Hit Rate Observer

All this talk of the swine flu got me thinking: Every successful strategy is a mutant. The strategy invades a pool of creatures who are blind to its advantages, and it exploits some central feature of the creatures’ day-to-day operation. This continues until the creatures recognize that the strategy is a winner and unleash defenses (usually in the form of mimicry), at which point the strategy has no edge to exploit and becomes harmless.

Suppose that we wanted to introduced a mutant to fantasy baseball. To be most effective, we want to attack the bias that has the widest spread—something that every league, and nearly every player, is vulnerable to. Does such an opportunity exist?

Yes. The most dominant strategy in fantasy today is the LIMA Plan. Devised by The Master, Ron Shandler, the LIMA Plan provides guidance on how to build one’s roster. To be precise, “LIMA” is both a plan (on how to allocate dollars) and a filter (on how to identify pitchers who are worth a bid—i.e., who are “LIMA-caliber”). In this piece, we’ll be focusing on the filter, which is generally expressed as 6.0 K/9, 2.0 K/BB, and 1.0 HR/9.

To a good approximation, the LIMA Plan has universal exposure. Every fantasy player worth his or her salt has antibodies against LIMA—you can’t name a pitcher with good base rates and expect him to slide past your leaguemates. This reflex among fantasy GM’s is nearly instinctive—

—Which is good; laziness is what we wish to attack. Rule LIMA, and you’ll rule the world.

But is there a mutant that can beat LIMA? That’s unclear. If there is, though, the secret may be staring us in the face:


Technically, “LIMA” stands for “Low-Investment Mound Aces.” However, the name was chosen in a contest because it evoked Jose Lima, the multiple Cy Young winner and first-ballot Hall-of-Famer whose finesse in the diplomatic arena brought lasting peace to—

Oh, you mean the Jose Lima from this reality. The one with a career 5.26 ERA in 1,567 innings. In his 11 seasons of consequential work, Lima had an ERA under 4.00 only twice. Yes, he posted a 3.70 ERA in 33 starts for Houston in 1998, and then a 3.58 ERA in 35 starts the following year; excellent seasons, no doubt. Remove those years, though, and Lima has an ERA of 5.98 and a record of 52-84. It’s true that Lima’s skills went downhill after 1999, but even in his three seasons before 1998, he had ERAs of 5.28, 5.70, and 6.11.

In other words, Jose Lima is not a pitcher whom you wanted to own more than twice in 13 years. And yet, our premier pitching plan is named after him. What a country!

This mismatch elicits a suspicion: How did Jose Lima come to be enshrined? Perhaps there is a flaw here.

For his career, Jose Lima had base rates of 5.6 K/9, 2.5 K/BB, and 1.5 HR/9. From 1994-1999 (when he acquired his allure), his rates were 6.6 K/9, 3.9 K/BB, and 1.2 HR/9.

Now, technically, Lima’s HR rate keeps him from passing his own filter. In fact, in his pre-collapse period, he achieved a homer rate below 1.1 HR/9 only twice—he had a 1.08 HR/9 in 1997, and a 1.096 HR/9 in 1999. So that we may include the man himself, we’ll expand the definition of LIMA here to parameters of 6.0 K/9, 2.0 K/BB, and 1.1 HR/9.

We looked at all pitchers since the year 2000 who have had at least one LIMA-caliber season of at least 100 IP. There were 131 such pitchers. We then asked, How many LIMA-caliber seasons did these pitchers have for their careers (possibly extending before the year 2000)?

Here are the results, grouped by the number of LIMA-caliber seasons:


Fifty of our 131 pitchers (38%) exhibited LIMA-caliber skills for one season only. (Scanning just the As, the list comprises Terry Adams, Jeremy Affeldt, Wilson Alvarez, Rick Ankiel, and Bronson Arroyo. Dare we tally the draft dollars spent chasing this fivesome?)

A Hardball Times Update
Goodbye for now.

Only 40% of our pitchers have passed the LIMA filter for three or more seasons. That’s a mere 52 pitchers, spread across nine seasons! Granted, some of the pitchers are still pitching today, and some of those can be expected to lift their LIMA counts. Still, it’s fair to say that fantasy fortunes rose and fell on whether you got in early on these 52 names. (Only one pitcher has a 100% rate of LIMA seasons in a career of four or more seasons. Hint: He pitches for Houston. And he has some work to do to run the streak to nine seasons.)

How about this: After these 131 pitchers put up their first LIMA-caliber season, what fraction of subsequent seasons were LIMA-caliber (and in at least 100 IP)?

A: So far, these pitchers have produced 481 subsequent seasons (of any length). Of those seasons, only 35% were LIMA-caliber in 100 IP. Nineteen pitchers were responsible for half of those seasons; collectively, they had a repeat rate of 73%. If you banked on one of the other 112 pitchers, though, your hit rate was just 23%.

If this be skill, I want nothing of it. A replay of LIMA skills is 1-in-3 overall, and 1-in-4 for the unestablished? Why bother? Admittedly, we’re using LIMA eligibility as a proxy for good results. But a homer rate over 1.1 HR/9, or a strikeout-to-walk rate below 2.0 K/BB, virtually ensures a troubled season.

These findings suggest that the LIMA Plan is vulnerable to a mutant. Far from possessing repeatable skills, most of these pitchers were merely in the right place at the right time—much like Jose Lima, when the LIMA Plan was looking for a name. That may be the real lesson of the LIMA Plan.

Still, the longevity of the LIMA Plan should count for something. And we haven’t done the hard work of showing that a superior strain exists. We have a notion that an answer may lie with the low-K, high-command pitchers whom we identified last week. But we welcome your suggestions.

Newest Most Voted
Inline Feedbacks
View all comments
Eric Hinz
14 years ago

Excellent work.  The issue with LIMA is identifying the LIMA qualifier before he qualifies and then selling him after.

Seems a sarcastic “Good luck with that” is appropriate.

Tube Sox
14 years ago

Who’s the outlier at 20 seasons?

John Burnson
14 years ago

Roger Clemens. Of his 22 seasons of at least 100 IP, only two weren’t LIMA-caliber (1999 & 2000).

14 years ago

John, why was 100 innings used as the filter?  Is there a reason for that, or was it simply arbitrary?

John Burnson
14 years ago

I used 100 IP because I wanted to work with starter-types but I didn’t want to have to decide among 120 IP or 180 IP or whatever; logging 100 IP in a season is generally a mark of at least occasionally starting games. I also wanted to err on the side of generosity (i.e., allowing pitchers—especially new pitchers to the majors—to post LIMA-caliber seasons).

I’m open to using a higher (or even lower) threshold if you think that would be more pertinent or revealing.

14 years ago

I think the number of innings pitched is important in defining what, exactly, one wants LIMA to accomplish.  The original concept was to figure a way to minimize the cost of a pitching staff in order to focus the remaining dollars on hitting—a much more “stable” investment arena.  The plan was not created to enable someone to put together a top pitching staff.  A fundamental component of the plan was to hit the league minimum for innings pitched and no more—the thought being that the ratio stats (ERA and WHIP) could be corrected more easily during the season if you snagged a $1 bust.  Thus, relievers who meet the criteria are important to the process.

John Burnson
14 years ago

I take your point. Part of the disconnect is that LIMA was created for—or at least *in*—a 4×4 world; you could plausibly pull off a two-SP staff. But the reliever-rich roster has a harder climb these days; I suspect that leagues that don’t count strikeouts (or innings, or something else that pitchers are rewarded for having a lot of) are rarities.

So lowering the IP threshold may be truer to “classic” LIMA, but I have my doubts about its revelance today. Still, I’ll consider re-running my analysis at 75 IP.

14 years ago

75 is probably too high, too.  When I think Lima, I see it as a way to identify a future closer, not sp’s.  Draft skills not roles, right?  Try the study at 40 or 50 ip.

14 years ago

From BaseballHQ:
“The LIMA filters were never intended to predict future instances of attaining those filters. LIMA was intended to predict short-term improvement in ERA… [in the case of Lima for instance] the only thing that mattered was Lima’s near-LIMA-worthiness in 1997 which presaged a sharp ERA improvement in 1998. That’s it.”