The #57 Rule:

Johan Santana has pitched well this season for the Mets, but due to lack of run support and a poor bullpen, he has totaled just 7 wins and 7 losses, whilst having a whole lot of no descisions...

There should be a statistic created called probable wins... where it takes how many runs you let up in that game, and takes the league average for runs scored in a game, and decides whether you pitched well enough to win that game...

for example if a pitcher was to 7 innings and give up 2 runs, and the league average for runs scored is 4.5 than that pitcher should recieve a probable win...

probable wins requirements should be, the pitcher went at least 6 innings, and gave up a full run less than the league average of runs scored...

lets see how many probable wins Johan Santana has this year:

the league average for runs scored this year by an offense is 4.5 per game...

so Johan should have 13 probable wins this year, out of his 17 games started... that is just 4 probable losses....

so his record hypothetically can be somewhere around 13-4... not bad eh?

Well rad, there sort of is a stat like that already:  quality starts.  You have a good thought there about using the league average for runs scored as a barometer, but quality starts does a decent job already.  Along the same lines you discuss above, Santana has 13 quality starts in his 17 games.

The basic idea is that, if you go 6 or more innings and allow 3 or fewer runs, you should win most of the time.  It's a very telling stat in general to say if you run the guy out there, what are odds that a team will win with a normal offensive output.

I love stats, but IDK about this one. 

I still don't understand the whole HOLD stat.  Other than for agents to get a few more dollars in a salary negiotiation I just don't see the point. 

PIcture it: 

Joe Smith comes in to the game during the 6th inning and gets one batter out and then gets pinch hit for.  He gets creditted for a hold when we win the game.

Give me a break! 

I know we will never have the old days back, but look at Roger McDowell in 1986: 14 wins 9 losses and 22 saves.  He was involved in a real decision with over 25% of the games.  In the words of Archie & Edith: THOSE WERE THE DAYS!

Instead, I would suggest maybe a "quality start winning percentage" or something like that.  Basically it would just be:  how many of your quality starts has your team won?  It's not as easy as saying that Santana has 13 quality starts (one off the major league lead by the way, held by Haren and LIncecum) and he has 7 wins.  The Mets might've won a game for him in a non-quality start, or he might've lost a quality start.  So you'd have to look game-by game.

It's amazing league-wide how many quality starts are wins for certain pitchers.  Again you can't take these completely at face value because of the problem I noted, but it's interesting:

Haren:  14/17 QS, 8-4

Lincecum:  14/16 QS, 9-1

Guthrie:  13/18 QS, 4-7

Cook:  11/17 QS, 10-5

Harang:  11/18 QS, 3-10

Lee:  11/15 QS, 11-1

That's a sampling.  It really goes to show who doesn't get the run support (Lincecum, Guthrie), who gets "lucky" and wins most of the time when they have a good outing (Cook, Lee) and who has clearly pitched a LOT better than his record shows (Harang).

I'll check about Santana later after I get some work done.

That Quality Start BS is also bad:

to go 6 innings and allow 3 or fewer earned runs...  ER* 9 / IP = 4.50 ERA, yes that is real quality