Much Ado About Freddy Lew
Author’s Note: I should probably hire someone to write titles for me. Because, boy, that’s a lame title.
I began writing this post with the intention of publishing it before the Giants made a move. Of course, that didn’t work out too well for me, as they’ve since sent him to Toronto for a player to be named later or cash. So much for that. In any case, I figure some can still find this post to be useful in the future, as I’m sure Lewis will spark some debate amongst Blue Jays fans. For me, the Giants trading Lewis is yet another example of how poor their player evaluations are.
I wouldn’t say that there’s been a hot debate over Fred Lewis amongst the lunatic fringe, but a number of debates have certainly taken place. The more statistically-oriented fans point to Lewis’ on-base percentage and positive UZR scores in the outfield as an indicator that he’s a league average Major Leaguer at worst, and should be playing every day, if not in a frequent spot starter. Lewis detractors point to Lewis’ propensity for the strikeout and the fact that he looks downright terrible in the outfield, with his questionable routes and brick hands. For them, these two reasons alone make Lewis an unworthy candidate for an everyday role- not to mention the fact that his limited defensive versatility essentially makes him a one-position bench player that takes a lot of pitches and strikes out a lot. Obviously, I have a bias. I’m of the statistically-oriented mind, and I have a hard time supporting the idea that we should rid ourselves of Lewis. I can’t argue their case- but I can certainly make a case for the statistical side.
Personally, I find the strikeout argument to be amusing. Henry Schulman of the San Francisco Chronicle downplayed Lewis’ on-base abilities due to his strikeout rate a little while back. And, honestly, I can’t blame him for thinking that way. Striking out has traditionally been frowned upon as an evil thing for a hitter to do. In a lot of cases, however, a strikeout is no different than a batted ball out. Certainly, the more balls in play you allow the greater the chance to have a hit. But when all is said and done, an out is an out is an out.
Don’t get me wrong, though- there certainly are instances in which a strikeout is not equal to a batted ball out. There are such things as productive outs- if you strike out, you can’t move a runner over (unless, of course, it’s a dropped third strike, etc.). Anyone versed in simple baseball language knows this, and anyone familiar with the concept of run expectancy knows this to be true as well. If you’re coming up with a lot of men on base, it’s not good to strike out. To play Devil’s Advocate, though, it also prevents you from hitting into the double play.
The beautiful thing about run expectancy is that it allows us to look at the difference an event makes in every single situation (sans the inning and run differential states, of course). We look at the number of runs scored as a result of the event, and find the run expectancy of the situation after the event has occurred compared to the initial state. By taking the average of these events, we find the marginal value of an event (also known as Linear Weights, or LWTS). There are other methods aside from the empirical one, of course- and we can generate marginal values through dynamic run estimators such as David Smyth’s BaseRuns. If we do this for the 2007-2009 National League, for example, these are the values we get:
LWTS = .46*1B + .75*2B + 1.02*3B + 1.39*HR + .49*ROE + .31*NIBB + .33*HBP + .17*IBB – .27*Outs – .29*K
(where “Outs” are AB – H – ROE – K + SH + SF)
Remember how I said that a strikeout is often no different than a regular batting out? Well, here’s some supporting evidence. On average, the run value of an out decreases run expectancy by -0.27. A strikeout is -0.29, a mere difference of 0.02 runs. So, yes, a strikeout is worse than a batting out. But the difference is often marginal.
We can estimate Lewis’ production level by using a weighted average (5/4/3) of his batting rates over the past couple of years. If we give him 600 plate appearances, for example, we get:
That’s a pretty respectable line, I must say. A fair amount of doubles, a number of triples, getting base on an error and a healthy OBP of .353. Not much power, obviously, but it is what it is. If we plug this batting line into our LWTS formula, Lewis comes out as a +7 runs above the average hitter. What if we replace his strikeout totals with a league average rate? Then it changes to +8 runs. That’s right. Changing Lewis’ 141 K’s to a league average of 108 gives him one extra run of value.
What about those productive outs, though? Do Lewis’ strikeout tendencies make him a poor situational hitter?
Perhaps. Lewis does poorly with situational hitting, but I don’t know whether that can be attributed solely to strikeouts. It is important to keep in mind, however, that not all batted ball outs create a “productive” out. A player can be absolutely outstanding at putting the bat on the ball, but with a man on that could very well result in a double play, or, with a runner on second, a weak grounder or fly ball that doesn’t allow them to advance. It might come as a bit of a surprise, but a league average hitter makes a “productive out” about 31% of his opportunities. That’s a rather low conversion rate. Lewis’ weighted productive outs rate is 24.1%, well below the league average by 7.7%. A player sees a maximum of around 90 opportunities to make a productive out in a season. The league average would be successful in 29 of those opportunities; Lewis 22. The difference, 7, is worth 1.7 runs*. That’s it. Given that Lewis hits at the top of the lineup, however, he’ll be seeing far less opportunities- around 30-35. Let’s assign him 35 opportunities just for the heck of it; this makes him a -0.75 productive outs hitter.
Avoiding the Double Play
Lewis’ weighted double play rate is 8.4%; league average is 10.6%. Over the course of 600 PA, Lewis is expected to see around 70 double play opportunities- meaning that he is expected to hit into 5.88 double plays, while the league average hitter is expected to hit into 7.42, a difference of 1.54. Converted to runs, Lewis is a +0.47 in double play opportunities. I’ve written a bit about staying out of the double play before, and based on that “research” and common sense would imagine that it’s the combination of Lewis’ speed and handedness that makes him an asset for keeping out of the double play. I’ll touch on the importance of this a little later in terms of lineup slot implications. For those keeping score, Lewis is a -0.28 runs overall when it comes to situational hitting.
According to Baseball Prospectus’ Equivalent Baserunning Runs, Lewis has a weighted rate of +0.008 runs above average per opportunity. That might not sound like much- but when you consider that he’ll have about 500 opportunities over 600 PA, that comes out to about +4 runs. EQBRR, by the way, not only accounts for stolen base runs, but for the player’s ability to advance on ground balls and fly balls. It might come as a bit of a surprise to see Lewis rated so nicely by EQBRR- this is because we often look to stolen base totals as a sign of speed or baserunning efficiency. This is, of course, only one aspect of it. Fred might not look like the second coming of Carl Crawford, but he’s pretty efficient on the basepaths.
This is undoubtedly the most controversial topic about Lewis’ value- to the statistically savvy, Lewis is a good defensive outfielder. To the non-statistical fan, Lewis is a train wreck waiting to happen. One thing is for sure- he certainly isn’t pretty to watch. I wrote about Mark Teixeira a while back, noting that as human beings we are very much subject to availability and confirmation biases. That said, it’s very possible that our eyes are giving us the wrong impression of him. The Fielding Bible, Vol. II had a nice little summary on his glovework:
Lewis is a gifted athlete who has good speed, a good first step, and a playable arm. But having tools doesn’t make you a good player, and Lewis definitely has room for improvement. This is not to say that he is a liability in the outfield, because he makes more than his fair share of impressive catches. To be a better outfielder, though, he’ll need to cut down on his mistakes. Far too often, he allows balls to bounce off his glove or eat him up after falling safely for a hit. Given his athletic ability, he simply doesn’t come close to catching as many balls as he should. His best skill currently is his ability to create outs with his arm. He doesn’t have the strongest arm in the world, but he plays it up by getting rid of the ball quickly and accurately. He also is prepared to throw on every play, preventing runners from advancing by throwing the ball to the correct base. Lewis has above-average talent for a left fielder. If he puts that talent to better use, he could be a superb defender. (303-4)
In other words, Lewis is a raw defender that botches some easy plays. The BIS scouts gave him a Good Play/Misplay Percentage of .486 (.500 is average, of course) for 2008. Despite his mental lapses and occasional misplay, Lewis grades out by UZR as a +4.8 defender over the last three seasons- and Plus/Minus has him at a +9. How could a player that makes so many mistakes on routine plays be graded as above average?
Quite simply, UZR and Plus/Minus don’t care one iota about how a player looks. It is irrelevant. All the metrics wish to know is how efficient the player was at his position, and compare his efficiency to the performance of other players at the same position. In other words, one of the main reasons why Lewis grades out so well is because his pool of competition includes players like Manny Ramirez and Jason Bay. He looks absolutely fantastic by comparison. This doesn’t mean that he’s a good defensive outfielder, or that he could play a great right field or a passable center field- but compared to other players at his position, he’s more efficient than they are. And from a value perspective, nothing else matters.
Using a 5/4/3 weighting, prorated to 140 games gives us a +2.2 runs saved above the average LF. Not a spectacular defender by any means, but solid in the outfield.
Putting it all Together
I’m not the biggest fan of WAR, but it’s become a common measurement for player value. Using FanGraphs’ replacement level and a standard runs to wins conversion (10 runs = 1 win), we get an estimate of Lewis’ current value.
Ignoring replacement level, Lewis grades as being +6.3 runs above an average Major Leaguer. Put another way: the Giants gave up an above average player for cash or a player to be named later. For a team starving for above average hitters and a nice boost of on-base percentage, they sent out one of their better options for next to nothing. Lovely.
Maximizing Lewis’ Value (a bit o’ game theory)
While WAR is a very good tool for player valuation, it is context-neutral. This works well, but it fails to account for how the player blends in to the lineup and how his team can maximize his productivity. Given what we know from watching Lewis and from analyzing his skill set, we can gather a pretty good profile of what his strengths and weaknesses are:
1. He gets on base.
2. He is very patient.
3. He has some pop in his bat that allows for him to hit for doubles and triples.
4. He has shown an ability to stay out of the double play.
5. He is a very efficient baserunner.
1. He’s not a great contact hitter.
2. He’s too patient and doesn’t put the ball in play as often as he should.
3. He’s not good at driving runners in.
4. He’s not efficient when it comes to productive outs.
5. Did I mention the strikeouts?
It’s relatively easy to see where he’d fit best in a lineup- in a slot that doesn’t see many men on, a place where he can make an impact by getting on base and maximize his baserunning abilities. That automatically rules out the 3, 4, and 5 slots in the lineup. Depending on the lineup, he might fit very well at the bottom of the order- but since we’re talking about the Giants here, it becomes glaringly apparent (if it hasn’t already) that he fits perfectly at the top of the order. Leadoff seems to be a natural fit for Lewis- a leadoff hitter sees men on base in around 36% of his plate appearances, according to the work done in The Book. Compare this to the second hitter (44%), third hitter (48%), and cleanup hitter (51%), and we see that the leadoff hitter sees a lot less men on base than the other hitters in the top of the order and heart of the lineup, by about 10-15%. If Lewis is given 600 PA, we would estimate for him to be at the plate with around 216 men on base (coincidentally, Lewis has seen men on in 37.7% of his career PA). Less men on base means less opportunities to fail at moving the runner over, so that’s good in a sense.
Interestingly enough, if we determine the strikeouts to be too much of an issue for whatever reason, Lewis also makes sense as a #2 hitter. If we look at the run values based on the slot in the lineup, it makes some sense to hit Lewis second as well:
#1 LWTS =.515*1B + .806*2B + 1.121*3B + 1.421*HR + .542*ROE + .385*NIBB + .411*HBP – .328*Out – .329*K
#2 LWTS = .515*1B + .799*2B + 1.100*3B + 1.450*HR + .536*ROE + .366*NIBB + .396*HBP – .324*Out – .322*K
These run values are based on data from 1999-2002, so they’re not ideal for the current run environment. Regardless, it’s interesting to see that a strikeout is better than a batting out while you’re in the #2 slot. And, if you think about it, this makes quite a bit of sense. The #2 hitter has a high amount of runners on first (GIDP opps), so a strikeout would be the out of choice. Coupled with the knowledge that Fred’s pretty good at staying out of the double play to begin with, and you’re looking at a player that would fit pretty well in either the #1 or #2 slot. The strikeouts might make you wince, especially when it comes with a man in scoring position, but we’re effectively limiting those chances to fail and maximizing his skillset.
The Giants never really gave Fred a chance to truly develop into the player he was meant to be: a top of the order hitter that provides good value at the plate and on the basepaths, with a solid above average glove in left field. They placed him in the #3 slot and expected him to develop more power. They happened to be wrong- he didn’t hit for power, and he wasn’t good at driving runners in. Rather than putting him back in a familiar role, they essentially gave up on him. What didn’t help was his reported dislike of hitting leadoff; to which I say, “who cares?” Sit him down and explain to him that it’s for the benefit of the team. Fred’s a kid that’s not afraid to show some humility. If Bochy gave him a thorough talking to, it’s not hard to imagine Fred relenting and accepting his rightful role. The Giants were also apparently concerned with him bringing a bad attitude into the clubhouse- well, what do you expect? If you give a good player inconsistent playing time and sit him in favor of lesser players, that’s going to create some problems.
Finding a place for him on the field was made more difficult with the signings of Mark DeRosa and Aubrey Huff. DeRosa’s currently seated in LF and Huff is entrenched at first base. To me, it seems like the Giants would have benefited from not signing Huff- move DeRosa to third, and shift Sandoval to first. This would seem to maximize the team’s defense as well, as Huff isn’t particularly efficient at the not-so-hot corner. Using a blend of CHONE and ZiPS projections in tandem with a lineup simulator, it looks like having Lewis lead off with Sandoval at cleanup (sans Huff) yields 6 more runs over the course of a full season than a lineup with Rowand leading off and Huff at cleanup (sans Lewis). Granted, these are merely projections- it is possible that Huff returns to his ’08 form and Lewis fades into obscurity. But the very fact that the simulator- which runs 2,000 simulations- finds the Giants offense better with Lewis and without Huff, strongly suggests that we’ve lost a pretty valuable asset to our lineup.
Even if this isn’t true, I find it very difficult to justify keeping Eugenio Velez on the roster and not Fred. I cannot for the life of me understand what it is about him that people like- he has the occasional timely hit and he’s got a lot of speed. Never mind the fact that he’s essentially a pinch-runner when all is said and done; let’s focus on the tools that have never developed and most likely never will. He’s not versatile. His “versatility” consists of putting a glove on his non-throwing hand and having him stand at a position. If that’s versatile, why not hand Bengie Molina an infielder’s mitt and have him take grounders at shortstop? I’m exaggerating, of course, but I want to make a point. Velez isn’t a player that does anything well except for run. He’s not a good hitter and he’s not a good defender. Lewis is clearly the better option of the two, but for some reason Velez remains and Lewis has been shipped off.
Rather than giving Lewis a chance, the Giants decided to let him go. It’s a tragic story, and it’s mind-boggling when you consider that he was one of the best hitters on a team that sorely needs offense. The lack of foresight and analysis by the Giants’ front office is astounding- they’ve given up a productive leadoff hitter for a player to be named later or cash.
Sometimes, it really sucks to be a Giants fan. This is one of those times.
*The run value from 2007-2009 is 0.249 for every productive out. This value is found by taking the average difference between an out that advances a runner and an out that doesn’t move the runner in all three base states.