As has been shown in other threads, there is room on every team for one relief pitcher who is just as important as the starters. Those high-leverage innings add up. Marmol at this point I would argue projects at just one win above replacement worse than Oswalt for the season and is younger and cheaper. I'd have to think hard before making that trade straight up. How do you factor win projections for relievers? Odds of winning before and after he enters the game. Take a chart like fangraphs.com uses. It determines each teams odds of winning based on the difference in the score, the baserunners, the inning and the outs. For example, in Marmol's last appearance, he entered the game with none on, none out, his team leading by one in the bottom of the 10th. Historically, the leading team wins 79.8% of the time of those situations. He finished the game with the Cubs winning (100% chance of win), so he is credited with .202 wins. If he had blown it and they lost right there, he would have been lost .798 wins. Like I said, a shut-down reliever used in close situations late in the games can have a huge impact and rack up a lot of wins for the team that VORP or ERA/IP wouldn't show.