Need an Excel/Access expert for Fielding Neutralization project

Garlon · 08-01-2014, 10:33 AM

Hey everyone,

Since Spritze is not available to help me with this for a while, I was wondering if anyone else might be able to step up.

We are working on creating a new set of fielding neutralized stats for all players 1871-present. I have new formulas for improved estimation of defensive inningouts for 1871-1950, where we do not have real data for players in the database. I also have new sabermetric formulas for neutralizing PO/A/E/DP for each individual position.

Once we have fielding stats properly neutralized, we will have improved defensive modeling moving forward with historical simulations.

I just need someone who is very talented with databases to work with me on this.

Garlon · 08-08-2014, 02:29 PM

Here is a sample:

Brooks Robinson 1968 Real Stats
3B Assists: 353
3B PO: 168
Defensive Inningouts:4305

MLB 3Bmen
3B Assists:6724
MLB 3B PO: 2989
Defensive Inningouts: 87701

The average MLB 3Bmen in 1968 would make 330 assists while playing 4305 defensive outs. Robinson had 353, so he appears to be about 6.9% above average based on his raw statistics.

However, after adjusting for his team's Balls In Play against, assists by his team's catchers and outfielders, +/- excess of LHP, and his team's Defensive Efficiency Record, Robinson's Neutralized Assists go up to 381.

So in 4305 defensive inningouts played, Robinson would have made 51 more assists than the average MLB 3Bmen playing on a league average team. This puts Robinson's assist rate at 15.45% above average. Michael Humphreys, who wrote the book Wizardry, has Robinson at +23 runs saved for 1968. He values infield plays made above average at +.45 runs. 51 * .45 = 23 So this neutralization process yields the same result that Humphreys reports with his Defensive Regression Analysis system.

If you neutralize his PO, which in general should be thrown out for 3Bmen defensive evaluation, he ends up at +32 PO above average for 4305 defensive inningouts played.

He would have +83 PO+A above average. With the average play made above average being worth approximately .43-.45 runs saved, Brooks Robinson potentially saved his team 35 runs with his defensive play in 1968. Baseball Reference has him at +33 runs saved in 1968.

With teams scoring an average of 3.4 runs per game in 1968, those 35 extra runs saved are worth about 5 wins to his team. His 1968 season is the top defensive season by a 3Bmen in terms of runs saved according to baseball reference.

I'm hoping we can model defense in historicals in a way that recreates this type of impact on scoring in the simulation.

Scruff · 08-17-2014, 05:43 PM

I would love to help with this ... are you going all the way back to 1871? I'm pretty good with both Access and Excel.

Garlon · 08-17-2014, 05:58 PM

Yes, 1871-present.

08-01-2014, 10:33 AM	#1
Garlon Hall Of Famer Join Date: Jun 2004 Posts: 3,756	Need an Excel/Access expert for Fielding Neutralization project Hey everyone, Since Spritze is not available to help me with this for a while, I was wondering if anyone else might be able to step up. We are working on creating a new set of fielding neutralized stats for all players 1871-present. I have new formulas for improved estimation of defensive inningouts for 1871-1950, where we do not have real data for players in the database. I also have new sabermetric formulas for neutralizing PO/A/E/DP for each individual position. Once we have fielding stats properly neutralized, we will have improved defensive modeling moving forward with historical simulations. I just need someone who is very talented with databases to work with me on this.

08-08-2014, 02:29 PM	#2
Garlon Hall Of Famer Join Date: Jun 2004 Posts: 3,756	Here is a sample: Brooks Robinson 1968 Real Stats 3B Assists: 353 3B PO: 168 Defensive Inningouts:4305 MLB 3Bmen 3B Assists:6724 MLB 3B PO: 2989 Defensive Inningouts: 87701 The average MLB 3Bmen in 1968 would make 330 assists while playing 4305 defensive outs. Robinson had 353, so he appears to be about 6.9% above average based on his raw statistics. However, after adjusting for his team's Balls In Play against, assists by his team's catchers and outfielders, +/- excess of LHP, and his team's Defensive Efficiency Record, Robinson's Neutralized Assists go up to 381. So in 4305 defensive inningouts played, Robinson would have made 51 more assists than the average MLB 3Bmen playing on a league average team. This puts Robinson's assist rate at 15.45% above average. Michael Humphreys, who wrote the book Wizardry, has Robinson at +23 runs saved for 1968. He values infield plays made above average at +.45 runs. 51 * .45 = 23 So this neutralization process yields the same result that Humphreys reports with his Defensive Regression Analysis system. If you neutralize his PO, which in general should be thrown out for 3Bmen defensive evaluation, he ends up at +32 PO above average for 4305 defensive inningouts played. He would have +83 PO+A above average. With the average play made above average being worth approximately .43-.45 runs saved, Brooks Robinson potentially saved his team 35 runs with his defensive play in 1968. Baseball Reference has him at +33 runs saved in 1968. With teams scoring an average of 3.4 runs per game in 1968, those 35 extra runs saved are worth about 5 wins to his team. His 1968 season is the top defensive season by a 3Bmen in terms of runs saved according to baseball reference. I'm hoping we can model defense in historicals in a way that recreates this type of impact on scoring in the simulation.

08-17-2014, 05:43 PM	#3
Scruff Minors (Triple A) Join Date: Sep 2005 Posts: 243	I would love to help with this ... are you going all the way back to 1871? I'm pretty good with both Access and Excel.

08-17-2014, 05:58 PM	#4
Garlon Hall Of Famer Join Date: Jun 2004 Posts: 3,756	Yes, 1871-present.