skyking162

Tristan Cockraft of ESPN.com has another one of those articles up about how Wins are a dumb pitching stat (they are) and how his new Quality Start off-shoot is the solution to that problem (it’s not). Quality Starts (at least 6 IP and at most 3 ER) are definitely better than Wins because they ignore run support, something outside a pitcher’s control. Cockcroft thinks his Winnable Starts (at least 7 IP and at most 3 ER) are better than Quality Starts because… well, he doesn’t provide much proof. But even if they are better, it’s a small improvement and completely misses the main shortcoming with Wins, Quality Starts, and Winnable Starts:

They are all black-and-white stats — a pitcher either earns one or doesn’t.

I think we all realize that there are many shades of gray between a perfect game and a typical Livan Hernandez start. So why pick one point in between and draw the line there? Why not instead recognize that there’s a sliding scale to performance?

I’ve advocated one option in the past, Game Score. And you cold also go the Support-Neutral Wins route. But both of those are all-encompassing — they’re too good. In fantasy, most leagues want a mix of different stats that value a pitcher in combination, not one stat that does it all.

So how about something a little simpler? For this simple new stat, let’s keep the general idea behind the Quality Start: more innings and fewer runs are both good things. We just need to find the balance between quantity and quality of innings.

A starter only has value if he’s better than a replacement pitcher, which means better than a 5.75 ERA (I’m going to round to 6 for the sake of simplicity.) He saves his team runs by allowing fewer runs than a 6.00 ERA pitcher. And the more innings he pitches, the more he helps, because the team can only be expected to provide a replacement-level pitcher for those additional innings. Here’s the formula for runs prevented:

(6.00 - ERA)*IP/9

We can get rid of the divide-by-nine thing because it just re-scales the first part. And we can replace the pitcher’s ERA with it’s definition:

(6 - ER/IP*9)*IP

By distributing the IP and factoring out a 3 (again, it just changes the scale), we get:

2*IP - 3*ER

That’s simple and intuitive: 3 points for pitching an inning and -2 points for giving up a run. It’s similar to ERA, but favors starters instead of relievers because starters get way more innings. A 180 IP starter with a 3.50 ERA is three times as valuable as 60 IP reliever with the same ERA. In fact, to match a 3.50 ERA reliever, a starter only needs a 5.15 ERA to be as valuable. I think that’s a good balance between just using ERA, which promotes using only relievers, or just using IP, which promotes picking up any hack off the waiver wire. All those league-average 4.50 starters will have value, which they should.

So do you think “two points per inning, minus three poing for each run” has any chance of catching on? Would a catchy acronym help it or hurt? Either way, it’s sure better than Winnable Starts (and Wins).

Popularity: 6% [?]

This entry was posted on Friday, July 25th, 2008 at 11:47 am and is filed under Baseball. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.