Today's Fantasy Rx

Before I begin our off-season pitcher reviews, I am going to spend this week discussing modifications to our LPR pitcher ratings. I will resume player reviews in a week.

Our search for a quick way to evaluate pitchers resulted in our introduction of the LPR pitcher ratings last year. Working from concepts originally developed at Baseball HQ, we devised a simple alphanumeric code that allows easy recognition of specific pitching abilities over a pitcher's recent history. Due to both our confidence in the system and the overwhelmingly positive response we received from readers, I updated our rankings for each league's pitchers on a weekly basis to provide a unique resource to help owners choose their pitchers.

Unfortunately I have grown uncomfortable with some of the inconsistencies in the system, particularly regarding the LIMA and PQS qualifications. Today I will discuss the new methodology we are employing for 2003, and then I will spend the rest of the week listing the pre-2003 pitching codes by presenting one division each day.

The LPR coding we used for 2002 stood for LIMA PQS Rating. Ron Shandler's LIMA qualifications for pitchers include a 2.0+ K:BB, 6.0+ K/9, and no more than a 1.0 HR/9. These benchmarks are easy to identify, however we do not feel they are optimal for identifying the most useful pitchers.

We both are familiar with and believe in most aspects of Voros McCracken's theory that most pitchers exercise little control over the number of hits allowed per batted balls in play. A team's defense accounts for most of a pitcher's hit rate, but we have not seen a satisfactory explanation for why certain pitchers continually wind up on the outlying edges of the study in terms of significant numbers of hits both allowed and prevented. Therefore we believe that identifying pitchers who allow less than 9.0 H/9 provides a welcome additional layer of qualification. Hit rate affects both a pitcher's ERA and obviously his WHIP, so applying a reasonable standard for our LIMA qualification makes sense to us.

Over the past couple of months we have also moved away from analyzing K:BB as a primary skill of pitchers since we believe that the K:BB ratio is similar to WHIP by reflecting multiple skills rather than a single skill. Therefore, since we already have a 6/0 K/9 requirement, we are implementing the equivalent walk rate minimum of 3.0 BB/9 or less, which matches the previous qualification of 2.0+ K:BB.

Finally, you have probably noticed our emphasis on G-F ratio as a central skill for both pitchers and hitters. The logic here makes intuitive sense: anyone who increases the number of flyballs they allow or hit should see their slugging percentage rise due to developing power. While this metric seems to fail for hitters with little natural strength, I believe it reflects an important skill for all players. Therefore we are instituting a 1.0 G-F minimum for pitchers to qualify for our numerical ranking. Since these numbers are published with most box scores, available online at many sites like ESPN, and are printed in publications like Sports Weekly, the increasing availability of the stat warrants its inclusion.

We will use these five skill ratio minimums to qualify pitchers for our numeric LPR rankings: 6.0+ K/9, 3.0- BB/9, 9.0- H/9, 1.0- HR/9, and 1.0+ G-F.

To eliminate inconsistencies in the numeric ranking, we are eliminating any use of half-season stats. Also, we are moving from a ten-digit code to nine-digits so that each pair of letters has a corresponding number. Lastly, we are adopting a 25 IP minimum with the skills listed above instead of the 15 IP we used in 2002. By requiring someone to amass at least half the number of innings towards the 50 IP limit at which pitchers lose their rookie eligibility, we hope to see less movement on the lists during the season.

We will use the following codes throughout the preseason:

1 - K/9 of 6.0+, BB/9 of 3.0-, H/9 of 9.0-, HR/9 of 1.0-,
     and G-F of 1.0+ in 25+ IP in 2002.
2 - K/9 of 6.0+, BB/9 of 3.0-, H/9 of 9.0-, HR/9 of 1.0-,
     and G-F of 1.0+ in 25+ IP in 2001.
3 - K/9 of 6.0+, BB/9 of 3.0-, H/9 of 9.0-, HR/9 of 1.0-,
     and G-F of 1.0+ in 25+ IP in 2000.

Once the season begins and pitchers begin reaching 25 innings, we will shift to 1-2003, 2-2002, and 3-2001. We will not employ any rounding to attain these numbers, and pitchers with a 5.9 K/9, 3.001 BB/9, etc. will not qualify.

Reworking PQS scores is a more intricate task since we love the simplicity of the system. Game scores, while quite useful in their own right, are much more complicated to formulate, and more importantly from our perspective, are not a metric we can easily tabulate on a daily basis. There also is no corresponding analysis of relief pitchers, and we wish to modify the LPR coding so that all skilled pitchers can achieve the maximum ranking of 123abcxyz.

The basics of the PQS system make a lot of sense. A 5 IP minimum to attain any score above zero is necessary since a pitcher cannot win the game without meeting that requirement. We also approve of the points for no more than one hit per inning(H<=IP) and no more than one home run per game(HR<=1).

We are not overly comfortable with the current requirements for either strikeouts (K>=IP-2) or walks(BB<=K/2), however the strikeout requirement is acceptable as long as we strictly enforce IP. Anyone with 3 strikeouts and between 5 and 6 IP, i.e. 5.1 or 5.2 IP, does not earn a point, and neither will anyone with 4 strikeouts and between 6 and 7 IP, etc. This decision leaves lowest acceptable single-game strikeout rate at 5.4 K/9.

As we separated walks from strikeouts for the numeric code requirements, we will make a similar change for the alphabetical requirements. We will award a point for any pitcher that posts a 3.0 or lower in-game walk rate(BB<=IP/3). While this move creates a stricter requirement, we were not comfortable with potential inflated WHIP numbers due to bad walk rates. Now we reward a pitcher for exhibiting good control rather than merely not demonstrating bad control.

The final, and likely most controversial change to the former PQS system is to remove the point awarded for pitching six innings and instead give the point to pitchers whose G-F is at least 1.00. We have seen good arguments for not only keeping the 6 IP point but even adding another point for anyone with 8 IP, a move that logically requires removing the hit rate point since the pitcher exercises the least control over this skill. However, as fantasy team owners, we would rather see a good performance from a pitcher in 5 innings than risk injury or skill degradation by unnecessarily increasing his pitch count. We are not comfortable rewarding a pitcher for IP since so many other factors, including managerial decisions and potential at-bats for pitchers in NL parks, reduce the importance of IP as a measure of stamina. This move from an IP point to a G-F point also reflects the new requirement for the numeric code listed above.

Our new Quality Appearance(QA) requirements therefore award up to 5 total points, one for each statistical goal, to any starter who pitches at least 5 innings and fulfills any of the following skill qualifications: K>=IP-2, BB<=IP/3, H<=IP, HR<=1, and G-F>=1.00. Any starter who does not reach at least one of the required statistical goals or fails to pitch 5 innings earns a 0 QA score.

As the QA name suggests, we are also introducing similar qualifications for relievers to allow every pitcher to reach the highest LPR code of 123ABCYXZ. We will award one point to a reliever who retires at least one batter(.1 IP) for each of the following:

A strikeout rate of at least 9.0 (one per inning). Relievers often enter games with men on base, and a strikeout rate of 9.0 or higher indicates an ability to strand existing baserunners.

A walk rate of 0.0 (no walks allowed). We have repeatedly observed managers pull relievers for walking even a single batter, and since most relievers do not regularly pitch more than a single inning, even one walk pushes their WHIP to 1.00 without even considering hits.

A hit rate of no more than 9.0 (one per inning). Since we are enacting a strict requirement on walks, I am less concerned about allowing hits even though hits increase run scoring more than walks. However a pitcher could demonstrate dominance and control while allowing a bloop single, or the official scorer could fail to award an obvious error for various reasons. This hit rate also parallels our requirement for starters.

A homer rate of 0.0 (no homers allowed). Unlike The Simpsons' episode regarding a similar concept, a reliever cannot allow any homers in order to have a good outing since even a single homer significantly increases his ERA.

A G-F rate of at least 1.00 (one groundball for every flyball). As I discussed above, this intriguing metric rewards pitchers who keep the ball in the infield, decreasing the risk of extra-base hits and improving the chances for double plays, force outs, etc.

Our new Quality Appearance(QA) requirements therefore award up to 5 total points, one for each statistical goal, to any reliever who retires at least one batter and fulfills any of the following skill qualifications: K>=IP, BB=0, H<=IP, HR=0, and G-F>=1.00. Any reliever who does not reach at least one of the required statistical goals or fails to retire a batter earns a 0 QA score.

Two years ago in Wise Guy Baseball, Gene McCaffrey introduced the concepts of DOM(dominant starts) and DIS(disasters). A PQS score of 4 or 5 qualified as DOM while a 0-1 qualifies as a DIS. Rotohelp reader Jon Ratner suggested DUL (dull) as a logical category for any PQS scores of 2 or 3.

In the process of formulating qERA last summer, Shandler presented a fascinating table grouping all 2001 pitcher starts by each of the six possible PQS scores. The aggregate totals for scores of 4 and 5 were logically stunning, suggesting a pitcher with a value no less than $20 in standard leagues. PQS scores of 3 also indicated respectable pitchers likely to earn double-digit value thanks to their overall contributions. However, while scores of 0 and 1 indicated simply terrible performances, the average 2 score's qualitative marks of a 5.78 ERA and 1.67 WHIP do not depict a pitcher I would ever want to roster. While we believe our Quality Appearance method will provide more help for us to identify skilled pitchers, we do not expect a revised study to reflect huge changes in the aggregate averages.

Therefore we are grouping all QA scores of 0, 1, and 2 into the DIS category. Only QA scores of 3 will remain DUL, while scores of 4 and 5 will remain DOM appearances.

Last season we used a 55% DOM minimum for an a, b, or c, and we employed a 20% DIS maximum for an x, y, or z code. We are returning to the more easily computable 50% DOM minimum given our new and stricter requirements, but we are also employing a set of capital letters in the ratings to arrive at the following possible pre-season coding:

Note: Pitchers must pitch in five games before scoring DOM or DIS points.

a - QA score of 4 or 5 in at least 50% of 2002 appearances
A - QA score of 4 or 5 in at least 75% of 2002 appearances
b - QA score of 4 or 5 in at least 50% of 2001 appearances
B - QA score of 4 or 5 in at least 75% of 2001 appearances
c - QA score of 4 or 5 in at least 50% of 2000 appearances
C - QA score of 4 or 5 in at least 75% of 2000 appearances

x - QA score below 3 in no more than 20% of 2002 appearances
X - QA score below 3 in no more than 10% of 2002 appearances
y - QA score below 3 in no more than 20% of 2001 appearances
Y - QA score below 3 in no more than 10% of 2001 appearances
z - QA score below 3 in no more than 20% of 2000 appearances
Z - QA score below 3 in no more than 10% of 2000 appearances

The significant changes described above, especially regarding the move away from traditional LIMA and PQS standards, require that we adopt a new acronym. I hope you do not mind the following brief interlude to explain our choice.

Jean-Baptiste Lamarck developed the theory of inheritance of acquired characteristics in the early 1800s. In short, Lamarck suggested that animals evolve by undergoing physiological changes in response to survival needs and then pass these new attributes to their offspring via normal reproduction. Of course Darwin soundly refuted this theory, but Lamarckian evolution, rather than applying to biological evolution, explains cultural evolution quite soundly as each generation incorporates the new advances and ideas of their predecessors.

Our focus on the underlying skills of pitchers echoes Lamarckian theory as pitchers can develop and refine their skills to improve their skill ratios(strikeout rate, walk rate, hit rate, homer rate, and G-F rate). As either veteran mentors or pitching coaches, they then can pass on the techniques that enabled them to succeed, specifically in regards to relatively new pitches like Bruce Sutter's split-fingered fastball and advances like using the slide step to decrease steals. The methodology of pitching accounts for skills and remains separate from the physical evolution that allows increasing numbers of pitchers to reach 100 mph. While a pitcher's tools develop in accordance with Darwinian theory, a pitcher's skills improve via Lamarckian evolution, so though we will keep the LPR designation when discussing our system, we have selected Lamarckian Pitcher Rating as the new term for these rankings.

I will briefly address a few issues that might arise in defending the validity of this system. First, we consider this system as merely another tool for our readers to use when choosing pitchers. The relatively simplicity of the code allows for easy identification of both sleepers and studs: simply look for a long code with emphasis on recent performance, such as the inclusion of a 1AX.

Second, we are employing minimum IP requirements to eliminate fluke performances. Sustained success over 25 innings necessitates either a good two-month stretch by a reliever or four consecutive starts over which a pitcher displays good overall skills by either avoiding disasters or dominating in his other appearances. I will be happy to take a chance on anyone who fulfills either of these provisions.

Third, we are ignoring minor league performance because the vast amount of AAAA talent indicates that pitchers must impress some front office personnel to receive an extended look in the majors. Perhaps more important for our purposes, practically no one plays minor league fantasy baseball, and therefore some major league track record seems appropriate for us to recognize a pitcher possesses skills that warrant rostering.

Finally, while these changes require more work in obtaining the necessary data, specifically regarding a pitcher's G-F numbers, we believe the inclusion of this metric is important in reducing the risk associated with taking chances on pitchers who often possess limited track records. Also, we consider pitchers with a .60 G-F like Jarrod Washburn far riskier than similarly skilled pitchers with more of a tendency towards groundballs, and therefore do not object to excluding them.

The LPR system identifies pitchers with solid recent skill histories through a relatively basic alphanumeric code. No pitcher can qualify for the numeric code without posting a WHIP below 1.34, and the new skill requirements distinguish pitchers with good dominance who can help fantasy owners accumulate strikeouts as well as hopefully lower ERA totals. While we recognize this system might appear somewhat complicated to new readers, we believe it provides an interesting shorthand to quickly and easily identify pitchers to target.

I will spend the rest of the week presenting one division each day listing all Major League pitchers over the last three years by the team for which they most recently appeared, along with their offseason LPR codes.

Today's Fantasy Rx: While we have spent considerable time this winter debating and analyzing the changes see you listed above, and we are very pleased with the new LPR system, we certainly welcome additional debate on this subject. Please send us your questions or comments if you have any opinions regarding what we have added, subtracted, or even what we did not change in the ratings.