Making an example of Ted Lilly

This article examines the 2011 prospects of one pitcher, but, more importantly, it is more about the methodology than the results.

For some time, I have been calculating expected ERA in this fashion and though it is hardly the most accurate or scientific method on the market, I find it does me well in fantasy from season-to-season. “xERA” is merely a logical way to see how a pitcher’s ERA should trend in subsequent years. It is less a number to rely on as accurate than a baseline number with which to work and from which to determine direction of numbers from one year to the next.

After a more-or-less middling young career in Canada and New York (with a sprinkle of Oakland), Ted Lilly, like Mark DeRosa, seemingly “came into himself” when the Cubs signed him to a four-year, $40-million contract that seemed quite ludicrous at the time. Lilly proved the skeptics wrong, however, and was worth his cost and more.

Over the past four seasons, few pitchers have been as consistent and reliable as Lilly. Over 2007-2010, he posted the following ERAs: 3.83 (4.31 xFIP), 4.09 (4.14), 3.10 (3.98), 3.62 (4.16). Those are hardly elite, but Lilly has proved himself as a solid, useful, quasi-durable lefty No. 3 starter in both real life and fantasy over the past several seasons. More evidence is by his four-year 3.2 WAR/200 innings production (valued at $12-$15 million per season, depending on whose free agency numbers you use), 1.13 WHIP and 3.68 ERA (tangible results effectuated, not talent).

Over the past three seasons, Lilly has posted an above-average strikeout rate (7.84 K/9, 21.3 percent K rate) coupled with a strong walk rate (2.25 BB/9, 6.1 percent BB rate) that has allowed him to limit the damage from his heavy and ever-increasing flyball tendencies (groundball percentages by season, 2007-2010: 33.7, 33.6, 31.9, 29.5).

Lilly’s biggest talent is impeccable control. He is no Cliff Lee, but his three-year K/BB ratio of 3.48 is top 10 in the major leagues (behind James Shields and Josh Beckett, but ahead of Roy Oswalt and Tim Lincecum). Earlier this year, Tom Tango established that the best measure of control is the differential between a pitcher’s strikeout and walk percentages. Lilly’s three-year differential is 15.2, which ranks 14th overall among qualified pitchers. That places him marginally behind Javier Vazquez, Justin Verlander and CC Sabathia, but ahead of such studs as Jon Lester, Adam Wainwright and Felix Hernandez.

Lilly’s three year FIP of 4.22 is right in line with his xFIP over that period (4.10). Despite a slightly higher tRA (a metric which is traditionally “inflated” in comparison to the average and pervasive ERA/FIP baselines), these numbers accord a good feel around what Lilly’s true talent level is. Splitting the FIP-xFIP difference, we get a 4.15 baseline ERA expectation to begin working with for 2011. From there, we will massage the numbers to reflect park/luck factors and garner a more accurate fantasy expectation.

After that, we will look at Lilly’s three-year strike and strikeout rates and then calculate his expected WHIP using version 1.4.3 of the xWHIP Calculator.

Ultimately, this should leave us with a good feel for Lilly’s expected ERA, WHIP and strikeout totals. Wins are essentially arbitrary, so let’s avoid that analysis entirely—plug and play your own numbers as you see fit. Just note that the Dodgers will be without Manny Ramirez next season (a .400+ OBP with quality power to boot is always hard to replace), though Matt Kemp should rebound some to cut the difference if he cuts back on the strikeouts and gets a better bead on balls hit into the outfield.

And now, let’s do some number crunching.

Given our expected ERA baseline of 4.15, first we need to do is determine how many innings we expect Lilly to pitch. Last season, he threw 193.2 innings over 30 starts—a little more than 6.1 innings per outing. That’s in line with the past four seasons. Using this as the expectation for 2011 seems reasonable, considering the durability he showed down the stretch for Los Angeles despite offseason surgery which kept him out until late April last season. Now, we must further ask how often Lilly will pitch.

Assuming that he is to be the Dodgers’ third starter, behind Clayton Kershaw and Chad Billingsley, and routinely given the four days of rest recommended in the unofficial “how to manage a starting pitcher” handbook which is inevitably given to all rookie managers, presumably including Don Mattingly, this would give Lilly a ceiling of 32 possible starts. Over the past five seasons, he has averaged more than 31 starts per season. Injuries in 2009 and 2010 limited him to 27 and 30 starts, respectively, however, though those injuries seemed to overlap the end of 2009 and beginning of 2010.

Assuming, perhaps foolhardily, that Lilly’s knees are truly healed, let’s set our expectations at 31 starts, with reason to believe he may get a couple more. That would give him about 196 innings. With an expected innings total in place and an ERA baseline, we need a runs allowed total that we can modify and adjust. An ERA of 4.15 over 196 innings would yield a context-neutral runs total of 90.4. Adjusting that to determine his expected ERA for 2011, we need to accord for two factors: park effects and defense.

According to Baseball-Reference.com’s multi-year park factor data, Dodger Stadium depresses offense by about 10 percent. Multiplying Lilly’s context-neutral runs allowed by 0.95 (as the Dodgers only play half of their games at home), we get a new total of 85.9.

Figuring out the defensive adjustment for the Dodgers in 2011 is a less exact process than the park factor adjustment. Mathematically speaking, you take the team’s cumulative UZR and divide it by the team’s total innings played in the field and then multiply by the individual pitcher’s expected innings total. You then subtract this number (which will either be negative or positive, depending on the quality of the team defense) to the pitcher’s adjusted runs allowed. Simple, right?

Not exactly. This formulation of defensive adjustment makes several assumptions which are not true. Dave Cameron recently explained in detail the problems of measuring defense as it applies to a pitcher’s value in a two-part post on Fangraphs. There are two primary problems:

First, not all pitchers will get equal defense. Defense measures a total of groundball, line drive and flyball fielding ability. Not all players have an equal distribution of fielding talent between all types of fielding scenarios. A groundball pitcher will probably not get the same level of defense as a flyball pitcher.

Second, this defensive adjustment approach assumes that a team’s cumulative UZR in a season reflects that team’s true defensive ability. This is incorrect for three reasons: roster turnover, the unpredictable distribution of fielder playing time (especially in the NL, where player substitutions define the predominant late-game strategy), and the unreliability of single-season fielding data.

All three of these problems are notably present with the Dodgers. Ramirez (-5.7 outfield UZR in 2010, -20.9 UZR/150 in 2010, -20.0 UZR/150 career) is out of LA, while Kemp (-24.0 UZR in 2010, -25.16 UZR/150 in 2010, -10.2 UZR/150 career) seemingly had a “down” defensive year after two average defensive seasons in center (-2.9 UZR/150 in 2008 and +3.7 UZR/150 in 2009). With Scott Podsednik also departing the Dodgers’ outfield, who will fill the hole in left field (Carl Crawford or Reed Johnson or someone else?) and how that body will affect the team’s overall defense are glaring question marks. Furthermore, which bench players the Dodgers will re-sign and how they will use them is always an uncertainty which provides little insight to a team’s prospective defense.

As you can see, defensive adjustments are a fickle thing. Unfortunately, some adjustment is necessary if we are to go beyond the mere peripherals. If you use the process that I have just cautioned you about, it is important to view it as an over/under baseline. That is, do we expect the Dodgers’ defense to be better or worse in 2011, compared to 2010, and by how much? Applying the answer to this question, quantifiable with 2011 defensive projections, which will come closer to spring training, will be essential.

For now, suffice it to say, we are going to make the “ignorant” assumption of stability.

In 2010, the Dodgers posted a less-than-inspiring cumulative team UZR of -32.0 over 1,441.2 defensive innings. This gives the team a UZR/per inning of -0.0222, or a UZR of -4.4 over of 196 innings. Note that the Dodgers’ outfield was particularly atrocious, posting a cumulative UZR of -48.5, last season. It’s bad even if you take Ramirez and Kemp’s poor gloves out of that defensive picture: -18.8. That number would still rank as the fifth worst outfield defense in baseball. That’s never a positive sign for a flyball pitcher. Nonetheless, adjust this number for how much you expect the team’s overall defense to improve or deteriorate in 2011.

Applying this defensive adjustment to Lilly’s park factor-adjusted runs allowed, we expect 90.2 runs, almost completely obliterating any pitching advantage of Dodger Stadium. Over 196 innings, this would translate into an expected ERA of 4.14. If the Dodgers add a Carl Crawford type, I am willing to bank on a sub-4 ERA for Ted Lilly, however. Until the dust that is the Dodgers’ outfield situation settles closer to spring training, the defensive adjustment will remain largely a matter of speculation (more so than usual). Perhaps, therefore, it is worth noting that Lilly’s park factor adjusted, peripheral-based ERA (based on 85.9 runs allowed) is 3.94. My feeling pegs his ERA over/under around the 4.00 mark. Feel free to pick a different number.

Ted Lilly’s three-year whiff numbers shows some healthy consistency. His swinging strike percentage in 2008 was 9.7. In 2009, it was 9.5. Last year, depressed by early season “extended spring training” starts in the majors following surgery, it was a still solid 8.9. His K/9 fluctuated a bit over this period (8.09 in 2008, 7.68 in 2009, 7.71 in 2010), but his strikeout rate as a function of total batters faced remained quite constant: 21.4 percent in 2008, 21.4 in 2009, and 21.1 in 2010.

For the sake of healthy pessimism, let’s peg the whiff rate of the soon-to-be 35-year old Lilly’s at a flat 21 percent for 2011. To figure out Lilly’s expected strikeout total, we would need to know his expected total batters faced. Over the past three major league seasons, an average of 4.31 batters have come to the plate per inning. Last season, that figure was 4.28. Using a 4.3 figure makes it 843. Applying the 21 percent strikeout rate, we find that Lilly would whiff an expected 177 batters, which would translate into a K/9 of 8.13. That seems a little high for Lilly, who has historically faced 4.23 batters per inning. So, using Lilly’s historic rate, we get an expected strikeout total of 174, for a K/9 of 7.99. My expectations are slightly tempered: 170 K (7.8 K/9).

Rounding out this analysis, let’s a look at Lilly’s xWHIP per xWHIP v.1.4.3 (version 2.0 is on its way. Since Lilly split time between the Cubs and Dodgers last season, I will need to plug in each set of numbers separately. I am going to ignore defense here; substitute your own expectations of how many hits the Dodgers defense is going to add to (or subtract from) Lilly’s defense-independent expectation in 2011.

Plugging in the relevant numbers, xWHIP pegs Lilly’s defense-independent performance on the Dodgers as being worth a 1.087 xWHIP, while his Cubs performance translates into a slightly worse, but still solid 1.195 xWHIP. The composite result is an xWHIP of 1.15.

Overall, Lilly’s 2011 expected over/under numbers (pre-defense) are 196 innings of 3.94 ERA, 174 strikeouts, 1.15 WHIP baseball. Obviously, the Dodgers’ defense (one of the worst in baseball) is going to have a huge impact on Lilly’s actual numbers, but at least we have a starting point from which to start the draft-day bidding.

As always, post your love/hate in the comments.