Let's Talk Sabermetrics

[wearinthefourfive]

OK, so this season I've been trying to step my sabermetrics game up, weeding out the bad ones and trying to find out which are actually pretty solid. So far I've just got the basics like how BA and RBI are bad stats to go by, and how OPS is probably the best way to value hitters. I'm still trying to understand the complex stats, like WAR or UZR. Any sabermetric pros out there want to drop knowledge?

 

[DeadsetAce]

you need to talk to osh kosh

 

[wearinthefourfive]

Why are you laughing? I'm frightened.

 

[DeadsetAce]

nah don't be. just that in every baseball discussion, he'll bring up these types of stats, and most of the people in the thread have no idea what he's talking about.

but he's a good guy and knows a lot, def someone you can learn a lot from.

 

[tim lincecum]

I need some schooling on this subject as well.

 

[doublejs07]

Originally Posted by Tim Lincecum

I need some schooling on this subject as well.

me too...

OKB loses me when he starts talking about Zone Rating, %'s of foul balls hit depending on which foot you put on the floor first when you wake up, etc etc....

 

[bijald0331]

OKB is the Dork Elvis of S&T. Word to Bill Simmons and Daryl Morey

 

[p mac one]

The key when dealing with advanced metrics is to have a context in which you're evaluating them.

Like for example, OPS is a great stat for getting the total offensive value of a player but SLG and OBP are valued equally in that stat. If you're comparing two leadoff hitters with a .750 OPS, you'd prefer a guy with a .400 OBP than a guy with a .400 SLG. There are a lot of people who would take a similar scenario where the guy with the higher SLG would say have an OPS slighly higher and stathead types would be quick to say "his OPS is better, therefore he's better." That's where it's important to know what you're evaluating and the context of that evaluation.

For me, I like to look at basically every stat in the book. Some "new" stats that are great are FIP and BABIP

 

[banks2pierce]

Simmons

This article breaks down a lot of the stats.

 

[djaman]

It's a progressive development in the statistical revolution, but I wonder if some of sabermetrics' key stats will ever make it into the vernacular of the baseball world. A couple of stats I like are BABIP (Batting Average on Balls In Play, which can indicate the luckiness or unluckiness of a hitter,) and FIP (Fielding Independent Pitching), which eliminates outcomes a pitcher cannot control like jam-shot singles and poor defense and evaluates what they can control like balls and strikes.) If a pitcher has a plus-FIP in comparison to his ERA, it possibly indicates that those uncontrollable variables skewed a pitcher's actual performance. I don't like stats like UZR (the numbers from year to year fluctuate too much, one season a guy like Carlos Gomez has the largest zone and the next he's below average) and WAR (just not certifiable proof a player's worth in my opinion.)

 

[djaman]

GD double-post.

 

[p mac one]

Everything I've read about UZR says that it's only valuable when used over 3+ years.

Also, WAR by nature is like a fantasy baseball type stat. Real life GM's (well the few that know what they're doing) can't make decisions based on stuff like that because they're operating in more of a vacuum than someone who goes on baseball-reference to see if Joe Mauer has more value than Albert Pujols based on salary and position and everything else. That said, it's a fun stat to keep an eye on for me, even though I don't really play fantasy baseball.

 

[juan baller]

In my opinion, the best offensive stat is RC/27 (runs created per 27 outs). It basically shows how many runs a lineup of 9 of the same player would score in a 9-inning game.

It's based on "run expectancy" of events, such as walks, stolen bases, strike outs, etc. By looking at data from past games, you can determine how often an particular event leads a run. If stealing a base leads to a run 20% of the time, then an SB should have a "run expectancy" of 0.2 runs in RC/27.

Here's a basic formula for RC/27 that shows what events have "run expectancy". The actual formula is a lot messier because of all the weighting that goes into to each variable.

Here are the Top 20 RC/27 seasons since 1990:

[table][tr][td]Player  [/td] [td]Year  [/td] [td]RC/27[/td] [/tr][tr][td]Barry Bonds [/td] [td]2004[/td] [td]22.46[/td] [/tr][tr][td]Barry Bonds  [/td] [td]2002[/td] [td]21.60[/td] [/tr][tr][td]Barry Bonds [/td] [td]2001[/td] [td]18.93[/td] [/tr][tr][td]Barry Bonds [/td] [td]2003[/td] [td]16.98[/td] [/tr][tr][td]Mark McGwire [/td] [td]1998[/td] [td]14.28[/td] [/tr][tr][td]Frank Thomas [/td] [td]1994[/td] [td]14.18[/td] [/tr][tr][td]Todd Helton [/td] [td]2000[/td] [td]13.68[/td] [/tr][tr][td]Larry Walker [/td] [td]1999[/td] [td]13.41[/td] [/tr][tr][td]Jason Giambi [/td] [td]2000[/td] [td]13.31[/td] [/tr][tr][td]Mark McGwire [/td] [td]1996[/td] [td]13.19[/td] [/tr][tr][td]Larry Walker [/td] [td]1997[/td] [td]13.18[/td] [/tr][tr][td]Manny Ramirez [/td] [td]2000[/td] [td]13.18[/td] [/tr][tr][td]Sammy Sosa [/td] [td]2001[/td] [td]13.16[/td] [/tr][tr][td]Carlos Delgado [/td] [td]2000[/td] [td]13.01[/td] [/tr][tr][td]Jason Giambi [/td] [td]2001[/td] [td]12.86[/td] [/tr][tr][td]Edgar Martinez [/td] [td]1995[/td] [td]12.67[/td] [/tr][tr][td]Larry Walker [/td] [td]2001[/td] [td]12.42[/td] [/tr][tr][td]Todd Helton [/td] [td]2004[/td] [td]12.29[/td] [/tr][tr][td]Jim Thome [/td] [td]2002[/td] [td]12.27[/td] [/tr][tr][td]Barry Bonds [/td] [td]2000[/td] [td]12.24[/td] [/tr][/table]

Notice no Ken Griffey or Alex Rodriguez. They both made too many outs to get their stats (similar to volume shooters in basketball). Griffey and A-Rod routinely made 450 outs in a season. Nobody in the Top 20 made more than 400 outs. Griffey's best RC/27 was 10.3 in 1994 (the strike-shortened season was the only time he topped 10 RC/27). A-Rod's best RC/27 was 10.5 in 2007. Neither sniffed the Top 20.

The Top 20 is dominated by sluggers. More specifically players that hit home runs and get on base at high clip.

Here are the Top 20 RC/27 seasons by players who hit less than 20 HR:

[table][tr][td]Player  [/td] [td]Year  [/td] [td]RC/27[/td] [/tr][tr][td]Chuck Knoblauch [/td] [td]1996[/td] [td]9.82[/td] [/tr][tr][td]Kenny Lofton  [/td] [td]1994[/td] [td]9.48[/td] [/tr][tr][td]Tony Gwynn [/td] [td]1997[/td] [td]9.16[/td] [/tr][tr][td]Paul Molitor [/td] [td]1994[/td] [td]8.74[/td] [/tr][tr][td]Rusty Greer [/td] [td]1996[/td] [td]8.73[/td] [/tr][tr][td]Dave Nilsson [/td] [td]1996[/td] [td]8.71[/td] [/tr][tr][td]Craig Biggio [/td] [td]1994[/td] [td]8.62[/td] [/tr][tr][td]Edgar Martinez [/td] [td]1992[/td] [td]8.56[/td] [/tr][tr][td]Todd Helton [/td] [td]2007[/td] [td]8.54[/td] [/tr][tr][td]Derek Jeter [/td] [td]2006[/td] [td]8.41[/td] [/tr][tr][td]Jason Kendall [/td] [td]1998[/td] [td]8.37[/td] [/tr][tr][td]Barry Larkin [/td] [td]1995[/td] [td]8.36[/td] [/tr][tr][td]John Kruk [/td] [td]1993[/td] [td]8.34[/td] [/tr][tr][td]Bill Mueller [/td] [td]2003[/td] [td]8.29[/td] [/tr][tr][td]Lenny Dykstra [/td] [td]1993[/td] [td]8.28[/td] [/tr][tr][td]Joe Mauer [/td] [td]2006[/td] [td]8.17[/td] [/tr][tr][td]Derek Jeter [/td] [td]2000[/td] [td]8.17[/td] [/tr][tr][td]Rickey Henderson [/td] [td]1999[/td] [td]8.15[/td] [/tr][tr][td]Todd Helton [/td] [td]2009[/td] [td]8.11[/td] [/tr][tr][td]Brian Giles [/td] [td]2005[/td] [td]8.10[/td] [/tr][/table]
I can't believe Rusty Greer is so underappreciated.

 

[ham city]

WHIP is a very relevant stat for pitchers, ^ that's the first time i've heard of RC/27 real interesting stuff.

 

[p mac one]

RC/27 seems to really favor guys who are Three True Outcome (BB/K/HR) players than anything else.