# You Made Me Do It

Joe Posnanski | Mar 12, 2013 |

The point of my previous post about relief pitchers was not to look at specific players but to look at the strategy of the ninth inning closer. I'm not saying that I made any breakthrough -- in fact, let me be clear, I am saying that I DID NOT make any breakthrough -- but I do find it fascinating that the closer, as a basic strategy, has not made teams more likely to win games they are leading going into the ninth.

And, even when you break it down to close games, to one-run leads entering the ninth, teams in general are STILL not more likely to win games going into the ninth because of the closer.

Well, of course -- because you are brilliant readers -- you ask about the specifics. What about the Yankees, for instance? Certainly Mariano Rivera makes the Yankees much more likely to win games they lead going into the ninth inning.

So, yeah, I tried to figure it out. What you find is that it's a pretty small sample size -- even for a long career like Mariano Rivera's. I went through a painstaking and probably incorrect process -- but doing that (and remember I was ONLY looking for this specific situation) I found that 158 times, the Yankees went into the ninth inning called on Mariano Rivera to protect a one-run lead. How many times did he protect that lead? Answer: 140.

What does that mean? Some quick calculations:

Baseball on Baseball Prospectus' expected win matrix, teams going into the ninth inning should win about 85% of the time. In the 2000s, which is when Mariano mostly pitched, the win percentage was .848.

And the Yankees with Mariano? They won almost 89% of the time. Win percentage .886.

So, the average team, with that lead would have been expected to hold the lead in 134 of those 158 games. With Mariano, the Yankees held the lead six more times over 16 seasons.

Just another thing to think about.

## Create your profile

## Only paying subscribers can comment on this post

Sign in## Check your email

We sent an email to with a link to finish logging in.