The *PLAYER* class has three types; BATter, PITCHer, and FIELDer. So far we’ve covered the former two with this model but haven’t addressed the *FIELDER* *PLAYER* type. I haven’t figured out how to do a value ranking for fielding. Sabermetrics tried to calculate this but I have issues with their methods and soon I’ll be running some numbers to demonstrate how inaccurate their measurements can be.

Baseball games are won and lost based upon runs given up and runs scored. Runs given up are the purview of pitching and runs scored are the responsibility of batting. So where does fielding fit into this? A great fielder can rob hits from opposing batters. As we calculated in a previous post, every two hits is equivalent to a single run. Thus, theoretically, for every hit a fielder saves his team means 0.5 runs didn’t score. The problem here is that runs that didn’t score cannot be reliably counted like a run that did score.

“Advanced” Sabermetrics use zone data to determine a fielder’s worth. How valuable is a great fielder and exactly how many opportunities does your average position player get where their superior fielding turned a hit into an out. My assumption is that these opportunities are negligible but I’ll run some studies using the newly released 2013 event data from retrosheet.org in future posts.

For now I divide fielders into 3 categories; Great, OK, and Liability. A GM should be concerned about players who fall into the latter category. If that player generates a lot of BATting or PITCHing value (pitchers are FIELDers too) maybe having them as a fielding liability is acceptable. The ERROR stat has been maintained in box scores since the beginning of baseball. An official score keeper will determine whether a fielder should have made an out but didn’t.

The following table lists MLB fielders sorted by the most errors showing how many unearned runs they gave up because of those errors. This table was derived from event data. A script goes through and notes an error, tabulates that against a player, then counts all associated unearned runs.

Name_tm | #Errors | #Uruns |
---|---|---|

Pedro_Alvarez_PIT | 27 | 13 |

Alexei_Ramirez_CHA | 22 |
23 |

Starlin_Castro_CHN | 22 | 11 |

Ryan_Zimmerman_WAS | 21 | 15 |

Ian_Desmond_WAS | 20 | 10 |

Pablo_Sandoval_SFN | 18 | 4 |

Pedro_Florimon_MIN | 18 | 17 |

Daniel_Murphy_NYN | 18 | 19 |

Jed_Lowrie_OAK | 18 | 9 |

Conor_Gillaspie_CHA | 17 | 11 |

Chase_Utley_PHI | 17 | 12 |

Juan_Francisco_ATL | 3 | 1 |

Juan_Francisco_MIL | 14 | 6 |

Anthony_Rendon_WAS | 16 | 4 |

Alberto_Callaspo_ANA | 11 | 9 |

Alberto_Callaspo_OAK | 5 | 4 |

Josh_Donaldson_OAK | 16 | 3 |

Jonathan_Villar_HOU | 16 | 8 |

Mike_Moustakas_KCA | 16 | 15 |

Matt_Dominguez_HOU | 16 | 12 |

Erick_Aybar_ANA | 15 | 4 |

Jordy_Mercer_PIT | 15 | 8 |

Brandon_Crawford_SFN | 15 | 7 |

Jean_Segura_MIL | 15 | 15 |

Adeiny_Hechavarria_MIA | 15 | 6 |

From this table you can probably figure out fielders that pose a liability to a team. Normally every error results in around 0.6 runs. I’ll get more into this ratio and how it varied throughout history in a future post. Alexei Ramirez for the Chicago White Sox led the league by letting up 23 runs through his 22 errors. These are real runs that scored against the White Sox. Either Alexei was unlucky (possible) and this is a statistical anomaly or he has panic issues where he commits errors in the worst possible situations. You can’t make any conclusion from this number other than perhaps the top players in the above list pose a liability. Considering Pedro Alverez posted a +4.2 WAA perhaps that makes up for the 13 runs he let score.

I don’t have a good handle on how to model fielding for OK and great fielders and to what extent fielding should impact their overall value. Sabermetric and WAR people do lump fielding into their overall rating which I think is wrong. Since fielding is its own class type it should have its own value and it can’t be tabulated like pitchers and batters. Throughout the winter I’ll be doing some more counting studies on the 2013 event data to see what is the best way to represent this and try and determine exactly what value a great fielder provides over an OK or Liability fielder.