Not really something that applies to current events in baseball right now, but I've been asked so many times about this recently, I've decided to create a single post to address the issue.
No statistic is perfect. Especially for pitchers. But, I feel as though the most effective way to measure a pitcher's ability is the record for starting pitchers and holds and saves for relievers. I understand what the majority opinion on this is, but the ERA and WHIP can be just as unreliable for measuring pitchers.
The objective of the game is to win. Teams make the playoffs based on wins, not based on earned runs allowed or baserunners allowed. There's no question that the WHIP statistic is not the best way to measure success; a good pitcher is able to pitch effectively with runners on base. Good pitchers don't give up hits in bunches, they scatter them. So, measuring the timing of hits would be a better indicator, not the number of hits.
Arguing against the earned run average is a little harder. It seems logical that a pitcher should not be held accountable for runs that wouldn't have scored but for a defensive error. It also seems logical to not reward pitchers based on the amount of run support that they recieve. But the ERA is not without its flaws either, and there are equally compelling arguments for wins and losses.
Any runs that score after the inning would've ended are considered unearned. A pitcher should still somehow be held accountable if he implodes on himself and is unable to "stop the bleeding" after an error. Similarly, the timing of when he allows runs should also be considered.
The pitcher's main responsbility is to keep his team in the ballgame. If the pitcher consistantly leaves the game always trailing, even if it's only by a run or two, it says more about his ability to pitch in tight ballgames, whether the score is 10-9 or 2-1.
A pitcher that goes four games with the following numbers would have a respectable 4.15 ERA.
8 IP, ER
5 IP, 5 ER
6 IP, 4 ER
7 IP, 2 ER
But his pitching was severely inconsistent and really only gave his team a chance to win two of the ballgames. A pitcher that consistently goes 7 innings and allows 3 runs however has a 3.86 ERA but has pitched considerably better. Over 200 innings, a short good stretch is enough to really alter the ERA.
Imagine a pitcher that has allowed 90 earned runs in 190 innings. He would have a 4.74 ERA. Say he has one good start and pitches a complete game shutout, adding 9 more innings to his tab. The ERA jumps down to 4.52. Has he really improved his entire season performance that much? He's really only helped his team by one potential game out of 162. (I say potential, as there's no way to really tell if his team would've won had he allowed 5 runs.)
There are two reasons why the W-L record is so essential.
A good pitcher alters his style of pitching, based on situation. Do not punish the pitcher that throws strikes with a 7-run lead and gives up a solo homer here and there; that's what he's supposed to do. Instead, punish the pitcher who risks flooding the bases with runners by nibbling on the corners. (But only if he blows the lead.) For the same reason, relievers and closers should be measured similarly for their ability to hold a lead, not the number of earned runs they allow. A closer with a three-run lead is expected to come out firing strikes. Everyone knows that, yet it's still the right thing to do.
All pitchers pitch in different conditions. On a given day, there are so many variables that have an impact on the game being played on the field. The most obvious choices include the ballpark dimensions, the wind, and the umpires, but other things such as the mood and pace of the game also need to be considered. Don't compare a pitcher pitching in Coors Field with the wind blowing out with a pitcher who's pitching in Atlanta. Instead, compare him with a pitcher that is pitching in similar conditions -- the other pitcher in the game.
W-L is resistant to outliers. A few bad starts here and there can really hurt a pitcher's ERA. If you're down by six runs, does giving up another three-run homer really hurt your team? How about giving up single runs in three different tie games? Both effect the ERA in the same manner, but one hurts the team a lot more than the other. No matter how poorly or how well a pitcher pitches, it only (directly) effects the outcome of one game. So let's consider a statistic that measures a pitcher's ability to pitch well in more games, not a statistic that measures a pitcher's ability to dominate in fewer games.
There are flaws with the W-L record as well. The main is that people will still claim that those pitching for teams with a good offense will have more favorable numbers. While that's true, shouldn't a pitcher pitching for a team that can hit be more aggressive within the strike zone? Isn't that smart pitching?
Another argument I've heard is the bullpen. Why penalize the pitcher with a no decision when his bullpen fails to hold a lead that he worked so hard to create? Well, perhaps that is a flaw that could be corrected; maybe there should be a stat like "number of games left with lead".
But since all pitching stats are flawed in one form or another, why not put more emphasis on the one stat that matters at the end of the day? Winning ballgames...