FANALYTICS: Mayberry 3.0

For the previous three seasons, we've been using the Mayberry Method to evaluate and project Rotisserie skills. For those willing to abandon the faux precision of player projections and embrace wide error bars, Mayberry has turned out to be a very "accurate" method for judging talent.

But there is always room for improvement. There is one area that has not proven to be as "accurate" as I would like.

We have been using our Statistically Scouted Speed (Spd) score as a measure a runner's potential to amass stolen bases. The Spd formula looks at elements like beating out hits on softly hit balls, triples, runs scored and body mass index. It is essentially a measure of raw speed.

However, raw speed does not necessarily translate to the stolen bases that we play our game with. In fact, there are many "speedy" players who do not rack up mega-bags yet still score high in the Mayberry system.

There are two disconnects. The first is "opportunity." No matter how fast a runner is, he will not steal many bases if he is not given a green light. The second is a runner's ability to read a pitcher and get an adequate jump. You could be the fastest man on the planet but you will not steal many bases if you don't have those skills.

So we need to somehow incorporate our Stolen Base Opportunity rating (SBO) and stolen base success rate (SB%) into the process.

I have toyed with several advanced formulas that combine these two elements with Spd to create a new gauge. However, in the spirit of the simplicity of Mayberry, there is a much more basic formula that yields perfectly acceptable results while sacrificing little accuracy.

Roto Speed (RSpd) = Spd x (SBO + SB%)

Roto Speed (RSpd) = Spd x (SBO + SB%)

The (SBO + SB%) adjustment ranges from zero to more than 1.5. It stratifies the pool of fast versus slow players into much more realistic groups. For those players who have a perpetual green light, the adjustment elevates their raw Spd score by as much as 50% or more. For those who never run at all, they are more appropriately valued at or close to zero.

For all players with at least 250 plate appearances this year, SB% naturally ranges from 0% to 100%, though the spread is much narrower for those who run more. SBO ranges from 0% to 91%. The sum of the two stats ranges from 0% to 170%, with 98 of 264 (37%) players having adjustments that will increase their Spd score.

The current Spd score ranges from 34 to 182. With the adjustment, the new RSpd now ranges from 0 to 260.

Let's look at a few players at the extremes.

Rajai Davis pretty much runs all the time. His SBO is a ridiculous 91% and he has a success rate of 78%. Adding those two together yields a 169% adjustment to his Spd score. That elevates his previous 114 level (and a Mayberry rating of 3) to 193 (and a more appropriate Mayberry rating of 5).

At the other end of the spectrum are players like A.J. Ellis (Spd = 99), J.J. Hardy (95) and Miguel Montero (94), three players who have not attempted even one steal all year. Their current Spd scores give them Mayberry ratings of 2. With an SBO of 0% and SB% of 0%, their RSpd scores are now 0, yielding more appropriate Mayberry ratings of 0.

To accommodate the new scores and ranges, we also have to make an adjustment to the lookup tables that generate the Mayberry codes.

Here is the current table and the percentage of players who've achieved these levels over the past three years:

Spd           MM   Pct.
==========    ==   ====
0 - 49        0      3%
50 - 79       1     35%
80 - 99       2     23%
100 - 119     3     22%
120 - 139     4      9%
140+          5      8%

As you can see, the majority of players congregate in the middle with very few achieving the highest or lowest scores. But now...

RSpd          MM   Pct.
==========    ==   ====
0 - 39        0     18%
40 - 59       1     21%
60 - 79       2     22%
80 - 99       3     15%
100 - 119     4     14%
120+          5     10%

The new table flattens things out just enough. It still takes good skill to merit a 4 or 5, but now 24% of players can achieve that level as compared to 17% before. And the split between the lower three and upper three Mayberry ratings is still about 60-40.

Who benefits most from these changes this year?

                                       Mayberry
Player       Spd  SB%  SBO  RSpd  Var  Old  New
===========  ===  ===  ===  ====  ===  ===  ===
Gomez,C      132  86%  74%   211  +79   4    5
Davis,R      114  78%  91%   193  +79   3    5
Gordon,D     140  79%  67%   204  +64   4    5
Cabrera,E    112  96%  52%   166  +54   3    5
Schafer,J    116  76%  68%   168  +52   3    5
Crisp,C      112  89%  56%   161  +49   3    5
Stubbs,D     140  82%  51%   186  +46   4    5
Upton,B.J.   126  82%  45%   160  +34   4    5
Victorino,S  123  84%  41%   154  +31   4    5
Venable,W    122  77%  48%   153  +31   4    5 
Casilla,A     83  94%  40%   111  +28   2    4

Who loses the most?

                                        Mayberry
Player        Spd  SB%  SBO  RSpd   Var  Old  New
===========   ===  ===  ===  ====  ====  ===  ===
Schumaker,S   114   0%   2%     2  -112   3    0
Ellis,AJ       99   0%   0%     0   -99   2    0 
Martinez,JD    97   0%   2%     2   -95   2    0
Hardy,JJ       95   0%   0%     0   -95   2    0
Montero,M      94   0%   0%     0   -94   2    0
Morse,M        91   0%   2%     1   -90   2    0
Soto,G         89   0%   0%     0   -89   2    0
Youkilis,K     89   0%   0%     0   -89   2    0
Plouffe,T      93   0%   7%     6   -87   2    0
Pierzynski     86   0%   0%     0   -86   2    0
Ransom,C       88   0%   4%     3   -85   2    0
Barmes,C       85   0%   3%     3   -82   2    0
Polanco,P      81   0%   0%     0   -81   2    0
Crawford,B     78  20%   8%    22   -56   2    0
Francoeur,J   101  33%  12%    46   -55   3    1
Thames,E      119  50%   6%    66   -53   4    2
Bloomquist,W  147  41%  24%    95   -52   5    3
Young,M       108  50%   4%    58   -50   3    1 

We'll be launching these new codes in the 2013 Baseball Forecaster.

We're taking advance orders now.
 


Click here to subscribe

  For more information about the terms used in this article, see our Glossary Primer.