Sunday, June 24, 2012

MLB Team Rankings - Part Two: Results

Texas Ranger 1846
Texas Ranger
This is part two of a post on creating betting market power rankings for major league baseball.  Part one focused on methodology.  You can find it here.

Part two of this post covers the actual rankings.  In a nutshell, I'm using the information contained in the moneylines and run totals set by the Vegas sportsbooks to "reverse engineer" an implied power ranking.  Part one of this post covers the methodology in more (excruciating) detail.


These rankings are based on results and betting information as of the morning of Sunday, June 24, 2012.  I plan to publish these daily in a format similar to my Ticker format for the NBA.


Here is a glossary of the fields in the ranking table:

Glossary

  • GWP - Stands for "Generic Win Probability".  It's what you would expect the team's win probability to be against a league average opponent on a neutral field.  The term itself was stolen from inspired by the rankings published at Advanced NFL Stats.
  • oRAR - Stands for "Offensive Runs Above Replacement".  It is how many more runs per game a team's offense would be expected to score, compared to a league average opponent.
  • dRAR - Stands for "Defensive Runs Above Replacement".  It is how many runs per game a team's defense/pitching would be expected to prevent, compared to a league average opponent.  The sign convention is such that a higher number means better defense/pitching (even though it implies fewer runs being scored).
  • GTOT - Stands for "Generic Total".  It's what you would expect the betting total to be for a the team when playing against a league average opponent.
  • W-L - The team's season-to-date win loss record and corresponding rank.  For comparison against the implied "true" strength of the team.


The Rankings


Rank Team GWP oRAR dRAR GTOT W-L
1  tex 0.59 0.71 1 0.18 7 8.9 3 44-28 1
2  nyy 0.57 0.58 4 0.15 9 8.8 6 42-28 2
3  ana 0.55 -0.05 17 0.47 2 7.8 26 39-33 10
4  bos 0.54 0.69 3 -0.26 25 9.3 2 37-34 15
5  det 0.54 0.10 8 0.23 6 8.2 19 34-37 20
6  tb 0.53 -0.10 20 0.41 5 7.8 25 38-32 9
7  stl 0.53 0.24 6 0.02 17 8.6 11 37-35 17
8  ari 0.52 0.24 7 0.00 19 8.6 10 36-35 18
9  was 0.52 -0.29 25 0.46 3 7.6 28 41-28 4
10  cin 0.52 0.07 10 0.11 13 8.3 17 39-31 6
11  atl 0.52 0.01 13 0.15 8 8.2 20 38-33 12
12  sf 0.51 -0.59 30 0.68 1 7.1 30 40-32 7
13  tor 0.51 0.28 5 -0.19 23 8.8 5 37-34 14
14  phi 0.51 -0.05 18 0.13 11 8.2 21 34-38 21
15  cws 0.51 0.04 12 0.04 16 8.3 14 37-34 16
16  mia 0.50 -0.08 19 0.11 12 8.2 22 33-38 23
17  mil 0.50 0.00 14 0.01 18 8.3 15 33-38 22
18  la 0.49 -0.50 28 0.46 4 7.4 29 43-29 3
19  nym 0.49 -0.21 23 0.15 10 8.0 23 39-33 11
20  bal 0.49 0.07 9 -0.18 22 8.6 9 40-31 5
21  sea 0.48 0.00 15 -0.20 24 8.5 12 31-42 25
22  oak 0.47 -0.18 21 -0.06 20 8.2 18 34-37 19
23  cle 0.47 0.07 11 -0.33 27 8.7 7 37-33 13
24  pit 0.47 -0.39 27 0.11 14 7.9 24 38-32 8
25  col 0.46 0.70 2 -1.12 30 10.2 1 27-43 28
26  kc 0.46 -0.20 22 -0.18 21 8.3 16 30-38 24
27  min 0.45 -0.03 16 -0.49 28 8.8 4 28-42 27
28  sd 0.44 -0.58 29 0.07 15 7.7 27 25-47 29
29  chc 0.44 -0.30 26 -0.30 26 8.3 13 24-47 30
30  hou 0.42 -0.27 24 -0.55 29 8.6 8 29-42 26

Observations

  • The spread between the best and worst teams in terms of win probability surprised me.  For my NFL and NBA rankings, the GWP's tended to fall in the 0.25 to 0.75 range, whereas for these MLB rankings, the range is much tighter (0.40 to 0.60).  I think this has to do with starting pitching.  Imagine that the NFL played every day and that teams had to maintain a "starting rotation" of quarterbacks to allow each quarterback a few days rest between games.  If the Patriots could only play a third of their games with Tom Brady and had to make do with Ryan Mallett and Brian Hoyer the rest of the time, the league would get a lot more competitive in a hurry.  Conversely, if MLB only played games once a week and the Tigers could start Justin Verlander for every single game, you would expect to see some significant separation in relative team strength.
  • The correlation between win-loss record and these rankings is fairly weak. You have teams like Boston and Anaheim at the top of these rankings, despite mediocre records, and you have teams like the Dodgers and the Orioles below average according to these rankings, despite posting very strong win-loss records so far.

Next Steps

The next step is to publish the starting pitcher rankings, and then move on to adding the day to day visualizations of the ranking movements (similar to my Ticker feature for the NBA).

1 comment:

  1. Yeah, I find it interesting that only 2 teams are projected to win more than 90 games. If this isn't an anomaly, Vegas seems to be saying most teams with 90+ wins are lucky.

    Starting pitcher ranking sounds like a great idea!

    ReplyDelete