Phillies fans on Twitter, Facebook, and Internet message boards were infuriated last night when Ryan Madson blew a three-run lead and as sure a victory as the Phillies are going to get this year. The disappointment was justified as the Phillies had a 98.7% chance of winning the game with two outs and a runner on first base in the ninth inning. Even when Troy Glaus hit the two-run home run, the Phils were 95.9% favorites.
Madson’s latest outing coupled with previous struggles in the ninth inning has led many to conclude that he is unfit psychologically to handle the responsibilities of closing out games. Really, it is no different than the criticisms of Cole Hamels being a sissy or mentally weak or what have you. It’s armchair psychoanalysis, as I like to call it.
And it’s bunk science based on small sample sizes, confirmation biases, and misinterpretation of data.
Here’s how it happened:
- Ryan Madson quietly performed his job well last year, but prominently struggled in several notable games which stuck in the memories of many Phillies fans. Phillies fans suspect that Madson may be unfit to be a closer.
- Madson succeeds as a closer several more times. Phillies fans don’t pay any mind because there are more interesting things to pay attention to and because most relievers, even bad ones, will convert saves every now and then.
- Madson fails to convert a save opportunity. Coupled with Phillies fans’ previous view of Madson, this only reinforces their belief and serves as evidence.
- Lather, rinse, repeat.
If you have been reading this blog with any regularity, you have heard me harp about small sample sizes over and over again. With a limited amount of data, we can draw almost zero conclusions with any confidence. If you’re trying to figure out if a coin may be rigged, you do not flip it twice and call it rigged if you get two heads or two tails. You flip it many times — 50 or 100 times if not more. The odds of getting peculiar results with a fair coin become much, much smaller with more trials.
The same thing applies to baseball statistics. In seven appearances, the odds of allowing two home runs with five fly balls is a lot higher than allowing 14 home runs in 35 fly balls in 49 appearances. Ryan Madson has, thus far in 2010, allowed two home runs with five induced fly balls. We know that pitchers do not have control over their home run rate aside from their ability to induce fly balls in general. A pitcher’s HR/FB% will normally hover around 10% so Madson’s 40% really sticks out.
Similarly, we know that pitchers have very little control over batted balls put in play. A pitcher’s BABIP will tend to hover around .300 and actually several points lower for relief pitchers. Ryan Madson’s current BABIP is .405. The odds of 40.5% of 22 batted balls falling in for hits is a lot higher than 40.5% of 220 batted balls.
The other statistical principle we know to be true is regression to the mean. Ryan Madson may have had 40% of his fly balls hit for home runs but that will not hold true for an entire season. Of the remaining 65-ish fly balls that Madson will likely allow over the rest of the season, 10% of them will likely be home runs.
With a little bit of logic and knowledge of statistics, we are able to write off Madson’s poor start to the 2010 season as aberrant. That should be enough for most rational people, but there will still be some out there who, upon reading this, will still claim that Madson doesn’t have the mental toughness to pitch in the ninth inning. Take a gander at Ryan Madson last year:
- 8th inning: 48.1 IP, 47 K, 15 BB, 3.54 ERA (3.20 xFIP)
- 9th inning: 23.1 IP, 27 K, 6 BB, 3.47 ERA (2.91 xFIP)
Ryan Madson is a relief pitcher who strikes out a lot of hitters, walks very few, and induces a lot of ground balls. I will take that pitcher every day and twice on Sunday. And so should you.