What does it mean, then, for a statistic to measure winning? ERA, for example, measures runs - preventing runs is obviously half of the way you win ballgames. Some statistics, like a pitcher's won-loss record, try to move past runs and try to measure wins directly. Runs are only important in how they affect wins, but, certainly, having an estimate of the number of wins a player is worth is even more valuable. We can take that one step further - the object of a team is not to win games, but win championships. In some sense, adding 10 wins to a 60 win team isn't valuable - it is indicative of a quality player, but not quality in and of itself, since the team doesn't have a chance at winning the World Series.

But we can't just look a the number of pennant or World Series wins for a player, since it's not a player's fault if they are surrounded by teammates who are not championship quality. Ernie Banks never made a World Series, but that doesn't mean he wasn't a Hall of Fame player. So how can we reconcile this? Here's what I propose - we'll take a look at pitchers, but a similar method would work for hitters. Take an average team - average pitching, average hitting, average fielding - but with the player you want to study on the pitching staff. This would be easier with an example - let's look at Pedro Martinez, 2000. First, we need to know how many runs the team would score and allow. (For the record, I'm only looking at earned runs, so the runs scored for this average team will be less than the average team scored in the AL in 2000). The ERA in the American League this season was 4.92 - multplying that by 162 games gives us 797 runs scored. This mythical team will allow fewer runs with Pedro on the staff - as opposed to an average pitcher (with a 4.92 ERA) pitching Pedro's 217 innings, we'll use his 1.74 ERA, which would save 96 runs. Therefore this team would allow 701 runs.

What would the winning percentage of such a team be? If you're familiar with the Pythagorean formula, you'll know that obtaining a winning percentage is a simple formula involving runs scored and allowed (if you haven't come across it before, just trust me for now - it will be the subject of another column). In any case, this team would have a winning percentage of .558, which translates to about 90 wins a season (this tells you something about the quality of the Red Sox aside from Pedro, since they didn't win 90 games). Now comes the crucial step - 90 games is not enough to win the pennant (well, it is now with the Wild Card, but for historical purposes, we're going to assume a two team playoff system). I am boldly going to assert that a team will win the pennant if they win at least 95 games (this may or may not be true, but too bad). So what is the chance that a .558 could win 95 games (or more)? You can figure it out using probability, but I'm stupid, so I'll just plug it into the computer, which spits out the answer: 21%. For reference, what is the chance that an average team without Pedro could win at least 95 games? The answer is about 1%. So Pedro, this season, was worth .20 pennants, which is pretty amazing if you think about it.

Okay, we have a new stat, but why is it valuable? I think another example would help - here are some career numbers for two Hall of Fame pitchers, Robin Roberts and Fergie Jenkins:

Obviously their numbers are pretty similar. But can we look strictly at career totals? Anecdotally, and looking at seasonal statistics, Roberts clearly was a better pitcher at his peak - he led the league in wins for four straight years, and in innings pitched for five straight years. Jenkins, on the other hand, wasn't dominant to the same degree, although he made up for it in consistency. The question is: which pitcher is more valuable? This question doesn't limit itself to these two players - given two players with identical career numbers, which player is more valuable, the one with the higher peak, or the one with consistency?

The answer in my mind is the one with the high peak - this player will get you into the playoffs more often. To take an extreme and invented example, imagine Ronald McDonald goes 20-0 and 0-20 in two seasons, while Mayor McCheese goes 10-10 and 10-10. If they both play on average teams, Ronald McDonald has a good chance of making the playoffs in that first season, while Mayor McCheese probably won't make it. The fact that Ronald's second season is so bad isn't too important, since neither a 0-20 or 10-10 would be enough to get an average team into the playoffs. Ronald McDonald, the player with the higher peak, is more valuable.

This article is getting long, so I'll get to the point - Robin Roberts was far more valuable than Fergie Jenkins. Although they ended their careers with nearly identical numbers, Roberts was worth .52 pennants in his career, while Jenkins was only worth .31 pennants. Roberts' peak, from 1950-1954, accounted for .33 pennants, more than Jenkins accumulated over his entire career. (By the way, I adjusted ERAs in this final analysis by park, which helped Jenkins a little bit, since he played in hitters parks over his career.) If you're curious, Walter Johnson is the all time leader in this statistic, at 2.09 pennants. That's all we have time for today, boys and girls, but they'll be more on this topic next week. Bye bye! <!-- Article Ends Here --!>