FAQ - Modern Stats and Player Evaluation

phillychuck 2005-02-05 15:35:58 UTC #1

The purpose of this thread is to list some principles and reasoning often employed here in baseball discussion threads. The purpose is NOT to suppress discussions of issues, but rather to offer information that has repeatedly been presented here (and on the now defunct phillies.com board) over the last 6 years. If you disagree with any of this in general, or in particular situations, please feel free to continue the discussion (albeit logically and rationally, supporting whatever points you make).

I've used lots of things I've personally written in this thread (because I can find them easily and I can edit them for continuity without offending the original author ) and the individual posts are quite long. Please PM me with anything you think should be incorporated that you've written or read here, and I'll try to add it (I reserve the right to edit for clarity and continuity).

People have suggested just linking to off-site resources, but I'd like this thread to be a compilation of some of the things we've written here, or things that have been written by people who post here.

This thread is for reference and therefore will be locked.

dsc25 wrote the following, an eloquent appeal for the use of more than just the naked eye in resolving baseball issues:

Q: What's wrong with using the naked eye to evaluate baseball players?

A: There's nothing wrong with using the naked eye and personal observation as one tool to evaluate baseball players as long there are other tools in the toolbox as well. And, many would argue, counterintuitive to the traditional evaluation crowd, that the naked eye should be one of the lesser used tools. Why? Because in baseball the naked eye deceives with its selectivity.

There are so many games played, so many plays within each game, and so many individual efforts within each play, that it is nearly impossible to see everything that a player brings to the game. And that's assuming you watch every play of every game. Watching only a select number of games and with the limitations of watching only the players involved in the play at the moment of their involvement limits the naked eye as well.

On the other hand, statistical analysis attempts to provide a picture of what the player does over the course of the entire season during every play that he is involved. There are certainly deficiencies in that analysis (like the attempt to develop a fielding standard), but the attempt is to alleviate the problem of selective perception based on the naked eye by counting and measuring a player's actual performance over the course of all the games in which he is involved.

An excellent quote from Bill James (quoted in Moneyball (pp. 68-69)) sums this up perfectly:

[quote]Think about it. One absolutely cannot tell, by watching, the difference between a .300 hitter and a .275 hitter. The difference is one hit every two weeks. It might be that a reporter, seeing every game that the team plays, could sense that difference over the course of the year if no records were kept, but I doubt it. Certainly the average fan, seeing perhaps a tenth of the team's games, could never gauge two performances that accurately--in fact if you see both 15 games a year, there is a 40% chance that the .275 hitter will have more hits than the .300 hitter in the games that you see. The difference between a good hitter and an average hitter is simply not visible--it is a matter of record.

But the hitter is the center of attention.[/quote]

We notice what he does, bend over the scorecard with his name in mind. If he hits a smash down the third base line and the third baseman makes a diving stop and throws the runner out, then we notice and applaud the third baseman. But until the smash is hit, who is watching the third baseman? If he anticipates, if he adjusts for the hitter and moves over just two steps, then the same smash is a routine backhand stop--and nobody applauds.

phillychuck 2005-02-05 15:38:16 UTC #2

EXPLANATION OF MANY SABERMETRIC STATS USED IN VARIOUS PLACES ON THE BOARD

Originally, with discussion at:
philliesphans.com/phorum/vie ... hp?t=16967

Q: What are considered traditional baseball stats?

A: Traditional baseball stats usually consist of counts of things (like RBI, AB's, IP's, etc.) or simple ratios (BA, ERA, HR-rate, etc.). Some of the most common are:

Stats that attempt to measure offensive performance: AVG, Runs, RBI's, Hits, Walks, Strikeouts, etc. Some of these are pretty good (AVG, and Walk Rate for example) but the counting stats like RBI and, to a lesser extent, Runs tend to be overvalued by the casual fan (and some baseball organizations). They receive a lot of airtime but are team-dependent to a large extent (for example, if a trained Iguana hit behind three guys with .400 OBP's he could get 100 RBI's).
Stats that attempt to measure pitching performance: Wins, Losses, ERA, Walks, Strikeouts, Saves, Holds, etc. ERA is a very good traditional stat (except for relievers that pitch a lot of partial innings), but Wins, Losses, Saves, and Holds all suffer from external influences.
Stats that attempt to measure fielding performance: Putouts, Assists, Errors, FPCT, etc.

Traditional stats are often worthwhile and germane to an argument, but they rarely provide the best evidence for any interesting argument.

Q: Why do many posters cite modern stats instead of traditional stats?

A: Modern baseball stats aren't really all that modern. They trace back to Branch Rickey (progressive Cardinals' GM in the 20's and 30's, Dodgers GM of the 40's, who I think invented OBP (on-base percentage)and isolated power, at least conceptually). Bill James is considered to be the father of most of the modern baseball stats, and he coined the term sabermetrics to describe the modern, statistical measurement of baseball performance.

So-called modern or sabermetric stats attempt to measure more complicated things, often things that directly impact fundamental questions like 'which of these players is better?' or 'which of these teams is likely to win this year?' They often accomplish this by combining several traditional stats mathematically and analyzing the effectiveness of the stats by complicated statistical methods. These stats (if posters understand them and agree about their accuracy) frequently provide better evidence on interesting questions than traditional stats.

Q: What are some of the more commonly cited modern offensive stats?

A: Here are a few:

OPS-this is the modern stat you'll see cited most frequently (not a Bill James stat). It is derived from adding on-base-percentage, OBP (a very good traditional stat that some argue is modern) and slugging-percentage, SLG. It's a quick way to get a reasonable measure of a player's overall offensive contribution without doing any complex math. It's become pretty universal, now available on ESPN's stat pages and baseball-reference and many others. A decent OPS for a corner OF or 1B is .800, a good one is .900, and a great one is 1.000 or more. Players in other positions have different levels of good or great depending on how challenging their defensive position is (this brings up the concept of the defensive spectrum, a James invention). Shortstops and catchers get the most leeway, followed by 2B, 3B, and CF. Jim Thome's 2003 OPS was .958, and David Bell?s was .579 (lowest in baseball for his number of AB's)

AVG/OBP/SLG-some people avoid using OPS and give batting stats this way. A very good offensive player has a .300 BA, a .400 OBP, and a .500 SLG as a corner OF or 1B (again, some slack given to harder defensive positions).

Runs Created-a Bill James creation that combines offensive stats to measure overall offensive contribution. It can be approximated by multiplying OBP by SLG by AB's. (OPS is a quick and dirty attempt to approximate this) There now exist very complex formulas for getting the exact answers for players (one of the most popular of these is the linear weights methods authored by Pete Palmer, which attempts to combine stats linearly with proper weighting coefficients). Regardless, Runs Created (RC) correlates very, very well with team runs and has a huge pile of evidentiary support. The best offensive players in a league in a given year create about 110-130+ runs, and most regulars create about 60-90. Jim Thome created 125 runs in 2003, while Jimmy Rollins created 76 (good for a SS).

RC's enable a baseball fan to get at least an approximate idea of a player's effect on winning, by using the next entry.

Pythagorean Win-Loss estimates-team winning percentages correlate very closely to team run totals in the following manner: (runs squared)/(runs squared) + (runs allowed squared)] is approximately equal to winning percentage. In modern baseball, that means a player must create or prevent about 9-10 runs to win an additional game.

Equivalent Average (EqA)-another overall measure of offensive performance intended to read like batting average. An average player is .260, good is .280, great is .300, etc. Clay Davenport (Baseball Prospectus) is the inventor (and posts the numbers on the net for free). Jim Thome's EQA for 2003 was .321, Pat Burrell's was .253 (bad for a LF, but not as horrible as some might think), and David Bell's was .216.

Park factors-adjustments for the relative tendency for parks to favor hitters or pitchers. The most important of many adjustments intended to refine offensive and defensive measurements. Baseball-reference and most other modern statistics sites often list park-adjusted and league adjusted figures. The Vet was pretty much a neutral park for most of its history, but has tended to be a pitchers' park in recent years. Coors is a great hitters' park (as pretty much everyone in the world knows) and Shea Stadium is a great pitchers' park. CBP seems to be a good HR park, and favors offense at least slightly (doubles and triples are depressed, at least partially offsetting the HR increase) as of the end of the 2005 season.

Replacement level (or replacement level player) (*from dsc25)- a player who is assumed to be freely available to any team when a fill-in player is needed. Different analysts have different methods of determining the exact parameters of this hypothetical replacement level player, but they all have the same basic concept: the level of production from freely available talent. It's important to consider how many runs a good player produces above that level, not above zero, for many reasons. RARP (Runs above replacement position) adjusts RC's for park, position, and replacement level, and is again available on Clay Davenport's page at Baseball Prospectus.

Stolen Bases-analysts focus on stolen base percentage rather than total number of bases. Runners must steal at least 2 out of 3 bases to add to their offensive value, otherwise the caught-stealings offset the SB's. Analysts recognize a subtle side benefit of SB attempts and other small-ball techniques is that they tend to even out run production from game to game by exchanging big innings for multiple one-run innings.

Rules of thumb: Modern statistical analysts use the many rules of thumb for measuring offensive performance. Here are a few: a batter ideally should walk at least 1 time for every 2 K's, should walk at least 1 time for every 10 AB's, should post at least 70-100 OBP points through walks and HBP's. (Walks are very important to sabermetricians) OBP is somewhere between 40 and 70 percent more important than SLG. Batter's strikeouts don't strongly correlate positively or negatively with runs created, and therefore are usually ignored.

Q: What about situational stats?-I hear them mentioned on broadcasts pretty often.

A: Managers and players use batter-pitcher match-up stats pretty often, and broadcasters often take situational stats to a ridiculous extent (by using arcane situations with very small sample sizes). Lots of statistics are now available on the internet for specific situations. Modern analysts find some of them significant, and some not so. Modern analysts strongly believe in the platoon differential (so lefty/righty situational stats are valued, since they seem to remain somewhat consistent from year to year). Analysts also value home/road splits, and sometimes day/night splits, and monthly splits. Almost all modern analysts completely discount so-called clutch stats like BA/RISP or late-innings-close. These stats are wildly divergent from year-to-year for players and teams, and are considered the product of small sample sizes and luck rather than ability.

Many people in the stat community are watching AVG and OPS split by COUNT fairly closely these days, especially on first pitch and two strike counts. Allegedly, Joe Kerrigan is a big fan of splitting batters' performance by count in order to adjust pitching approach.

Q: What about pitching stats?

A: There's not as much room for improvement in pitching stats, since ERA is already pretty good. Here are a few interesting ideas:

WHIP-(walks plus hits per inning pitched) this stat is seen frequently and is not accorded very high value in the sabermetric community, in part, because of its connection with fantasy baseball (it was one of the original eight categories in Rotisserie League Baseball, I think). It makes a good companion to ERA, which, especially for relievers, can fluctuate fairly dramatically due to lucky or unlucky breaks. Great pitchers have WHIP?s below 1.200, good pitchers between about 1.200 and 1.300, average 1.300-1.400; worse than 1.400 is pretty mediocre. Millwood's WHIP was the best among starters on the 2003 Phils team at 1.224 (11th in the NL), with Padilla and Wolf also among the league leaders at 1.240 and 1.245.

BAA or OBA-(batting average against or opponent's batting average)-a good stat, easy to interpret. Millwood, .232; Padilla, .230; and Wolf .219 were the Phils 2003 numbers.

OPSA-(OPS against) a better stat than BAA but harder to find (although it is available on ESPN). Examples from the 2003 Phils: Millwood, .691; Wolf, .713; Padilla, .726. Anything under about .740 is pretty good for an NL pitcher.

K/BB and K/9-rather than just count strikeouts, statisticians prefer to compare them to walks and to innings pitched. A good pitcher has a K/BB ratio above 2/1 and a K/9 IP ratio of at least 5 (James did a famous study where he compared pitchers who were equally successful in ERA and Wins but who diverged on K/9 rates and found that pitchers with K/9 rates below about 4.5 almost never had subsequent success.) Minor league followers use both K/BB and K/9 extensively to evaluate prospects. Top prospects should have K/9 of 9 or over in the minor leagues. Brett Myers' minor league K rates were low for a top prospect, and that is why many stat-oriented Phils' Phans are still skeptical about him in 2003 (but hopeful, seeing his K-rate improved dramatically).

The next two stats are 'improvements' for ERA's

Component ERA-a recent Bill James stats that uses the same general principle of RC (above) to estimate how many runs a pitcher 'should' have given up in the innings he pitched. This is a slightly better predictor of pitcher's subsequent performance than pure ERA.

DIPS ERA ($ERA, sometimes Indicated ERA)-perhaps the most controversial modern baseball stat, DIPS ERA is the brainchild of Voros McCracken (inventor of defense independent pitching statistics, thus DIPS), based upon the counterintuitive but statistically well-supported hypothesis that pitchers have little to no control on what happens to batted balls in play (other than HR's). DIPS ERA is calculated solely from the three 'true' pitching outcomes (BB, K, and HR allowed), with adjustments for quality of defense and park effects mathematically recombined with the true outcomes to produce a stat that mirrors ERA. DIPS ERA is a much better predictor of pitchers' subsequent performance than simple ERA. Millwood's DIPS ERA in 2003 was 3.71, Padilla's was 4.19, and Wolf's was 4.36. Cormier's was 3.12.

PRAA (pitcher runs above average), PRAR (pitcher runs above replacement), Runs Prevented, ERSAA (earned runs saved above average) etc.-these and other stats try to measure how many runs a pitcher saves versus average or replacement. Park adjustments and defensive adjustments are employed in many of these stats. ERSAA for 2003 Phillies pitchers were 24 for Cormier, 16 for Padilla, 13 for Adams, 8 for Millwood, and 2 for Wolf.

ARP (adjusted runs prevented)-used for relievers, an interesting stat derived from play-by-play accounts of the games which calculates runs prevented by employing data tables for expected runs (how many runs SHOULD score from an inherited bases loaded, no outs situation, for example, versus how many runs did score). Data tables are park and team adjusted, but NOT adjusted for lineup position, which would require an immense amount of data. Cormier's ARP was 30.2, Mesa's was -19.8, and Adams was 10.1.

Leverage Index-a tangotiger (well known internet sabermetrician-see links below) idea for comparing the contributions of relief pitchers, this one also derived from play-by-play accounts and focusing on relief pitcher's increased value due to how the pitcher is used in game situations. All game situations are weighted by leverage, average being 1, the highest (leading by one run, runners on second and third, one out in the ninth) being 10. Ratings come from comparisons of win expectation grids. Closers have the highest leverage indices in general, approaching 2 (meaning each inning the closer pitches is worth 2 'normal' innings). From 1999-2002 the highest leverage index belonged to Troy Percival at 1.87, Jose Mesa was at 1.43, and Tim Worrell was at 1.03.

Two other interesting stats that are mentioned occasionally: Quality Starts (QS's) and game score. A quality start is when a pitcher goes 6 or more innings with an ERA of 4.5 or under-this gives a quick and dirty idea of how often a starter keeps his team in the game. Game scores are computed from a table of rules (adding points for good things, subtracting for bad), with 90-100 point games being just about perfect, and an average start being around 50. Additionally, a useful rule of thumb for adjusting ERA is that a pitcher should be held responsible for about half of the unearned runs he gives up.

Q: What about fielding stats? Aren't they pretty meaningless?

A: There are no fielding stats that describe fielding skill broadly and show as much year-to-year correlation as the best offensive stats, but some modern fielding stats are pretty good. Even the old classic Fielding Percentage (FPCT or PCT) gives a decent idea of what it's intended to measure, how sure-handed and accurate-throwing a fielder is, and correlates well from season to season and with wins. Here are some of the modern stats:

Range Factor (RF)-a simple count of the number of plays made per game (a Bill James stat). This was the first attempt historically to get at the difference between a high-percentage fielder and one who got to a lot of balls. The argument for its significance is that since almost all fielding percentages are quite high, the best and worst percentage fielders are separated by 10 or 15 plays per season, but the difference in total chances can be as high as 100 plays or so in the same season. RF is subject to many external influences, and thus is more important historically than as a current analysis metric.

Zone Rating-a new counting stat, in which multiple scorers at each ballpark assign all batted balls to zones and count the number caught in each zone by each fielder. Fielders are then assigned percentage ratings based on the number of balls converted into outs in each zone with respect to number of balls that could reasonably be fielded (according to very specific criteria). ZR's are pretty good defensive stats, but have several weaknesses-they omit line drives and pop ups for IF-s, contain no adjustment for DP pivot ability, do not consider outfielders' arms, and are not applicable historically. Judgments are still too subjective (which balls 'should' be fielded?) and zones are not optimally shaped (twenty six pie shaped wedges).

UZR (ultimate zone rating) and UZR Runs-an improved zone rating obtained by using more and better shaped zones, categorizing errors, and adjusting for the run values of outs and unplayed balls. Still not historically available, and still flawed with respect to IF fly balls, outfielder throwing, catching thrown balls, and pivot ability. It's still probably the best widely accepted and available fielding metric intended to measure range, certainly the best yet constructed from the STATS, INC. play-by-play data-its year-to-year correlation for fielders is about what BA is for hitters (good, but not as good as OPS and the various RC's). Mitchel Lichtman, the inventor, (and the other folks at baseballprimer.com) are still in the process of tweaking the stat to improve correlation, but improvements at this point are small.

Defensive Win Shares (DWS's)-a new Bill James stat customized by position, a subset of Win Shares (below). Defensive Win Shares measures fielding on an arbitrary 4-3-2-1 weighting of four varying factors for each position, and uses creative, team-wide and league wide methods to get at proxy measurements for several hard to quantify categories (How good is a 1B at catching bad throws? How does the OF range affect IF assists?, etc.). A disadvantage is DWS's are not controlled for playing time, so they must be adjusted for part-time players like Marlon Byrd in 2003.

Probably the best way to employ fielding stats is to look at all of them for a particular player, and compare that player to the other players in the league. I'll do it here for 2 Phils (mostly in 2002, since 2003 numbers are not available to me in many cases):

Jimmy Rollins

Fielding Pct. .980 (T1/15)
Range Factor 4.58 (8/15)
Zone Rating .861 (4/15)
UZR .724 (13/15) (2003)
DWS's 6.49 (6/15)

Bobby Abreu

Fielding Pct. .981 (7/10)
Range Factor 1.92 (10/10)
Zone Rating .929 (1/10)!
UZR .755 (source unsure) (5/8) (2000-2003 weighted)
DWS's 2.17 (9/16)

Q: Are there any modern team-level stats (other than simply computing the ones above for a whole team)?

A: Team level stats don't get quite as much attention-offensively, team runs is a great summary stat, and ERA or runs against is a great summary defensive stat. At the team level, one tends to look at deviations-in other words, does the team score more or less runs than it should (via runs created) or win more or less games than it should (via Pythagorean projections)? The Phils have been on the bad end of these stats lately, while the Braves have been on the good end. One team stat that gets some play is

DER (defensive efficiency record or rating or ratio)-this is just the (sometimes park adjusted) percentage of balls in play that are converted into outs by the defense. It is considered a very good measure of how good a defensive team is, much better than total errors. The Phils' unadjusted DER for 2003 was .7160 (meaning 71.6% of balls hit into the field of play against them were converted into outs), 5th in the league for the second consecutive year, above the NL average of .7101.

Q: Are there any all-encompassing stats that give a good idea of how effective any player is, taking into consideration offense, fielding, and pitching?

A: Probably the most frequently cited 'overall' stat is Bill James' Win Shares (WS).
Each win share represents 1/3 of a team win, and is obtained from evaluating offensive, defensive, and pitching contributions with various modern stats. The Phils were led in WS's last year by Jim Thome with 30 and Bobby Abreu with 28. League leaders were Albert Pujols with 41, Barry Bonds with 39, and Gary Sheffield with 35. Eric Gagne was the league's leading pitcher with 25. Win shares are controversial outside the sabermetric community because of their complexity, and inside the community because of the seemingly arbitrary nature of some of James's choices (such as excluding loss shares and the choice of zero-level).

Q: What links can I use to find these stats, and the methods for calculating them?

A: For the more mainstream stats, espn is pretty good -- sports.espn.go.com/mlb/statistics

Here are baseballprospectus's articles (best sabermetric site):
baseballprospectus.com/current/

Lots of info is at baseballprimer (but you must search through the articles):
baseballthinkfactory.org/

Traditional stats (and some modern ones) for all players historically at baseball-reference:
baseball-reference.com/

Win shares at:
baseballgraphs.com/ --for 2003
baseballtruth.com/bbt_winshares.htm --for 2002

Other interesting sites:
baseball1.com/c-sabr.html --Sabermetric Articles
geocities.com/tmasc/ --tangotiger on baseball
diamond-mind.com/weblog/index.htm --diamondmind baseball
retrosheet.org/ --best historical day-by-day info
mb2.theinsiders.com/fbaseballfrm8 --a pure sabr board-pretty deep stuff
bluemanc.demon.co.uk/basebal ... tracts.htm --baseball contracts (not sure this is up-to-date anymore, but contracts are now on usatoday)
sports-wired.com/players/default.asp --stats with minor league numbers
baseballamerica.com/today/Stats/index.html --absolute must for minor league fans
hardballtimes.com --home of some of the best baseball writing anywhere (especailly Aaron Gleeman's stuff)

julio 2006-10-21 21:21:37 UTC #3

I don't know what happened to the item on stikeouts...PC?

In that item's absence, donmoney posted the following in another thread. Until such time as we get a FAQ item back up, this provides links to the better discussions of the issue on philliesphans.

Thanks, DM!

phillychuck 2006-10-22 00:09:46 UTC #4

The FAQ was in this thread, and I didn't delete it...bug in the system somewhere?

I'll check to see if I have a backup copy.

phillychuck 2006-12-20 19:22:14 UTC #5

DO STRIKEOUTS MATTER? DO GOOD TEAMS GET MORE PRODUCTIVE OUTS THAN BAD TEAMS?

This discussion stems from comments like "the Phils strike out too much to be a good offensive team" or "Bobby Abreu would be a much better player if he didn't strike out so much." Everyone agrees that it is better to put a ball in play in most situations than to strike out. That isn't the point of discussion here, the point is whether strikeouts are WORSE than other kinds of outs--that is, do lots of strikouts indicate bad hitters or bad teams, and do good teams get lots of "productive outs" (GBs and FBs that advance runners)?

In many particular situations, a strikeout is worse than most other kinds of outs...man on second, nobody out, there are many types of outs that will help the offensive position of the team more than a strikeout. We all knew that, we all see that on a daily basis, and we're all so conditioned by that situation and similar ones that when the question arises, "Do batters' strikeouts matter?" or "Do team strikeouts matter?" we immediately flash back to the "failure to advance the runner" scenario where it clearly DOES matter, and say "lots of strikeouts are a bad thing."

HOWEVER, we ignore or fail to perceive the hidden data from other situations and other at bats--a strikeout is a better out sometimes than a hard grounder (DP), or a hard line drive (DP), or a soft FC grounder where the lead runner gets cut down, or a shallow fly ball where the runner is thrown out advancing. We also don't directly see other positives--guys who strikeout a lot often share two other characteristics--they swing hard, resulting in high SLG, and they take lots of close pitches (resulting in K's AND BB's and high pitch counts) which results in high OBP and early entry to the opponent's bullpen. This is a form of perception bias--we see the obvious evidence and its association with the strikeout (guy K's, runners don't move) but we miss the subtle evidence (guy who swings hard and often K's also hits 52 Dongs, and walks 120 times BECAUSE of his approach at the plate), or at least we don't associate that evidence directly with K's.

So strikeouts DO matter (negatively) all the time in specific situations, but over the course of the year the associated contributions of high-strikeout hitters in other situations and in other, non-strikeout at-bats outweigh their failure to make more productive-type outs. That's why EVERY scientific study has found that high strikout teams from about 1921 forward score slightly MORE runs than low strikeout teams.

Here's a brief study by Shore, and some commentary from him:

Since 1982, teams have averaged 947 strikeouts, and 726 runs. I broke all of those teams into two groups: whiffed MORE than 947 times and whiffed 947 times or LESS.

Who scored more runs?

The "MORE" teams. By a lot.

The 276 "MORE" teams averaged 753 runs per year, and 1056 strikeouts.
The 300 "LESS" teams averaged 701 runs per year, and 847 strikeouts.

Strikeouts, for batters, correlate as a slight positive for run scoring, throughout history. It's PROBABLY due to the fact that many of the best (read: Most Productive) hitters strike out a lot. By best, I mean "do the most things to make runs score". That includes, specifically, getting on base, and driving in runners on base.

How does one do that? The 2 most common ways to reach base are, obviously, getting a hit, or a walk. Hits are obvious... but how do you get a walk? You work a deep count. You take pitches. You do things that ALSO lead to striking out more. Driving in runners is a function of batting average and slugging percentage. Frequent hits, and hits for extra-bases, are good things. Despite the occasional genetic freak, MOST people who are hitting for extra bases are also striking out more, due in part to the violent nature of their swing.

There you have it... the basic take & rake philosophy. Both halves lead to more strikeouts, but also more baserunners and more power. Increased Ks, increased runs.

An important point to remember is that "cutting down" on their swing would more than likely impact a player's ability to hit for power. Thus, fewer strikeouts, more balls in play, less power.

To reiterate, no one is claiming that in particular circumstances other outs are not more productive than strikeouts (thus the sabermetric terminology "productive outs"). What is being claimed is that in some circumstances throughout the game, strikeouts are more valuable than other kinds of outs (DP grounders being the obvious example, but there are others), AND, most importantly, that the approach to hitting that leads to more strikeouts also leads to more of certain undeniably good things, like BB's and HR's.

An important assumption in this argument is that hitters like Jim Thome can't become Placido Polanco at will. This is unproven, of course, and its opposite is part of baseball lore ("take two and hit to right", etc.). The modern ballplayer, however, prides himself in consistency of approach regardless of the count or situation (see Ted Williams' writing and Mike Schmidt's comments on hitting).

I like to say to people who write posts that say things like "every major league ballplayer should be able to bunt" or "Thome shouldn't strike out that much, Dimaggio didn't, and Bonds doesn't" that the game is not as easy as they seem to think. Being able to bunt or make contact with tough pitches is a skill (remember Richie Ashburn!), just like hitting 500 foot homers is. Just because you can do one doesn't mean you can do the other; in fact, power hitters almost never bunt well since they've never been asked to do so in games, from little league on up.

Another problem with cutting down on swings in specific situations in that while you may be more successful in producing productive outs, you'll likely be less successful in getting productive homers and doubles. When Abreu homers with runners on first and third and one out, no one says, "Gee, I wish he had cut down on his swing and hit a Sac Fly instead!" As dsc25 mentioned in a different thread title, you have another case of perception bias--the strikeout is mentally linked to a specific failure, but the homerun has no obvious mental link to the batting approach that leads to the strikeouts.

What stat folks are saying is that a batter's K-rate, or a team's hitters' K-rate, is NOT indicative of their offensive ability, while a pitcher's K-rate, or a pitching staff's K-rate IS a pretty good indicator of pitching ability. In other words, people who say a certain hitter strikes out too much, or the Phils in particular strike out too much, are full of baloney.

Here's why:

Let's look at two hitters, call them Larry and Jim, and see how they approach their craft. They both recognize that every time they come to the plate it's a good thing if they can put the ball in play. It gives them a chance of getting a hit, a chance of advancing the runner, a chance of forcing a fielder to make a misplay (not a large factor in the big leagues, but much more significant when a player's younger, which contributes to the misperceptions). Larry gets pretty good at this, rarely striking out, and playing small ball well, but to do this he finds he has to choke up on the bat, cut his swing down further with two strikes, swing a bit more early in the count, etc., but that doesn't bother him, because he's a 5'9" shortstop with below average arm and hand strength, and he wasn't hitting lots of shots to the alley or over the fence anyway. And pitchers generally challenged him when he was ahead in the count because they knew he was very unlikely to hit anything more than a single anyway. His batting average on balls in play goes down a bit, but he puts lots and lots of balls in play leading to a reasonable batting averge overall. That plus his fielding give him a major league job.

Jim, on the other hand, is a big strong farm boy, and he finds out that if he cuts down his swing and chokes up, he does put the ball in play more, but his home-run numbers plummet, he doesn't drill as many balls into the gaps, and his BA on balls in play drops. He also finds that if he swings at borderline pitches early in the count it does help him cut down on strikeouts, but he walks less and he can't hit those pitches out of the park. Jim decides that he'll be patient and swing hard when he gets a pitch he can drive, accepting the fact that he's going to strike out more than Larry, leading to a career as a slugger who hits HR's and walks a lot. Good thing, too, because his glove wan't going to keep him in the big leagues.

For these two theoretical players, Larry might have the higher BA and Jim would probably have a good OBP and higher SLG, so the high strikeout total for Jim would be more than offset by the increase production in extra bases and walks. (And, in reality, Jim's BA on balls in play could be very much higher than Larry's, so Jim's overall batting average could be higher than Larry's, too.)

But, even if the two players' production numbers were identical (lets say each hit .300, posted a .400 OBP and a .500 SLG), all the strikeout totals would represent was which approach was necessary for each player to achieve those numbers. If Larry K'd 20 times a year and Jim K'd 180, they'd still be almost equally productive offensively (see dsc's thread on productive outs recently for the arguments as to why).

So, in summary, all batter's Ks indicate is how hard and agressively a batter has to swing to produce the OBP and SLG he does. The high K hitter with a .900 OPS is about equally as valuable as the low K hitter with the same numbers.

Pitchers, on the other hand, are nearly always trying to strike out a hitter (some sinker-ballers are exceptions, but even sinker-ballers have to be able to strike hitters out at a reasonable rate to succeed over time). For a pitcher, there are almost no bad side effects to striking out a batter (I'll deal with pitch count later), so getting the strike out is an unmitigated good thing. (unlike avoiding the strike out is for the hitter, who has to sacrifice power and walks to REDUCE K's).

The relative constancy of batting average on balls in play for pitchers makes it very difficult for the pitcher to reasonably say, "I'm going to try to hit more bats and strike out less players, and I'll cut down my pitch count and still be equally effective." That's because with certain rare exceptions, hitting more bats causes a pitcher to give up more hits, which is a bad thing (unless the pitcher is able to reduce walks or HR's allowed simultaneously, and that's not easy to do. (Part of this is tied up with how hitters approach the game--if the pitcher makes a pitch intending to draw a ground ball (low and outside corner, for instance), or a pitch intending to jam a hitter, lots of hitters will just not swing at it, leading to more called strikes and, ultimately, more strikeouts, which once again associates strikeouts with effective pitching.)

Bottom line--if two pitchers have the same ERA, and one has dramatically more K's than the other, the one with more K's is much more likely to succeed subsequently, meaning he's actually the better pitcher. K's appear to be a true reflector of pitching ability.

The result of all this is that batter's K's, for individuals or teams, tend to be (counterintuitively) slightly positively correlated with run scoring (the more K's, the slightly more runs a team scores, or an individual batter produces). But pitcher's K's, on the other hand, strongly correlate with runs prevented on both a team-wide basis and an individual (ERA) basis.

When a stat person says "batter's K's aren't bad," that's all he means--lots of batter's K's usually indicate teams with power and patience (often good offensive teams or players), and power and patience are good things. He doesn't mean that in any particular at bat, a K isn't a bad thing for the hitter. But lots of pitcher's K's almost invariably indicate a good pitcher.

dsc25 adds that strikeouts are reflective of skill-sets:

A pitcher who strikes out a lot of batters tends to have a skill-set that enables the pitcher to make batters swing and miss more often as well as fool batters, so that even if they do make contact, the contact isn't so great. Plus, the pitcher who strikes out a lot of batters has better control of the strikezone, thus probably decreasing walks (although that's obviously not always the case).

For hitters, the skill-set that comes along with striking out a lot can be anything. It can be someone like Doug Glanville who just stinks. Or, it can be someone like Bobby Abreu or Jim Thome who have such great command of the strike zone that they don't swing at pitches they know they can't drive but are still strikes. Also, the two of them swing at pitches so hard sometimes they are bound to miss (this is baseball after all). The great command of the strikezone maps to walks (beneficial) and the power of their swing maps to extra base hits (also beneficial).

So, the way I think about it, strikeouts for hitters can mean anything from being downright awful to excellent, and anywhere in between. Strikeouts for pitchers, while not perfectly mapped to good skills, do imply that the pitcher is probably doing something right that is going to be reflected in other parts of the pitcher's game.

ZP chimes in with the following commentary on expanding the strike zone in RBI situations:

Iâ€™d like to add something that somebody told me about "expanding the strike zone" with a runner in scoring position. We've all heard the traditionalists say that a good hitter should "expand the zone" and try to get the run home, rather than take the walk if he doesn't get anything.

These same people say the same three things when a player gets what looks to be a solid hit, yet turns out to be a routine out.

It jammed him.
He hit it off the end of the bat.
He just got under it.

"Expanding the zone" would just lead to a greater occurance of any of these.

Plus, it'd make the pitcher throw these types of pitches.

I always thought it was stupid when a player says that it is his job to get a runner on third home with less than two outs. No, it is not his job to get that runner home, rather it is his job to create the best possible chance for that runner to get home. In many cases, taking the bad pitches and walking is much more effective (and productive) than trying to just hit it in the air.

This year, ESPN began listing productive out percentage (roughly, the % of outs that advance a runner) on its stat pages, and dsc25 made this mid-season observation:

ESPN should be ashamed for promoting this concept. Here's another in a long line of [evidence] showing that a productive out has nothing to do with winning or scoring runs.

Here are some of the findings from BP:

[quote]
- The 15 teams above average in POP have scored 4.77 runs per game, have a .334 POP and a .468 winning percentage.

- The 15 teams below average in POP have scored 4.74 runs per game, have a .270 POP and a .532 winning percentage.

- The top five teams in POP have scored 4.33 runs per game, have a .351 POP and a .392 winning percentage.

- The bottom five teams in POP have scored 4.74 runs per game, have a .230 POP and a .534 winning percentage.
[/quote]

So, evidence shows that, bluntly, teams that have high percentages of productive outs don't score more runs, and that teams that have lots of strikeouts do, marginally at least. Counterintuitive, but not after you think about it a bit.

And, finally, BP did a large sample study supporting all of this:

baseballprospectus.com/artic ... cleid=2617

phillychuck 2011-10-08 13:51:51 UTC #6

There are a lot of contentions that seem to arise, many from new posters, that involve Bobby Abreu, the Phillie with the longest continuous service on the 2006 team. In this FAQ I’m going to try to respond to those contentions in a factual balanced way, but, to be honest, the reader should be aware that I’m going to struggle to find factual support for many of those claims.

Bobby Abreu is not a winner, and therefore should be traded

In his Phillies career, it is undeniably true that Bobby Abreu’s teams have never reached the playoffs—8 seasons, no rings. But, in his first major league season with Houston (in a limited role), the Astros won their division, and Bobby played on two minor league championship squads. Two things about this argument stand out as illogical to me: first, in baseball, we do not hold any one player responsible for a team’s failures, and most especially we do not blame the team’s best player. Should the 1972 Phillies have traded Steve Carlton? Should the 2000 D’Backs have traded Luis Gonzalez? Should the mid-to-late 70s Philies have traded their core of Schmidt, Luzinski, Bowa, Boone, etc? There are hundreds of other examples, of course, of good or great players who don’t win until late in their careers, if ever. Second, in two of the last three years, management decisions and bad play by teammates have been much more responsible for the Phils barely missing the playoffs. For example in 2003 and 2005, David Bell was arguably the worst regular player in baseball, and cost the Phils enough games that merely replacing him with a player that just substantially below average would have put them into the playoffs. As allentown wrote:

The comedy gets better and better. Stats are worthless unless you win. So a star surrounded by insufficient talent is dreck. Don't blame the manager, don't blame the GM. If a guy is the team's star it is his job to make sure the team wins regardless of overall team talent and managerial competence. A baseball player who raises the game of his teammates significantly? Highly doubtful.

The Phillies were within one game of the playoffs last season. They had more than enough talent to make it. Bobby played more than well enough for them to make it.

Let's see, where can I find one more win? Yanking pitchers to soon. Manuel has said he yanked pitchers earlier because he thought the relief corps was stronger than the starters. But he had some very good starters who could go longer on occassion and had thrown few pitches through 6 and were yanked anyway. The pen wore down. I think I can get a win there.

Padilla and Bell were injured in ST and brought back almost immediately at start of season without normal rehab. Polanco/Utley and Floyd were doing well so they weren't needed. They stank early on. The mishandling of Floyd coincided with his confidence nose-dive and stinkitude. I think I can get a win there.

Thome was hitting .200 with no power for months. The Z-boys were a disgrace. Yet Manuel/Wade went into interleague play using Thome and a Z-boy for 1B and DH. Thome was known to be hurt. Howard would have done much better. Another 100 AB for Howard last year is at least a win.

We stuck with Worrell right up to the point he said he had too many mental woes to pitch. Management didn't sense a problem? Just trotted him out there. That may be another win.

The talent was there to reach post-season last year. The managerial and GM smarts were not.

Abreu is not clutch

This is a claim made by many, many fans, possibly inspired by some people in the local media who have harped on this point. The factual support for this claim is primarily anecdotal, although if you restrict “clutchness” to just ABs in the last inning with the Phillies tied or down one run, Abreu had a bad 2005. However, taking a larger sample, and expanding that very narrow definition of clutch, Abreu’s been just fine. hoyalawya wrote:

And Rob Neyer responded on espn.com:

Abreu is lazy, and doesn’t play to his potential

This is a particularly subjective claim, perhaps stemming from Ed Wade and other Phillies officials proclaiming Abreu as a future batting champ and 40/40 player. While it’s possible his potential is even greater, we need to start by looking at where his performance places him historically. DS wrote this about his all-time rankings:

Hard to argue that his potential is even higher, right? And players who don’t work hard (are lazy) usually show signs of it during their development. I wrote this about his early career:

”phillychuck”:

To fill out a bit of the missing evidence, here is a link to Abreu's minor/major league stats:

sports-wired.com/players/pro ... ?Name=BDCA

He was signed at 17 and played in the GCL for Houston, and rose step-by-step through the Houston system until he reached AAA in 1995 at age 21. He had hit around .300 at every level, but had not shown a lot of power or outstanding plate discipline (though it was good). he played full AAA seasons in '95 and '96, getting a cup of coffee at the end of '96. He didn't play well. He played another half-season of AAA in '97, then spent a half season in Houston, where again he hit poorly for a corner OF. At this point in his career he looked like a classic AAAA player, a good AAA guy who couldn't make the jump (although his plate discipline improved markedly in '96 and '97 in AAA). Houston let him go in the expansion draft to Tampa Bay, who then moved him to the Phils for Kevin Stocker, who had proven to be a reliable fielding, weak hitting SS since his breakout in '93.

So Houston and Tampa Bay in December '97 had let Abreu go for a) nothing and b) a SS who was barely above replacement level. This guy is now among the top 3 RF's in the NL, but some say he isn't as good as he could be, that he's too lazy to live up to his potential.

I seriously don't understand this.

Mvpsoft has repeatedly pointed out Abreu’s incredible durability:

Abreu’s stats may look good, but they’re empty numbers

Many claim he hits an “empty” .300. Bill James originated the concept of secondary average to measure the non-BA contributions of hitters. Babby Abreu is 8th among active players in that stat and 23rd all-time.
(baseball-almanac.com/hitting/hisecavg1.shtml)
That seems to be the exact opposite of empty, especially when you factor in his 40 SB each year (with a high percentage, too!). allentown (again) deals with this argument based on Abreu’s below average (for him) 2005 season:

The top 10 BA hitters in the NL, with a minimum of .313 BA were Pujols, Helton, Renteria, Sheffield, Kendall, Giles, Castillo, Loretta, Podsednik, and Lieberthal. All of them exceeded Abreu's average of .300. So if Abreu's .300 is a ridiculously empty .300, all of these guys should be vastly better than Abreu as offensive players, right?

1. Since batting average is viewed as less important than obp, let's look at obp. Abreu had an obp of .409. The top 10 BA players in NL had obp of .439, .458, .394, .419, .399, .390, .381, .372, .379, .373. So Abreu is better than all but 3 of the top 10 NL batting average leaders.

2. What about slugging percentage? Again, Abreu had an off-year with a SLG of .468. The other 10 log in at .667, .630, .480, .604, .416, .526, .397, .441, .443, .453. So Abreu is below the midpoint, but certainly not showing the least power among the top 10. Castillo is the lightest hitter and also had an obp significantly below Abreu, so perhaps he can be said to be the emptiest .300+ hitter.

3. Let's look at runs scored. Abreu has 99. The top 10 scored 68 to 137 runs. Four of the top 10 scored fewer runs than Abreu and a 5th was tied with him.

4. Let's look at rbi. Abreu has 101. The top 10 have 39 to 132, with Castillo at 39 and Kendall and Podsednik tied at 58. Only 4 of the top 10 batting average leaders have more rbi than Abreu.

Other posters have posted Abreu's high OPS with RISP.

So again, on what basis can it be said that Abreu is an empty .300 hitter?

A couple of short claims:

Abreu is a poor defender.

His range is clearly below average by most measures, and he is wall shy by almost unanimous consensus. (Though not all agree that’s bad…) On the other hand, he has a top-flight arm, and has won a gold glove (which means someone respects his fielding ability somewhere). This is probably a valid criticism, at least the most valid among all those here.

Abreu takes too many walks with RISP.

Hardball Timesdid a compilation study of all players in Sac Fly situations, and Abreu was one of the top three (in SFs above expected), and his additional SFs came from reducing BBs and Ks, indicating he’s MORE aggressive with a guy at third, at least.

hardballtimes.com/main/artic ... ice-flies/

Shore added this GWRBI info:

and added that he was in the top 20 in all of baseball for the three year period 2001-03.

Abreu steals in bad team situations, taking the bat out of Thome’s hands

hoyalawya again did the legwork on this:

Of Abreu's 40 steals in 2004, 7 came with no outs, 9 with one out, and 24 with two outs. This seems to help your point, but let's take a look at those 24 steals more closely...

In only 14 occassions was Thome at bat when Abreu stole with 2 outs. On the other occasions, Thome was either not in the lineup, did not immediately follow Abreu in the batting order, or had already made an out with Abreu on base before Abreu stole.

Of those 14 occassions, then, when Thome batted after an Abreu 2-out steal, guess how many times Thome was walked (either intentionally or unintenionally)...Three.

Those three instances led to a Ricky Ledee 3-run homer, a Chase Utley RBI HBP and a David Bell RBI single.

Therefore, the total # of times in 2004 that Abreu stole with two outs, took the bat out of Thome's hands and cost the Phillies a run....(drumroll please)....ZERO!

Abreu is an interesting talent--for his career he has the following full-season averages:

2Bs--41
HRs--23
RBI--95
BBs-107
SBs--30 (with only 10 CS)

His most comparable year-to-year modern players are Fred Lynn, Dave Parker, Reggie Smith, Larry Walker, and Bernie Williams, all of whom were also under-appreciated during their playing days. Maybe it will take Abreu's departure for the fans to realize what an excellent player he was (and still is) for the Phils.