Feel Lucky, Punk?

Dave Paisley

One of the more curious aspects of baseball is the attribution of wins to the pitcher. Perhaps it's inevitable given the American predilection for having to credit someone, anyone, with credit for winning game. In baseball it's the pitcher, in football, the quarterback, in hockey the goaltender.

To a greater extent than the other examples, in baseball it's possible to argue that the player on a team most directly in control of the outcome of a game is the pitcher. Every play begins with a pitcher in control of the ball, and even though the catcher may call the game, the pitcher still has to execute.

But even though baseball has always felt a need to attribute wins to the pitcher, it doesn't mean that the attribution is correct, or that it's a good way to evaluate pitching talent. After all, there is another equally large component in determining whether a team wins a game, and that's runs scored. For a pitcher, it's better to be on a good team, of course, where the offense is likely to be better. However, there is always a wide variation in run support among pitchers on the same team.

While some fans argue that it is somehow a pitcher's fault if his team doesn't score runs regularly for him, it seems more reasonable to attribute this non-uniform run distribution to blind luck, unless we are to believe in mystical powers -- or God forbid, chemistry.

So in evaluating a pitcher, it seems reasonable to take into account more his runs allowed than his Win-Loss record. However, being the creatures of habit we are, it is kind of interesting to take a look at some of this run scoring variation and imagine what pitching records would be with a more even distribution of runs. We can do this by taking a look at the qualifying pitchers in a league to determine what their average run support is, then adjust their win-loss records to account for the change from actual to average.

In most cases we actually find that a pitcher with a stellar record can thank his offense to a much larger degree than he can thank his own pitching ability. The rare exceptions are the Randy Johnsons of the world, who will win at a phenomenal rate with all but the lousiest run support. Remember last year when he went winless in about five games, being on the losing end of a no-hitter and a one-hitter? Even so, he still managed to eke out a 17-9 record and the Cy Young award.

The best way to see some of this variation is in a table form (unbelievable, I know, but there it is) and the following is a summary for most pitchers of interest in the National league for the first half of this season. All of these guys pitched enough to qualify for postseason awards.

First, an explanatory note about the numbers. The first number column is OPS. This is the on-base percentage plus slugging allowed by the pitcher. As a guide, a number below .600 is outstanding, below .700 is very good, above .800 is getting bad and above .900 is atrocious. The next column, RS, is run support per nine innings while the pitcher is in the game. RA is his runs allowed average, similar to ERA, but including unearned runs.

W and L are actual wins and losses, while WPn and LPn are adjusted wins and losses based on neutral run support, as calculated by the Pythagorean method* for the same number of decisions. Finally, W-Wpn is the difference between actual wins and adjusted wins. A positive number indicates additional wins the pitcher received due to luck and high run support, while a negative number indicates bad luck and lousy run support. The table is arranged from worst luck/support to best.

Pitcher	Team	OPS	RS	RA	W	L	WPn	LPn	W-WPn
Holt	Hou	.835	5.07	5.23	3	10	7	6	-4
Lima	Hou	.976	4.15	7.87	2	13	5	10	-3
Parris	Cin	.855	4.35	5.80	5	11	8	8	-3
Yoshii	Col	.864	6.18	5.64	4	9	7	6	-3
Benson	Pit	.665	4.16	3.54	8	6	10	4	-2
Rusch	NYN	.717	3.72	4.43	6	7	8	5	-2
Dotel	Hou	.872	5.99	5.99	1	5	3	3	-2
Schilling	Phi	.780	4.47	4.26	4	5	6	3	-2
Daal	Ari	.902	4.32	8.44	2	10	4	8	-2
Tapani	ChN	.777	4.31	5.03	5	7	7	5	-2
Millwood	Atl	.795	5.65	5.57	5	8	7	6	-2
Dreifort	LA	.824	6.70	6.24	4	7	5	6	-1
Wolf	Phi	.707	4.54	3.91	8	5	9	4	-1
Glavine	Atl	.678	4.98	3.75	9	5	10	4	-1
Rueter	SF	.777	5.19	4.15	6	4	7	3	-1
Hampton	NYN	.698	6.04	3.95	9	5	9	5	0
Pavano	Mon	.711	6.31	3.71	8	4	8	4	0
Ortiz	SF	.939	5.97	7.48	4	8	4	8	0
Dempster	Fla	.719	4.82	4.01	9	5	9	5	0
Brown	LA	.581	6.05	2.59	8	2	8	2	0
Park	LA	.719	5.58	4.46	9	6	9	6	0
Perez	LA	.856	6.24	5.33	4	4	4	4	0
Ankiel	StL	.728	7.69	4.53	6	4	6	4	0
Johnson	Ari	.572	5.16	2.12	14	2	14	2	0
Hentgen	StL	.791	6.79	5.42	8	6	7	7	1
Maddux	Atl	.652	4.55	3.73	10	3	9	4	1
Neagle	Cin**	.727	6.50	3.67	8	2	7	3	1
Estes	SF	.733	10.23	4.26	8	3	7	4	1
Mulholland	Atl	.854	5.79	6.03	9	8	8	9	1
Stephenson	StL	.814	7.65	4.96	9	5	8	6	1
Leiter	NYN	.695	5.46	3.37	10	2	9	3	1
Anderson	Ari	.804	5.54	4.54	8	3	7	4	1
Kile	StL	.767	5.40	4.81	11	5	9	7	2
Benes	StL	.801	7.54	4.64	9	3	7	5	2
Average			5.66	4.91

*Pythagorean method: win percentage = RS^2 / (RS^2 + RA^2)
**Pre-trade numbers only.

Take a look at Chris Holt, for example. With an OPS allowed of .835, he doesn't deserve much in the way of a winning record. Still, if he had received the league-average 5.66 runs per nine innings support (see the average at the bottom of the table), his 5.23 RA should have been good for a 7-6 record. Even his actual run support of 5.07 would normally be good for 6-7, but his poor luck has saddled him with a 3-10 record.

Conversely, at the bottom of the table, Andy Benes hasn't pitched much better (.801 OPS against, 4.64 RA, and not in Enron), yet his 7.54 run support made him a 9-3 pitcher in the first half. Give him league-average support, and he'd be 7-5.

Normally you'd think Andy Benes is doing great at 9-3, and Chris Holt is awful at 3-10. And yet, take out luck and team offense and you have two pitchers that would be difficult to tell apart.

I'll leave you to peruse the list yourself for other anomalies, but here are a few more examples to chew on.

Randy Johnson is having exactly the kind of success you'd expect. His actual run support is a little below average, but not by enough to make a real difference. With an ERA of 2.12, any average run support over 4.00 is enough to guarantee he'll win all but a handful of games in a season.

Darryl Kile: for some reason, he is making up for the atrocious time he had in Colorado with some good luck this year. As an average pitcher with average run support, he should be around 9-7, rather than the 11-5 record he put up in the first half. Contrast that with Rick Ankiel, who has a better RA and much better run support, but has a pedestrian 6-4 record to show for it.

Kris Benson has a peachy ERA, but lousy run support. Even with his piddly actual run support he had an 8-6 first half, but with average offensive support, he would be looking at more like 10-4.

There are a few conclusions that can be drawn definitively from looking at these numbers, and the first one is that you can't judge a book by its cover, or in this case, W-L record. Second, apart from a few radical exceptions (Johnson, Kevin Brown), there is very little to separate the vast majority of major league pitchers.

about the author

Dave Paisley has not yet taken us up on our dare to tell Andy Benes he's just lucky. Offer to back Dave up by holding Alan back at drdjp@strikethree.com.