teams' MOV with different days of rest:

Code: Select all

DR :    0      1      2     3+
Mil    2.2   10.4   10.9   21.0
Tor    9.1    5.3    7.0   -2.7
Phl   -7.3    5.7    3.5    4.0
Bos   -4.9    6.1    3.4   13.0

Ind    5.5   -0.1    9.2   15.0
Brk   -8.7    1.6    0.7    2.5
Orl   -0.9    2.7   -2.7   -8.4
Det   -1.0   -3.5    9.0    1.3
Cha   -5.8    0.0   -3.3    6.7

avg   -1.3    3.1    4.2    5.8
  
              
DR :    0      1      2     3+
GSW    3.7    7.6    5.1   21.5
Den    8.7    2.1   13.2    2.4
Hou    9.4    6.1    0.7   -8.3
Por    0.0    4.7    4.2   10.4

Uta   -0.9    7.9    1.4   11.0
OKC    5.2    3.0    3.6    4.0
SAS   -4.9    1.9    6.1    4.0
LAC   -0.2    0.5    2.0    7.0

avg    2.6    4.2    4.5    6.5
In playoffs, there are no b2b; and no rest over 2 days except between series.
Rather than discard data in the zero and 3+ columns, I've given them half-weight, multiplied by games, and rolled them in with the 1's and 2's respectively.
Last year, one day rest was about twice as common as 2 within a PO series. Series-opening games matter, though, so I'm giving equal weight to 0-1 day rest and 2+.
Here the PO column is just the avg of the first 2; RS is reg-sea MOV:

Code: Select all

east   0-1    2-3     PO     RS
Mil    9.4   11.8   10.6    9.1
Tor    5.7    6.0    5.8    6.1
Phl    4.2    3.5    3.9    2.5
Bos    4.8    4.7    4.8    4.4

Ind    0.6   10.0    5.3    3.4
Brk    0.3    0.9    0.6   -0.3
Orl    2.2   -3.2   -0.5    0.6
Det   -3.2    8.3    2.5   -0.6
Cha   -0.8   -2.2   -1.5   -1.0

avg    2.6    4.4    3.5    2.8
                

west   0-1    2-3     PO     RS
GSW    7.2    5.9    6.5    6.7
Den    2.8    9.6    6.2    4.0
Hou    6.4   -0.2    3.1    4.8
Por    4.3    4.9    4.6    4.2

Uta    6.9    2.3    4.6    5.4
OKC    3.2    3.6    3.4    3.3
SAS    1.1    5.9    3.5    1.6
LAC    0.4    2.3    1.3    0.8

avg    4.0    4.3    4.2    3.5
The Bucks are mediocre in b2b, strong any other time.
Rockets are the opposite; dominant without rest, allergic to downtime

Mike G
Posts: 4303
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Predicting playoffs from days of rest records in regular season

Post by Mike G » Thu Apr 11, 2019 1:06 pm

With no further refinements or insights, the rest-adjusted team strengths shown above indicate these 1st-round results.
First column is 1st-vs-8th etc. PO are values from opening post.
Higher seed is predicted winner except where marked by *-*, indicated in both 4-vs-5 matchups.

Code: Select all

east  tm  PO    PO   tm   mov   G
1-8  Mil 10.6   2.5  Det  8.1   4
2-7  Tor  5.8  -0.5  Orl  6.3   5
3-6  Phl  3.9   0.6  Brk  3.3   6
4-5  Bos  4.8   5.3 *Ind* 0.5   7
                           
west  tm  PO    PO   tm   mov   G
1-8  GSW  6.5   1.3  LAC  5.2   5
2-7  Den  6.2   3.5  SAS  2.7   6
3-6  Por  4.6   3.4  OKC  1.2   7
4-5  Hou  3.1   4.6 *Uta* 1.5   7
Uta > Hou by 0.5 after one day off, and by 2.5 after two days.
Bos > Ind by 4.2 with one rest day; Ind > Bos by 5.3 after 2. Does not account for missing Oladipo (or anything else).

Post Reply