xERA caculations

[tedpower]

The 2008 Baseball Forecaster lists the 2007 league average for xERA - listed seperately for both the NL and AL in the "2008 Cheater's Bookmark". It has the AL at 4.34 and the NL at 4.37.

I am calculating an xERA for all of MLB in 2007 as 4.79! I double checked - it is not an error in my formula - I am extremely close to the xERA listed in the playerlink on all the pitchers.

So then I did 2006. The 2007 Baseball Forecaster lists the xERA for the NL as 4.06 and the AL 4.11

I calculate a 2006 xERA of 4.74 for all of MLB!!!

Are the league xERA's calculated differently than idividual players because this is not even close????????

I continued to go back to 2003 and here was the data that I ended up with:

2003 MLB: 4.40 ERA, xERA 4.72
2004 MLB: 4.46 ERA, xERA 4.69
2005 MLB: 4.29 ERA, xERA 4.67
2006 MLB: 4.53 ERA, xERA 4.74
2007 MLB: 4.46 ERA, xERA 4.79
2008 MLB: 4.32 ERA, xERA 4.67

This brings another question to mind. It appears to me that xERA is clearly too pessimistic. Shouldn't it be adjusted??

[Tom_T]

I'd say your formula is wrong. Obviously the MLB value can't be 4.79 if the league values are 4.34 and 4.37.

What do you get if you calculate the AL or NL value yourself?

Where are you getting the raw G/L/F numbers?

[Michael@HQ]

Replicating BBHQ's xERA calculation is do-able but not easy. I had a thread last year that has far too many details.

My suggestion is (1) make sure you are using the xERA2 definition, the one by Burnson, because that's what is in production (2) read through that thread to see if any of the suggestions that helped me correct my calculation also works for you. I don't think I can email you my xERA spreadsheet -- my home computer crashed since early 2007 so I believe this is one of the many files I lost.

[tedpower]

I'd say your formula is wrong. Obviously the MLB value can't be 4.79 if the league values are 4.34 and 4.37.

What do you get if you calculate the AL or NL value yourself?

For 2007 NL I get xERA of 4.68
For 2007 AL I get xERA of 4.73
For 2007 MLB I get and xERA of 4.69 (sorry mistyped it as 4.79 earlier)

Where are you getting the raw G/L/F numbers?

all data is from BaseballReference.com - league pitching splits

[tedpower]

Replicating BBHQ's xERA calculation is do-able but not easy. I had a thread last year that has far too many details.

My suggestion is (1) make sure you are using the xERA2 definition, the one by Burnson, because that's what is in production (2) read through that thread to see if any of the suggestions that helped me correct my calculation also works for you. I don't think I can email you my xERA spreadsheet -- my home computer crashed since early 2007 so I believe this is one of the many files I lost.

thanks for the link Michael, i have been meaning to go back and make my spreadsheet formula exact and I think I will have to do that this weekend. As is, I am using xERA2 - and on individual players I am almost identical to what is listed in the PlayerLink.

Does anyone else have there own spreadsheet where they can run the league data and help me figure this out?

[Tom_T]

I've made some progress on this. It looks like the individual xERA calculations take the final result and increase it by 10 percent. My spreadsheet works for individuals only when I include this extra factor.

However, I think the league calculations in previous Forecasters did not use this factor. When I calculated the AL xERA, I got a high number as well. When I removed the 10 percent factor, I got a much more reasonable 4.27.

[Michael@HQ]

I sure don't remember an extra "increase it by 10%" factor not mentioned in the xERA2 glossary definition.

[Tom_T]

I sure don't remember an extra "increase it by 10%" factor not mentioned in the xERA2 glossary definition.

I am going to get more information about this from our formula gurus.

[RON@HQ]

The variables in these formulas are -- just that -- variable. We tweak them all the time. 2008 was a down offensive year so the data used last year doesn't work anymore. Rarely, it requires a complete revamp. Sometimes, it's just a matter of doing a minor global adjustment, like perhaps a 10% across the board. We start by plugging the league-wide totals in the current formula and see how close it comes to ACTUAL BA and ERA. Then we adjust from there. Nothing magic.

I routinely look over xBA and xERA during the Baseball Forecaster production process and see exactly what we need to do. That's what is going on now.

[Hapj]

Where are you getting the raw G/L/F numbers?

This is the most interesting question of all.... my line drive could be your flyball, there is bias in observation.

[RON@HQ]

This is the most interesting question of all.... my line drive could be your flyball, there is bias in observation.

Yes, and this is exactly where variances are most likely to occur. We get our data feed from Baseball Info Solutions, who compiles the data independently, and thus reports their findings often differently, from ESPN or STATS or a host of other data providers. Some don't even separate out line drives. BIS even breaks line drives down further....