Sabermetrics: 10 best stats & more

Top 10 Statistical Acronyms

Last week, I invited sabermetric questions from the readers, and I will answer the first one.  Iowacowboy asked: Could you name the top 10 (or so) sabermetric abbreviations or acronyms . . . and then define them in simple terms . . . and don’t forget to add the ones that have been modified using a plus sign or similar change... Good question.  Here's my take on the Top 10.  You can get more information on these and other statistics in the glossary at Fangraphs.

1. wRC+ Weighted runs created is a good overall measure of a player's hitting.  The plus added to the acronym means that the statistic is shown as a ratio to league average (100).  For instance, a 110 wRC+ means the player's runs created were 10% above average, and 90 means 10% below average. Keep in mind that this is based on the average of all position players, and is not based on the average for a position.  For instance, the average wRC+ for shortstops is likely to be below 100.  This statistic is also adjusted for ballpark.  Runs created, like the statistic wOBA, is based on weighting offensive events (e.g., walks, hit by pitch, single, double, triple, HR) by its relationship to scoring runs.  The weights are recalculated each year, as shown in the Fangraphs' "guts" table.

2.OPS+ On Base % Plus Slugging % (OPS) was an early advanced stat for hitting,  but its use in sabermetric research has waned with the advent of wRC+ and wOBA, which are more accurate.  Adding OBP and SLG together is somewhat arbitrary, unlike the more precise weightings of the newer stats.  That said, OPS is a decent approximation of offensive performance in most cases.  For most players, OPS+ is nearly the same as wRC+.  However, players whose value is disproportionately related only to OBP or SLG will have larger deviations between OPS+ and wRC+.  OPS+ is used by Baseball-Reference.com, but not Fangraphs.   If you are using splits data available only on B-Ref, you may have no choice but to use OPS+ instead of wRC+.  OPS+ is ballpark adjusted and adjusted for run environment (meaning that OPS+ can be compared across eras).

3. FIP and x-FIP These defense independent pitching stats are used by Fangraphs.  Fielding Independent Pitching (FIP) and x-FIP are intended as a measure of a pitcher's performance unaffected by the defense behind him.  FIP is based on K%, BB%, and HRs per 9 innings.  This reflects the pitching outcomes which are most within the control of the pitcher.  x-FIP normalizes actual HR rates to the league average HRs per flyball.  This reflects the belief that a pitcher's home runs per fly ball will regress to league average.  x-FIP is sometimes used as a forward looking version of FIP.  These stats are scaled so they can be compared to ERA.

4. SIERA This is skill-interactive ERA, which you can find among the advanced pitching stats at Fangraphs.  SIERA uses a more complex formula than FIP and x-FIP.  SIERA reflects both the defense independent characteristics of x-FIP and additional batted ball characteristics which explain why pitchers are successful.  In many instances, a pitcher's SIERA is nearly the same as his x-FIP.  However, SIERA is a better metric for some pitchers who rely on getting outs by suppressing or inducing certain types of batted ball.  SIERA is more predictive of next year ERA than either FIP or x-FIP.

5. RE24 This stat is from the win probability family, and it stands for Run Expectancy - 24 base-out states.  There are 24 base and out possibilities (e.g., runner on 1st and 3d with 2 outs is one base-out state).  For the outcome of each plate appearance, RE 24 calculates the resulting  change in probability of scoring one or more runs in the inning.  RE 24 sums the player's impact on expected scoring over a season and reflects a runs above average value.  RE24 can be used to examine the contributions made by both hitters and pitchers.  Unlike other advanced hitting stats, this stat recognizes the value of productive outs.  REW is the same as RE24, except it is expressed as wins instead of runs.

6. WAR Wins above replacement player is a handy measure of a hitter's or pitcher's overall contribution to wins.  WAR encompasses a hitters' defense, hitting, and baserunning.  WAR for pitchers is based on defense independent pitching and the effect of leverage on the bullpen.  Wins are expressed relative to a theoretical replacement player, meaning minor league players who would be called up to replace openings on the roster.  Baseball-Reference (b-WAR) and Fangraphs (f-WAR) use different data for calculating WAR.  For instance, B-Ref uses DRS (defensive runs saved) and Fangraphs uses UZR (ultimate zone rating) to measure defense.  Pitching tends to show the largest deviation between bWAR and fWAR.  That's because Fangraphs uses FIP for pitching WAR, while bRef subtracts the results of defensive metrics from a pitcher's runs allowed.

7. ISO ISO (isolated power) is a measure of a hitter's extra base power.  ISO is calculated as SLG% minus batting average.

8. MD and SD MD is meltdown and SD is shutdown.  Both are relief pitcher stats from the win probability family of stats at Fangraphs.  In my view, these two stats are superior to saves, blown saves, and holds as a measure of relief pitcher success.  These two stats are  based on the change in win probability from the beginning to the end of a relief pitcher's stint.

9. DRS, UZR, TZ These are the three most used advanced defensive stats. DRS is defensive runs saved; UZR is ultimate zone rating; and TZ is total zone.  DRS and UZR probably are more accurate than TZ, because of the use of granular zones around each position on the field to measure how well fielders convert batted balls to outs.  TZ has the advantage of providing defensive estimates for the era prior to 2002.

10. BABIP Batting Average on Balls in Play is a remarkably useful statistic for measuring the sustainability of a hitter's or pitcher's performance.  BABIP includes fair batted balls but excludes HRs; the denominator excludes at bats which end in strikeouts, walks, HRs, and hit by pitches..  BABIP requires a very lengthy period of time to stabilize as a statistic.  As a result, BABIP for both pitchers and hitters tends to be volatile (i.e., subject to substantial random variation from year-to-year).  Average BABIP for pitchers is concentrated in a relatively narrow range between .290 and .300.  Average BABIP for late inning relievers may be somewhat lower.  Typically pitchers are expected to regress in the direction of league average BABIP.  The range of sustainable BABIP for hitters is larger than for pitchers. This is because hitters have a more personalized level of sustainable BABIP, based on their batted ball characteristics and hitting skill.  Various x-BABIP (expected BABIP) formulas are used to estimate the liklihood of a hitter's regression. Alternatively, if a hitter has an extensive BABIP history, his career average BABIP can be used as the regression target.

Now that we've gotten through the alphabet soup, I'll include two bonus sabermetric tidbits.

Astros' Defensive Shifts

Last year, the Astros led the major leagues in net runs saved by the shift, with +27 according to the Fielding Bible's DRS system.  I wrote about it here. How are the Astros currently faring with the defensive shifts?  Before I answer, let's get the required small sample size warning out of the way.  It's very early, and defensive stats will fluctuate over the course of a season.

The Astros (5 runs saved by the shift) and Rays (6 runs saved by the shift) lead the American League in defensive shift productivity.  The Pirates lead the National League.

Teams With Fast Starts

A few days ago, Dave Cameron wrote an interesting piece on the Mets' fast start to the season. His thesis was that the Mets' great start to the season had come against a weaker part of its schedule.  The part of his analysis which is most fascinating is comparing pre-game winning odds to the actual results.  Fangraphs publishes winning odds before each games.  The odds take into account the starting lineups for that game, thus accounting for injuries as well as the luck of the draw in terms of the actual starters that day.  So, for the Mets' example, the odds indicate that the Mets should have a .current .535 win percent, based on their opposition.  The Mets have still beaten the expectations.  But not by as much as it appears from the gaudy record. The Royals' fast start has also come against a weaker part of the team's schedule. This kind of comparison provides some context for evaluating how much the fast starts are likely to improve the season's projected W/L record.

Keep in mind that this article was written several days ago, and the evaluation of schedule strength may have changed a little bit since then.  The Astros, Mets, and Cubs are teams which had losing records last season but have gotten off to great starts to the season.  However, one difference is that both the Cubs and Mets are making hay in games that the pre-game odds said they should win. The Astros, on the other hand, were expected to have something like a .470 win percent during this part of their schedule (according to fangraphs' pre-game odds), but they have clearly done much better, with more than a .600 win percent.

We can see why the Astros were expected to lose games on the road trip, given the quality of the opposing teams and the fact that the Astros are the visiting team.  So, it's somewhat encouraging that the Astros have put wins in the bank against teams that were supposed to beat them.

Also, a commonality among the Astros, Mets, and Cubs is that all three teams are directed by sabermetric-oriented front offices which were brought in to rebuild the teams from the bottom up.