New Catcher Framing Values

[Wookiee of the Year]

With correlation comparisons, a perfect 1 is a perfect correlation, indicating causality (doing A causes B to happen) while a negative one is a perfect negative correlation, meaning there is a completely adverse relationship in causality (doing A causes the opposite of B to happen). In most statistical analysis, you want a correlation of 0.5 or better to show some sort of causality between variables. For instance, you can say that the consumption of tea is higher in the summer than in the winter. If your data showed a correlation equivalent of 0.3, you'd likely conclude that tea and summer seem to go together, but there doesn't seem to be strong causality. Basically, if you look at the number it's a measure of how much of 1 is cause and how much is coincidence. A score of 0.3 would mean that 30% of runs allowed and catch framing has something to do with catch framing causing a runs allowed difference and 70% coincidence. Once you get over a 0.5, there's a lot more causality to assign to catch framing changing the runs allowed by a team.

biggentleben--exactly. So if catcher framing had a correlation of 1 with Runs Allowed, it would mean your catcher's framing ability determined it completely. (If the two correlate perfectly in perfect tandem, no other independent factor could contribute. Otherwise, if they diverged, you'd have a paradox--the dependent variable wouldn't know which way to go. And the same would hold true for why you can't have two independent factors correlating at 0.9--if they diverged, the dependent variable couldn't be 90% determined by both of them simultaneously.)

 

So to say 30% of Runs Allowed has to do with catch framing sounds huge to me. That other 70% isn't coincidence, it's all the other factors/variables that affect Runs Allowed, right? That means that pitcher ability, team defense, and non-catch framing catcher skills like pitch selection don't correlate in their totality more than 0.7, because 30% of Runs Allowed is already accounted for. ...Right? (I'd imagine this isn't perfect because it didn't sound like the 0.3 correlation involved a multiple regression analysis to isolate catch framing and make sure it was fully independent of all other factors--but since it's generally considered an ignored skill right now, I'd think it's closer to isolated from other variables than most factors would be.)

 

To come at this from another angle: How well does a team's best player's OBP correlate with Runs Scored? My guess is it's pretty low. (The Twins have had one of the league leaders in OBP on their roster over the past three seasons, after all.) I would think it would be difficult to find correlations over 0.5 between one individual player's performance and a team's overall performance--because no individual player's performance can have that much of a causal relationship with the team's overall performance, even more so if you're measuring just one skill within that player's skill set. ...Right? So to say it's a "weak" determinant at best of team outcomes has nothing to do with whether it's a real, repeatable skill.

[biggentleben]

biggentleben--exactly. So if catcher framing had a correlation of 1 with Runs Allowed, it would mean your catcher's framing ability determined it completely. (If the two correlate perfectly in perfect tandem, no other independent factor could contribute. Otherwise, if they diverged, you'd have a paradox--the dependent variable wouldn't know which way to go. And the same would hold true for why you can't have two independent factors correlating at 0.9--if they diverged, the dependent variable couldn't be 90% determined by both of them simultaneously.)

 

So to say 30% of Runs Allowed has to do with catch framing sounds huge to me. That other 70% isn't coincidence, it's all the other factors/variables that affect Runs Allowed, right? That means that pitcher ability, team defense, and non-catch framing catcher skills like pitch selection don't correlate in their totality more than 0.7, because 30% of Runs Allowed is already accounted for. ...Right? (I'd imagine this isn't perfect because it didn't sound like the 0.3 correlation involved a multiple regression analysis to isolate catch framing and make sure it was fully independent of all other factors--but since it's generally considered an ignored skill right now, I'd think it's closer to isolated from other variables than most factors would be.)

 

To come at this from another angle: How well does a team's best player's OBP correlate with Runs Scored? My guess is it's pretty low. (The Twins have had one of the league leaders in OBP on their roster over the past three seasons, after all.) I would think it would be difficult to find correlations over 0.5 between one individual player's performance and a team's overall performance--because no individual player's performance can have that much of a causal relationship with the team's overall performance, even more so if you're measuring just one skill within that player's skill set. ...Right? So to say it's a "weak" determinant at best of team outcomes has nothing to do with whether it's a real, repeatable skill.

 

Not exactly. What 0.3 means is that for every 1 run change, there's a 30% chance that catcher framing had something to do with that, and 70% chance that it was coincidence that better framing led to a better runs allowed number.

 

Also, realize that catching impacts the game much differently than center field defense or single hitter OBP because you have one player who receives every pitch of the game for a team, typically. If Mauer hit every single time, you'd be able to see how his OBP translated to higher runs scored. The best way to view causality in hitting statistics is through team statistics, not individual, because an individual's hitting has such a tiny bearing on the overall outcome of the game. In a perfectly pitched game against the Twins, for instance, Mauer would come to the plate 3 times. He's only responsible for 11% of the team's plate appearances that day, so 89% of the Twins offense on that day was decided by someone other than Mauer, and the percentage gets smaller the longer the game goes.

 

With catcher framing, it's not true that everything else only accounts for 70% of runs allowed. The statistic is saying that, for example, the average major league team with average defense and an average catch framer will allow 300 runs in a full season. A team could have allowed 290 runs, and they had a good pitch framer at catcher, but the issue is that that framing could be because they went out and signed every Brad Radke and Greg Maddux type in the game, and no one was missing a spot at all, so runs went down because there weren't mistake pitches. That would increase the framing scores because umpires in general like pitches that hit exactly where the catcher put the glove, even if it's a bit off of the plate. What that number is saying is that for a runs change, there's a 30% chance that pitch framing contributed to it. There could still be 100% chance that a slick fielding shortstop contributed to it as well as that 30% chance of pitch framing contributing. They're looking at each thing as independent, so having good pitch framing and average everything else has a 30% chance of leading to better run prevention. I'd be curious to find out what having everything else average, but having high-end defense from shortstop would correlate to in run prevention, because I have a good feeling it's well over 0.5.

[Wookiee of the Year]

Thanks for the reply and for taking the time, biggentleben. I took a couple stats classes in college, but it's obvious I'm rusty when it comes to drilling down much.

 

Let's see if I'm understanding this: Would it be fair to say that if catch framing has a 0.3 correlation, than if the Twins re-ran their 2013 season but replaced Joe Mauer's defense with, say, Jose Molina's, there's a 30% chance they'd have had fewer Runs Allowed?

[biggentleben]

Thanks for the reply and for taking the time, biggentleben. I took a couple stats classes in college, but it's obvious I'm rusty when it comes to drilling down much.

 

Let's see if I'm understanding this: Would it be fair to say that if catch framing has a 0.3 correlation, than if the Twins re-ran their 2013 season but replaced Joe Mauer's defense with, say, Jose Molina's, there's a 30% chance they'd have had fewer Runs Allowed?

 

A better way to look at it would be that if the Twins allow less runs this season with Kurt Suzuki than last season with Joe Mauer, there's a 30% chance that it would have something to do with pitch framing.

[old nurse]

A better way to look at it would be that if the Twins allow less runs this season with Kurt Suzuki than last season with Joe Mauer, there's a 30% chance that it would have something to do with pitch framing.

 

The run scores formula for the extra strike assumes that the extra strike is equally divided throughout the game. This has been shown to be not the case in different studies. The higher likelyhood of getting the extra strike comes from when you are the home team and ahead by a margin, sorry I can't remember the number. The likelyhood goes down of getting the exra strike with men on base. The more men, the less the chance. For there to be a 30% less chance, the other 70% factor would have to be the same. It won't be. The staff is not that static.

To answer Wookie's question, I don't think with the staff of the 2013 Twins Molina would have had a dramatic impact.