Berardino on Pelfrey

[JB_Iowa]

Thanks Oxtung. Good stuff.

 

I had actually stayed with Berardino until he hit SIERA, then it got too much for me.

 

There's a reason I took stats for social science majors rather than the harder version.

[Winston Smith]

Of the other 29 teams how many of them have fans talking about how good it would be to bring in Pelfry as a free agent this winter? My guess zero.

Time to move on for him!

[nicksaviking]

I go either way on Pelfrey and his 2.1 WAR.

 

Perkins had 36 saves in 40 save opportunities in 2013

 

Perkins also had a 1.7 WAR (wins above replacement).

 

therefore, any other bullpen guy or chump off the street coulda come in and saved 34 out of 40 games last year, am I understanding that correctly?

 

As far as I know, WAR doesn't consider save percentage at all. Relievers have low WAR mostly due to low innings pitched. Perkins has a good WAR for a reliever because he has terrific stats that are dependant/derived from a stellar ability to strikeout batters.

[nicksaviking]

It's too early in the offseason to want to bring back Pelfrey. At this stage, my hopes are way higher. I don't mind Pelfrey. I have a hard time watching when he pitches, but if he can be as good in 2014 as Correia was in 2013, that would be solid. Like Nick said, if he's your 4th or 5th starter, that's fine. But I certainly wouldn't go multi-year, or much (if any) more than he got last year. Just not enough upside. But, one thing it is clear that Terry Ryan likes is a pitcher who can throw 180+ innings. And that's understandable. Meyer, May and Gibson should be up in 2014. It'd sure be nice for them to not be the top three pitchers in the rotation.

 

Yes, at the very best, Pelfrey is an impulse buy when you're standing at the checkout line at the supermarket. You get him if you've already loaded your cart with the main course. You don't first go to browse the tabloids and gum.

[gunnarthor]

I feel something has to be wrong if on Fangraphs Pelfrey is a 2.1 WAR and on Baseball Reference he was a -0.3 WAR.

 

Fangraphs WAR is based on fip while baseball reference uses RA. It usually is fairly similar but there can be some pretty big differences - Fangraphs WAR has Jack Morris with more WAR (in fewer seasons) than Jim Palmer. bWAR has him less than Brad Radke. Tom Glavine is another, BR has him 10 WAR higher than fangraphs. Jim Kaat is a clear HOFer on fWAR (70 WAR) but 25 WAR lower on bWAR.

 

I'm not sure why there can be some much variation on some pitchers. Radke has 45 WAR on both, for example. Santana is only a 3WAR difference.

 

I've read (but not sure it's true) suggestions that fWAR gives too much credit to bad pitchers although both rated Blackburns 08-09 seasons basically the same. fWAR isn't as harsh on Blackburn's disastrous 2012. And I've also read that stat heads like fWAR more than bWAR. So who knows.

[Jim Crikket]

It's too early in the offseason to want to bring back Pelfrey. At this stage, my hopes are way higher.

 

Exactly. If, come January, you're still in need for a pitcher of Pelfrey's caliber and he's still on the market, you can sign him then. But the last thing you want to do is be faced with an opportunity to sign a pitcher with more upside in December and feel like you have to pass because you're SP shopping cart is already full and Pelfrey is one of those guys already in the cart. And, honestly, it's not fair to him to sign him early and give the impression he's going to compete for a starting spot and then go out and sign 2-3 higher end guys.

[The Wise One]

Now my basic statistics was quite a few years ago, so please feel free to correct me. In any of these ERA predictors you have to consider their correlations SIERA is said to be 0.72. Good, but hardly precise. There is a regression factor. RMS data shows that factor to be greater than 1 for all the calculators for IP less than 200. Standard deviation of 0.5. Essentially their measurement could be up to a 0.5 worth of ERA wrong. One way or the other. 4.64 by SIERA is the predictive value, by deviation then the predictor says anywhere in the 4 range. Now SIERA does not ignore balls in play. To be an accurate predictor things have to be static. The one thing Twins fan hope for is for is for things not to be static. The predictors say he will be a better pitcher than last year. None say he will be a great pitcher. That is about all you can take away from this.

[Oxtung]

There is another line of thought however. Some people don't care about ERA because it is too vague of a statistic. A home run is a home run; the previous statistics don't differentiate between a home run allowed when your team is winning 7-0 or when your team is tied in the bottom of the 9th inning. Clearly the home run when you're up 7-0 means much less than a home run in the bottom of the ninth of a tie ball game.

 

There are two basic premises of Win Probability Added (WPA) and similar stats. First, a pitcher (or batter or fielder as these stats work equally well for offensive players) is more valuable, in either a positive or negative way, in close or late game situations than he is early on during the game or in blow outs. This is referred to as the Leverage Index. The later in the game and the closer the score the higher the Leverage Index.

 

Second, because we have an extensive history of the play by play of baseball we can determine the average outcome for any given situation. We can then compare how much better or worse Mike Pelfrey did than the average pitcher. Another way to say that is we can determine how much Mike Pelfreys actions affect the Twins probability of winning a game, for better or worse, when compared to an average pitcher. Imagine Pelfrey faces a bases loaded situation. We can look back and determine what an average pitcher actually did in that situation, let's say the generic pitcher gave up a hit and 2 runs. If Pelfrey strikes a batter out we know that he actually increased the Twins odds of winning the ball game when compared to the generic pitcher that allowed 2 runs. If Pelfrey's next pitch is hit for a home run he has now decreased the probability of the Twins winning compared to the generic pitcher. When you add up all of the changes in probability throughout a game, or season, and then factor in the Leverage Index you get the stat Win Probability Added. For WPA a 0 is average and for every whole number above zero one win has been added to the team's win total. Similarly every number below zero is a game lost that an average pitcher wouldn't have.

 

Well I didn't mean to write a novel, but for the second time today it seems I have. I didn't touch on all the statistics but those are several. If you have questions about any others or feel I misrepresented a statistic above feel free to ask or correct me. Again, sorry for the novel.

 

I feel I need to clarify this a little bit because my explanation of WPA wasn't entirely accurate. I was trying to convey the idea in the simplest way and I lost some of the accuracy.

 

WPA doesn't explicitly use the Leverage Index. It instead is automatically built into the way WPA is calculated. For instance, if the Twins hit a home run in the first inning their chances of winning go up but not dramatically because there are still 8 innings to play. So the WPA added for that home run is small. On the other hand if the Twins hit a home run in the 9th inning of a tied game their probability of winning that game increases significantly. So the WPA add for the 9th inning home run would be significant.

 

So you can see that Leverage Index, which is in itself a statistic, is not actually used in the Win Probability Added statistic. I was simply trying to show that late game and close game situations are more valuable than early or blow out situations.

 

There is a statistic, WPA/LI, that uses both Win Probability Added and Leverage Index. It attempts to strip the late and close game bonus that is automatically a part of WPA out so that 2 players contributions can be compared even if they don't fill the same roll on a team. For instance, Ryan Pressly pitched almost exclusively in low leverage situations. That isn't his fault, the manager made those decisions. So what if you want to compare him to Glen Perkins? Well if you use WPA Perkins (2.8 WPA) is clearly going to be better than Pressly (0.2 WPA) because Perkins was frequently in high leverage situations. So WPA/LI is a way to remove the advantage that Perkins gets and puts them on a level playing field. So what does WPA/LI say about Pressly (0.5 WPA/LI) versus Perkins (1.2 WPA/LI)? Well as you can see Perkins was still a better pitcher than Pressly even if they were used in the same roles.