“That there are so many blessedly different ways to the same end is why we watch the games. It’s why winning baseball will always enrapture us to no end.” – from the epilogue of Winners
Correlation is not causation—so the phrase that every college statistics student has drummed into his or her head goes. And of course that’s true, except when it isn’t, and one of those times is the annual correlation of the start of spring training and a spate of new baseball books.
As pointed out by Dave Studeman, this spring there seems to be more than the usual number of new books, and today I’ll discuss one of them, which features a subtitle that’s a wee bit condescending: Winners: How Good Baseball Teams Become Great Ones (and It’s Not the Way You Think) by FoxSports.com and Baseball Prospectus contributor Dayn Perry.
I’ll begin with a short overview of the book and then move on to highlighting and addressing a few of the specifics to which Perry draws our attention.
Who are the Winners?
In a nutshell, Perry examines the 124 teams that made the postseason between 1980 and 2003 (excluding the strike-shortened 1981 and 1994 seasons) and attempts to cull from their collective success a few pearls of wisdom regarding what strategies and tactics lead to winning and which are perhaps overrated or nonexistent.
The various strategies and tactics are laid out in a series of chapters where Perry uses modern sabermetric analysis—complete with references to Bill James and Voros McCraken (our own Dave Studeman and Maury Brown earn nods as well)—to come to his conclusions, with heavy reliance on Baseball Prospectus analysts such as Nate Silver and Michael Wolverton, and, of course, their player evaluation tools Value Over Replacement Player, Pitching Runs Above Replacement, and Support Neutral Lineup-Adjusted Value Added For Pitchers.
And those chapters are …
Perry concludes with a final short chapter on the role of luck, and an excellent epilogue that summarizes his conclusions by putting them in the context of a fictitious team—a touch that I wasn’t expecting and that neatly summed up the book. For those who’ve read Mind Game, you’ll note the synopsis of the final chapter “Beat the Yankees, Be the Yankees,” also written by Perry.
At the same time, however, the book contains a number of short profiles of players from those teams, ranging from Phil Neikro of the 1982 Braves, to Willie Hernandez of the 1984 Tigers, to Cesar Cedeno, who I had almost forgot played a pivotal role for the 1985 Cardinals down the stretch, and many, many more. In each portrait Perry provides a quick career path of the player, often including interesting back stories like the troubles of Pedro Guerrero and Dwight Gooden off the field, and the personalities of Ben Oglivie, Graig Nettles and Gorman Thomas.
In fact, for a fan like myself, whose first-hand baseball memories date from only a few years before 1980, this was perhaps my favorite aspect of the book because the concise and well-written profiles recalled to mind many of the teams, players and moments that contributed to my initial interest in the game.
Overall, this structure works well and allows Perry to both make the points that will, at the very least, make the “stathead” crowd think, as well as fashion an entertaining book for more casual fans.
But of course since this site is frequented by many of the former variety, I’d be remiss if I didn’t point out a few of the more interesting statistical conclusions that Perry draws from his data set, peppered with a few of my own criticisms of his approach.
Where Art Thou OPS?
Right out of the gate in The Slugger (or, Why Power Rules), Perry includes a nice overview of the frequent discussion regarding the relative importance of pitching, defense and hitting, and comes to the correct observation that, of the three, hitting is the most important, because 50% of the game (run prevention) is made up of pitching and defense, while the other 50% is offense. As a result, a single offensive player can usually have more impact over the long haul than a single pitcher or fielder.
He goes on to show that of the basic rate stats—batting average, slugging percentage, on-base percentage and isolated power (which he calls isolated slugging percentage)—slugging percentage correlates more closely with run scoring than the rest, and that therefore it is the most important.
While I don’t argue with the approach, I do have some subtle criticisms (call them nitpicks if you will) of the general presentation of the material.
First, the table he uses on page 13 to make his case is broken down by eras that include 1871-1900, 1901-1925, 1926-1950, 1951-1975 and 1976-2000. In doing so he found that SLG’s correlation with runs scored in the 1976-2000 era was .868 while OBP was .811.
These eras seem a bit contrived to me (and Perry didn’t indicate whether adjustments were made for the strike- and war-shortened seasons of 1918, 1972, 1981, 1994 and 1995), so I reran the correlations using the eras Michael Schell defined in his book Baseball’s All-Time Best Sluggers, which conform better to changes in the game.
Era AVG SLG OBP OPS ISO Notes 1904-1919 Deadball Era 0.879 0.880 0.888 0.922 0.748 1918 excluded 1920-1946 Lively Ball Era 0.841 0.934 0.913 0.957 0.866 1947-1960 Post WWII Era 0.823 0.861 0.864 0.935 0.738 1961-1968 Big Strike Zone Era 0.758 0.845 0.805 0.891 0.719 1969-1992 DH Era 0.817 0.910 0.877 0.951 0.787 1972, 1981 excluded 1993-2005 Lively Player Era 0.826 0.908 0.883 0.954 0.807 1994,1995 excluded
1904-2005 Modern Era 0.774 0.901 0.864 0.943 0.756
As the table shows, in the most recent era (which Schell calls the “Power Era,” but I couldn’t resist using George Will’s characterization) the correlations of slugging percentage and on-base percentage are closer than they had been before, since OBP outpaced SLG in the Deadball and Post-WWII eras. As Perry found, however, overall SLG does retain the top spot across eras.
Secondly, curiously Perry made no mention of On-Base Plus Slugging (OPS), when it might have been appropriate in this summary of his results.
“Despite the ‘OBP is life’ movement spurred along, in part, by Moneyball and the success of the Oakland A’s in recent seasons, hitting for power is more important than getting on base. However, both SLG and OBP are substantially more important than AVG.”
Now, as a fan of Earl Weaver’s third law (“The easiest way around the bases is with one swing of the bat”), I agree with the point. However, Perry leaves out three critical facts that bear on this discussion.
In my opinion, Perry, rather than leave the impression that slugging percentage is of greater importance, should have made the point that both elements are required and that putting runs on the board (as shown by OPS) is really a function of the interaction of SLG and OBP.
With all that said, interestingly, Perry finds that of his 124 playoff teams, 65.3% are better than the league average in ISO, which has a lower correlation than all the rest, while only 61.3% bested the league average in SLG and just 58.9% in OBP. Go figure.
In the chapter “The Glove Man (or, There Are Worse Things Than Making Errors),” Perry includes a nice overview of the role Defensive Efficiency Rating (DER) can play in evaluating the run-prevention capabilities of defenses while acknowledging that pitchers do retain some control over whether a ball put into play becomes a hit.
There were two features though of this chapter that piqued my interest.
First, in the context of informative profiles of John Olerud, Steve Garvey and Graig Nettles and to highlight the inadequacy of traditional fielding measures, he shows the following table related to Olerud.
Period Total Errors by 2B, SS, 3B Before (1994-1996) 192 During (1997-1999) 129 After (2000-2002) 181
This illustrates that with Olerud manning the position, Mets infielders were charged with one-third fewer errors in 68 fewer games due to the strike. Looking at this vast difference brought to mind the fact that most defensive rating systems, with the exception of those of David Gassko, don’t take this factor into consideration.
More interestingly, Perry notes that the 10 teams in his study that ranked highest in park- and league-adjusted DER employed a “hydra-headed center fielder” by deploying two true center fielders in their outfields. The 1999 Mets with Brian McRae, Darryl Hamilton and Roger Cedeno recorded a league-adjusted DER of 104.2 and came out on top. The 1991 Blue Jays with Devon White and Joe Carter at 104.1 came in second. Does this indicate that teams looking to shore up their defense should start at one of the corner outfield positions?
Do We Really Need a Closer?
In the chapter “The Closer (or, Why the Old and the New Both Have It Right…And Wrong),” Perry takes a look at the usage patterns of closers since 1980, as well as the postseason teams where closers fared the best. It probably comes as no shock that Willie Hernandez tops the charts at 76 PRAR for his 1984 performance with the Tigers when he pitched 140.3 innings and 80 appearances racking up 32 saves and 1.75 ERA.
Of course, the reason Hernandez saved as many runs as he did is because he threw twice as many innings as closers of more recent vintage. Even so, Perry is adamant that closers “simply don’t do enough in the way of preventing runs to ever justify MVP honors.”
But what I found most interesting here is that despite closers of the 1980s racking up more innings Perry illustrates, using Baseball Prospectus’ Leverage Score (a number that indicates the difference giving up a run makes towards winning and losing at a specific point in the game as opposed to the start of the game), that managers have actually improved in their ability to leverage their primary reliever (defined as the pitcher with the most saves). This runs contrary to the popular notion that managers of the 1980s employed their best reliever only in tight situations. In actuality, even Goose Gossage had Leverage Scores below the 2000-2004 average of 1.69 in five of his 12 seasons as a reliever, and two of his highest came when he pitched just 79 and 64.2 innings for the Padres in 1985-86.
And this makes sense when you think about it. There are only so many high leverage situations to be had, and if your primary reliever pitches in 80 games instead of 50 coupled with more innings per outing, he’ll likely end up pitching more often when the game isn’t on the line.
The other interesting tidbit that Perry reveals in this chapter is that while a good closer is certainly an advantage, winning teams don’t require one—at least the same one. The Braves of 1991-2003 had 10 different pitchers lead the team in saves. Ten.
Speed, Not Stealing
One of the longest chapters of the book is The Base Stealer (or, Uses and Misuses of Speed) and Perry uses the space to drive home two points. The first has become common knowledge to statheads, and that is that in the long run base stealing simply doesn’t add that much to offensive production and that break-even success rate for stolen bases is around 72%.
He uses an entertaining profile of Rickey Henderson to make this point, by illustrating that approximately 2,900 runs that Henderson created over the course of his Hall of Fame career, just 270 or less than 10% can be attributed to his record 1,406 stolen bases. As far as the teams in his study, their aggregate runs credited to the running game was -174—and these are postseason teams. The non-postseason teams were even worse, largely because they attempted more stolen bases.
However, Perry also makes the case that the stolen base is not the strategic weapon that many make it out to be by noting that the “leveraged value of the stolen base varies negligibly from the random value. It may sound implausible but the numbers bear it out”. Unfortunately, he doesn’t provide those numbers.
In addition, he references a study done by Michael Lichtman that indicates that when stolen base threats reach first, their presence doesn’t seem to help the hitter at the plate. This corresponds nicely with the work of John Walsh here at THT, who concluded in a two–part series earlier this year that stolen base threats very likely add no more than two runs per season.
Finally, Perry acknowledges that speed isn’t all about the stolen base, by referencing Nate Silver’s work that showed that speed is correlated more strongly with long and productive careers. I should also note that speed in terms of baserunning is also starting to be quantified in the work of James Click at BP and yours truly. That research shows that the difference between the best and worst baserunners in a particular season is likely on the order of 10 to 12 runs.
The Art of the Deal
The Deadline Game (or, Why It’s Hard to Win a Pennant in Two Months) was perhaps my favorite chapter, as it looked at in-season trades that approached the trade deadline for the 124 teams in the study.
What interested me, and what I had suspected, was that of the 108 teams that made such deals the average team realized only 2.2% of their total VORP from the trades, largely due to the fact that the season is two thirds over at the trade deadline and the remaining two months offer a sample size that can be heavily influenced by luck.
That said, Perry then goes on to discuss the teams that benefited the most from these deals, with the 1987 Giants and GM Al Rosen far surpassing the rest (15.5% of VORP), by snagging Dave Dravecky, Craig Lefferts, Kevin Mitchell, Rick Reuschel, and Don Robinson. Mitchell, Dravecky, and Lefferts were acquired almost a month before the trade deadline, a lesson that other GMs should note.
He also takes some time to credit Cardinals GM Walt Jocketty for making a series of excellent deals in both 2000 and 2002, and lauds his performance by saying that “whatever your standard for evaluation, Jocketty is peerless among modern GMs in making impact deals for his organization, deadline or otherwise”.
Winning Takes Money
In the penultimate chapter, The Money Player (or, Why This is No Place for the Faint of Wallet), Perry discusses the correlation between making the postseason and payroll among the 124 teams he studied.
What he finds won’t surprise many. Fully 72% of the teams surpassed the average league payroll but he also shows that the percentage decreased in the 2000-2003 period from the 1995-1999 period, a sign he says that the “dark days” of competitive imbalance are behind us.
Recently I ran a similar measure but instead of looking at postseason teams (since they include wild card winners) I looked at division ranking, and found that competitive balance doesn’t appear to be increasing, and in fact actually decreased in the 2002-2005 period covered by the existing collective bargaining agreement.
However, the most fascinating part of the chapter is his explication of how these teams are constructed. For example, he shows that while both high and low payroll playoff teams obtain about 37% of their players from trades, high payroll teams obtain almost 30% from free agency and 33% from their farm systems, while for low payroll teams the percentages are 12% and 52% respectively.
This chapter also includes a very readable history of the dawn of free agency and the sordid history of the A’s and Yankees mentioned previously.
It’s a Winner
In the end I definitely enjoyed the book, both for its application of performance analysis to the strategies and tactics used over the past quarter century, and for brief portraits of some of the more interesting players to have put on a uniform during that time.
Perry’s work is always well-written and when combined with great topics, this one certainly comes out a winner.
References & Resources