Manuel Corpas has been one of the strongest members of our pen this season. He's been absolutely excellent, as he still approaches the plate with a solid control over the strike zone, a low walk rate, and a career-low homerun rate.

Corpas has been one of the weakest members of the pen, however. He's struggled in pressure situations and allowed back-breaking hits and runs where we've come to expect outs.

So would the real Manny Corpas please stand up?

Statistically speaking, we can paint pictures of both versions of Corpas this season.

The first picture is coming from my typical analysis of a pitcher's performance. Right now Corpas is sporting a 6.42 K9, which is right in line with his career totals. His HR9 is at 0.80, which right in line with his career numbers. However, the most impressive part of Corpas' basic rate stats has to be his BB9. Coraps is walking a miniscule 1.87 batters per 9 innings. Only Mariano Rivera, Chad Qualls, Scott Downs, Matt Guerrier, Joe Nathan, and Dan Wheeler have a lower BB9. Summing that whole total out is a 3.49 FIP, which may not be an elite number, but it's hardly a number to sneeze at. His 3.74 tRA supports the notion that his 5.88 ERA isn't really painting the whole picture of Corpas' season.

Join us after the jump to see what the opposition says about Manny.

Corpas' 5.88 ERA bridges us perfectly into the second picture: Shaky, undependable, and absolutely ready to get shelled when he needs to be nails. Corpas is currently sporting a .366 BABIP, which any of the SABR guys around the site will inform you is completely unsustainable, given a .315 career BABIP. Basically, he's getting a bit unlucky - or is he? For the 2009 season, Corpas is allowing a .813 OPS to batters, which may be a bit inflated by the way batters are hitting him since he returned from the DL: 1.355 OPS (.400/.455/.900).

According to Fangraphs' analysis of Corpas' pitches, his fastball has been somewhat hittable, and his slider has been a positive pitch for him this season. He's slowly working in a passable changeup, which if he keeps mixing in properly, could be a solid out-making pitch for him. Batters won't be able to just sit on the slider and wait for the fastball much longer.

But this brings us to a new statistic that we've been seeing in the postgame Fangraphs in the game wraps: WPA.

Win probability added is based, once again, on Tom Tango's work with past data. Fangraphs gives an excellent explanation of how WPA works, and you can read it here. I'll attempt to give an explanation of it as well here.

What Tango's work did was to take every single base-out state (for example, 0-on, 0-out is a base-out state, and 1^st and 3^rd with 2 out is a base-out state) for every score differential and determined how many games the home team won after that particular base-out state.

So let's show an example here. Rockies vs. Astros, 6/4/09, Jason Hammel v Wandy Rodriguez.

The game starts with both teams having a 50% (.500) probability of winning the game.

To lead the game off, the base-out state was 0 out, no runners on. Ryan Spilborghs grounded out to 3B to create a base-out state of 1 out, no runners on, which gave the home team a win expectancy of .565, or a 56.5% chance of winning. Therefore, Wandy Rodriguez now has a WPA+ of .065, and Spilborghs has a WPA- of .065.

With 1 out and nobody on, Clint Barmes inexplicably drew a walk, leading to a base-out state of 1^st base occupied, 1 out, which changed the Astro's win expectancy to .534. Therefore, Rodriguez earned himself a WPA- of .031 (.565-.534), and Barmes a WPA+ of the same, .031.

With 1 out and Barmes at 1B, Helton singled to LF, creating a base-out state of 1^st and 2^nd, 1 out, with a win expectancy of .500. Helton gets a WPA+ of .034 (.534-.500), and Wandy a WPA- of .034.

With 1 out and 1^st and 2^nd, Matt Murton struck out, boosting the ‘Stros win expectancy to .559. Murton gets a WPA- of .059 (.559-.500) and Wandy a WPA+ of the same, .059.

With 2 out and 1^st and 2^nd, Brad Hawpe grounded out to 3B to end the inning, and the Astros now had a 59.1% chance of winning. Hawpe gets a WPA- of .032, and Wandy a WPA+ of .032.

Does your head hurt yet? Mine does.

Fast forward to the top of the 3^rd inning. With Ryan Spilborghs on 1B with no outs, Barmes triples to drive Spilborghs in. Barmes receives a WPA+ of .166 (.487-.321), and Wandy receives a WPA- of the same level. Run scoring plays clearly give good WPA+ to the batters.

As the game continues, players score runs and create outs, leading to varying levels of WPA+ and WPA-. At the end of the game, you take each player and find the difference in their WPA+ and WPA- (WPA+ - WPA-) to determine their overall Win Probability Added for the game.

At the end of this particular game, Clint Barmes was the big hero, providing a .150 WPA, or in other words, he added 15% to his team's chances of winning. Wandy was the big loser, with a WPA of -.334, or he cost his team 33.4% of a chance of winning.

Now there is one quick nuance to look at when reading over the WPA charts at the end of the game. The winning team's WPA will always add up to .500, and the losing team's to -.500. My intial thought was "But wait shouldn't the winning team have 1.000 and the losing team 0.000??" but then I remembered how arithmetic works: Both teams start with .500, and if the winner gains .500 and the loser loses .500, that means the the winner ends up with 1.000 and the loser with 0.000.

Hurray, the stats guy knows how to do math!

So what does this have to do with Manny Corpas, many of you may be asking?

Well, this season Manny Corpas has a WPA+ of 1.94, and a WPA- of -2.78, giving him an overall WPA of -0.84 for the season. What does this tell us?

Corpas hasn't really done well for himself this season. While he's posted good peripherals, he's saved his moments of weakness for the most untimely of situations. Granted, he's nursing bone chips in his elbow, but if Jim Tracy plans to continue sending him out, Corpas has to figure out how to turn some of these ugly hits into outs.

So to sum up our Corpas discussion, the Pros include: Good FIP, decent tRA, good K9, BB9, HR9. Cons: bad ERA, bad BABIP, bad WPA. Is he still a good tool for our bullpen? Absolutely. Is his elbow still hurting him? Most definitely.

Do we need him?

You bet we do.