Not surprisingly, an analysis of Bert's prime years of '70 to '79 demonstrates that despite his superlative ERAs he didn't significantly improve his team when he was on the mound. Yes, Bert didn't get good run support from his teams, who scored .35 runs/game fewer for Bert than they did for other starting pitchers. It is also true that in measuring Bert's performance against his teams' Bert was competing against some pretty good pitchers. For the entire decade, Bert pitched on staffs that were slightly above average even without Bert's contribution, and the staffs on his '70, '72, '77 and '79 teams were among the very best in their leagues. But the Team Relative analysis controls for these factors, of course.

Even after increasing Bert's run support to team average, and adjusting his team's W-L record downward to reflect what it would have been with an average pitching staff, Bert still only outperformed his team's W-L record by 10.2%. That's down in Drysdale territory. As I've previously noted, Bert hugely underperformed his Pythagorean projection during those ten years, compiling a .536 winning percentage as compared to a .599 PythPro. If Bert had been able to perform to his PythPro he wouldn't be such a hot topic today because he would have been inducted into the Hall years ago.

Bert's Team Relative performance was worst during his first six years with the Twins, the period that prompted the Sports Illustrated article wondering why Bert wasn't a bigger winner. Bert's outperformed his team by 8.3% during those years. He improved slightly in his stints with the Rangers and Pirates from '76 to '79, outperforming his team by 13.9%, still well short of what we'd expect from a top flight pitcher. Bert's worst year in this regard was '72, when he performed only 2.2% better than his team despite receiving .44 runs/game more than the other Twins starting pitchers. This is one year where Bert can truly be called a victim of a poor distribution of run support, with a disproportionate number of games falling at either end of the spectrum - a large number of games in which he received three runs or less and a large number of games where he received seven or more. Amazingly, in Bert's 38 starts there were only five games in which he received 4, 5 or 6 runs of support. Bert's run support distribution was also very poor in '75 and, to a lesser extent in '73 and '74.

Bert's run support distribution was more conventional in '76 to '79, although his average run support in '76 was terrible - only 2.75 runs/game. But remember, the Team Relative analysis controls for poor average run support*; it doesn't control for poor run support distribution. In '77, '78 and '79, Bert's run support was almost precisely team average. After controlling for Bert's run support for the '76 to '79 period, a period during which his run support distribution was more conventional and his ERA+ was a very good 125.5, he still only managed to outperform his team by 13.9%. And since we've acknowledged Bert's poor run distribution in the period '72 to '75, we should also acknowledge that Bert's Team Relative performance in the period '76 to '79 was skewed by his W-L record in the '79 season, when the Pirates' bullpen and bats bailed out Bert an extraordinary 13 times after Bert left the game in a position to lose. To give you some idea of how extraordinary this "bailout" total is, consider that Bert was similarly bailed out only 18 times in the preceding nine years. Bert's 12-5 record in '79 is extremely misleading, and if not for his good fortune and the Bucs late-inning dramatics for Bert in '79 his record would have been something like 12-15, which more than eliminates the improvement we see in Bert's Team Relative performance in the late '70s.

I've not modeled Bert's projected record from '72 to '75 assuming a more optimal distribution of run support. It could be done using a system that generates a random distribution of run support and Bert's projected W-L record using such a system would no doubt benefit. There's also no doubt, however, that any benefit to Bert from a more conventional distribution of run support from '72 to '75 would be largely offset by his '79 season, when Bert easily could have lost an additional 10 games and his 12-5 and .706 winning percentage was not reflective of his  performance: he had a 109 ERA+, a LevERA+ of 99, and won only 12 of his 37 starts for a World Series championship team.

Again, this Team Relative analysis is limited to Bert's peak decade of '70 to '79. As the Bert Backers would no doubt argue, Bert had some great seasons outside this period, primarily '84 and '89. But the '70 to '79 period forms the overwhelming bulk of Bert's argument for the Hall. For the balance of his career he had a .533 winning percentage and 108 ERA+, and despite the excellent '84 and '89 seasons the 80's were an exceedingly erratic period for Bert, a period in which poor seasons ('80 and '88) and injury limited seasons ('82 and '83) detract from his case for the Hall. The argument of the Bert Backers is almost exclusively based on the '70s, during which he posted his best ERA+ figures, almost 2/3s of his shutouts and six of his eight 200 strikeout seasons. But the fact is that for all of Bert's statistical achievements in the '70s, they didn't translate into a commensurate win total and W-L record and Bert didn't improve his team as much as he should have. And the problem wasn't run support.

__________________________

* For example, Bert posted a 13-16 record in '76, worse than his team's 79-83 record. But the Team Relative analysis has Bert outperforming his team by 32% after adjusting for his run support.