TomG's Robot Professional MLB Betting

[TomG]

At +149 I felt like we were on the right side of the Boston game but we lost in the 11th (results oriented thinking, I know). Maybe once I figure out run line calcs we would have played Bos +1.5 instead FTW!

Updated YTD:
84.49 units wagered
+20.68 units
+24.5% ROI

[basementproject]

Thanks for the link, Irish. RC is a decent measure of individual hitters' performances, in its more complex variations, definitely.

There's a bunch about RC in a few Bill James abstracts, as well as the Baseball Prospectus paperback... Which by the way, is worth every penny of the 20 dollars it costs. Highly reccomended.

It's funny, I never thought something as simple as OPS could give a good picture of a team's hitting performance either, but as I said earlier, when I took TeamOBP+TeamSLG values, and ran a correlation against runs scored for each respective team, the coefficient was 0.97, which is near perfect.

[crockpot]

[ QUOTE ]
[ QUOTE ]

What is the formula for xFIP? I can't seem to find it.


[/ QUOTE ]

It's park factor adjusted FIP. Hardball Times has it, but I imagine the actual formula would be very difficult to calculate manually since it must take into account venue for each start/appearance.

[/ QUOTE ]

actually, the adjustment is only for home runs. since HR/flyball regresses heavily to the established mean for the pitcher's home ballpark--much like BABIP--FIP can still be full of 'noise' over a small sample.

[crockpot]

[ QUOTE ]
i use dips as well, when i emailed murray himself he said that he uses FIP because he found that the correlation was stronger for FIP vs. XERA..

[/ QUOTE ]

odd, his 2007 book endorses the xERA formula. (although it does also mention FIP) maybe he just didn't want to rewrite everything.

[kyleb]

So, I bought this book and it helped me with the model I'm already constructing.

This book is really, REALLY thorough.

[TheRover]

Interesting discussion.

xFIP, I want to have your babies.

[TomG]

[ QUOTE ]
There's 16 games scheduled for Tuesday. But only one game that I like...

Tuesday MLB
Detroit at Texas
Texas +130
Risk 1 unit to win 1.30 units

I see this game much closer to a coin flip. Robertson and Padilla appear about equal. I give Detroit the edge offensively, but even assuming Robertson goes 6 innings, Detroit's bullpen is enough to give the overall pitching edge to Texas. Throw in home field advantage and I like Texas with the overlay.

[/ QUOTE ]

Pitching change. Padilla scratched and replaced with Rheinecker leaving us with no action for today. [img]/images/graemlins/frown.gif[/img]

On the plus side, The Hardball Times has xFIP stats for all pitchers from 2004 onward that I will be work on incorporating...

[irisheyes]

One thing I have noticed is that the difference between FIP and xFIP is often very large, e.g.


2007 Dan Haren 1.70 ERA / 3.38 FIP / 4.13 xFIP

Use at your own risk.

[irisheyes]

Thanks, Crock. If the only adjustment is for HR/flyball at the pitcher's home ballpark, that would seem to introduce a lot of noise in small sample sizes. I realize that statistically half of their starts will be at their home ballpark, but over a small sample (like 1/3rd of the 2007 season) there haven't been enough starts to properly 'regress to the mean'.

Just thinking about how to properly weight FIP v. xFIP for smaller sample sizes. And if HR/fly @Home is the only adjustment for xFIP, I think I have my answer.

[AllinDan]

[ QUOTE ]
basement, OPS is not well regarded, but mostly because as a single number it captures little about a hitter's performance.

The formula discussed here comes out of the Murray book, but the below link is a solid reference.

http://en.wikipedia.org/wiki/Runs_created

[/ QUOTE ]
in the model I'm working on I'm using EQA rather than Murray's OBP*SLG*.975.

basics on EQA