Why Wins is a Bad Statistic, Part 1

Whenever evaluating pitchers, most commentators, fans and managers tend to look at the player's wins and losses to see if they were good or not. Although this trend is beginning to fade (as seen by King Felix's Cy Young in 2010), wins still hold a lot of weight. To be honest, however, they very often are fairly bad indicators of pitcher performance. Take Sid Fernandez and Dwight Gooden for the 1990 New York Mets. El Sid went 9-14; Doctor K went 19-7. Who had the better season? Most people would say Gooden; they would be wrong. Fernandez's ERA was 3.46 (ERA+ of 109), while Dwight's was 3.83 (ERA+ of 98). What caused such a large disparity? Find the answer below the fold:

If we were to neutralize both pitchers' seasons to 716 runs scored environments and equal run support and defense, Fernandez would be 10-9 and Gooden would be 11-13. This makes a lot more sense. So where exactly do all of those extra wins come for Doc, and the losses for Sid? Well, Sid had 0 cheap wins and 6 tough losses; Doc had 5 and 0, respectively. A cheap win is a win not coming in a quality start, while a tough loss is a loss despite a quality start. Sid had 18 QS in  30 GS, and Gooden had 20 QS in 34 GS. If we were to thus calculate each pitchers' win-loss according to quality starts in their decisions, Fernandez would be 15-8, and Gooden would be 14-12; compensating for the bullpens, they would be 16-12 and 15-16, respectively. Again, those numbers make a lot more sense.

Clearly, Sid was quite unlucky in 1990, and Doctor K was exactly the opposite. The difference in wins and losses had very little to do with the pitchers themselves; rather, it was the fielding and run support that made the difference. Looking at each pitchers UERA (unearned run average), Sid's is .503 and Gooden's is about .270. That means that Fernandez gave up about twice as many unearned runs per 9 innings as did Gooden. This is generally a sign of bad defense. Even more shocking, however, is the massive difference in run support. Sid averaged 4.2 runs of runs support per game started; Gooden averaged 6.0. I guess it would make sense that despite a higher ERA, a player with 50% more run support would have a significantly better record, wouldn't it?