Baseball - Bullpen Value

[Dead__Red__777]

Hello Everyone,

I am pretty new to this forum and I am finding a ton of great information so thank you all. I have a baseball model I have been working on over the past few seasons and a lot of these posts are helping me to tweak it so I can continue to improve my model.

My question for today is about bullpens. I know a good bit about baseball so I know a good bullpen is key. However, what I am not sure of is how predictable a good bullpen can be. Sure - a bullpen is good for a while and gets labeled as a great bullpen - then when it misfires it is "struggling." The opposite can also be true. Plus there are so many different pitchers than can come in for so many different reasons it always seemed very difficult for me to quantify so I just left bullpens out of my analysis.

In Trading Basey Joe Peta actually did the same thing. He claimed that his research showed the bullpens are not consistent year to year (mostly due to changes in staff I would imagine). But if I recall correctly he goes on to say that even bullpen pitchers wavier a good bit and do not show consistency (except for a few obvious examples).

I think some of this can be predicted by accuracy. In a separate project I started to look at accuracy and opposing plate discipline to show how good a reliever is and how likely it is for a bullpen pitcher to have a severe fall-off. I was going to call this study "The Carlos Marmol Effect" since I think he was the best example. A player who was lights out until people literally just stopped swinging. Marmol went from stud to nobody seemingly overnight.

Anyway, sorry for rambling. Please let me know your thoughts and lets talk bullpens!

Red

[TheGoldenGoose]

You don't have to look far to see what a good Bullpen can do for a baseball team. The KC Royals Bullpen literally carried that team on their backs to Game 7 of the World Series last season.

[gryfyn1]

Bullpens will always appear inconsistent because relievers throw so few innings. When throwing just 60-70 innings per year the samples are so small that an average pitcher can look fantastic one year and terrible the next. Only a few bullpens carry more than one or two high quality relievers, the rest are average arms that can swing from looking great to terrible from month to month and year to year.

[EXhoosier10]

http://www.fangraphs.com/library/principles/sample-size/

Basically K% and BB% are the only two semi-reliable stats after 50 IP. Creating/using a statistic that focuses only on K% and BB% is probably your best bet of predicting future bullpen performance. Or just use any projection system that has shown success in the past. With that, you can predict a pitcher's performance and then assign a pitcher to a team and you have a team's bullpen stats

[Dead__Red__777]

http://www.fangraphs.com/library/pri...s/sample-size/

Basically K% and BB% are the only two semi-reliable stats after 50 IP. Creating/using a statistic that focuses only on K% and BB% is probably your best bet of predicting future bullpen performance. Or just use any projection system that has shown success in the past. With that, you can predict a pitcher's performance and then assign a pitcher to a team and you have a team's bullpen stats

Very quality stuff. Thank you, EXhoosier10!!!

[MikeTizzy]

WHIP, k/bb, OPP BA, recent ERA's/9 ip, lefty and righty splits, who's healthy and rested, or if a bullpen been taxed over a 5 game stretch where the starters been gettin shelled. This are the decisive metrics that usually shows how effective bullpens are at any given time. A talented bullpen with a perfect scenario will often dominate late innings, making it very short contests. And vice versa for crap bullpens, and leads are never safe with often blown saves and extra inning play, bad bullpens do equate to more OVERStho, since its a given dynamic by default.

[MikeTizzy]

and yes Fangraphs will teach u a ton.

[Big Bear]

You don't have to look far to see what a good Bullpen can do for a baseball team. The KC Royals Bullpen literally carried that team on their backs to Game 7 of the World Series last season.

if the oakland A's had a better bullpen the Royals wouldnt have made further than the Wild Card game.

they took a 7-3 lead into the bottom of the 8th.

[statnerds]

why would you intentionally introduce additional unknown variables into your model? or do you feel the 5 inn market is so extremely sharp the lines open at a perfect number each and every time? seems to me if you feel the only thing lacking from your model is accurately accounting for BPs, just eliminate them.

side note, not intended to fire anyone up. recognize you are better off modeling the Market and its moves than you are building a model that in practice will be going head to head with the absolute best modelers that money can buy.

GL

[TPowell]

good stuff, will save this for later

[roc024]

Awesome insight, thanks statnerds and everyone. I'm finishing up Trading Bases and will be creating a model for myself very soon.

[Squared Box]

I wouldn't expect to learn anything from Peta's book. It's garbage.

[roc024]

Any material you can suggest Squared Box?

I've been checking out Fangraphs and it has some great articles.

[Squared Box]

Fangraphs and Baseball Prospectus are the best websites. As for books and stuff like that, this one shows how to translate Fangraphs-type data into a line.

http://www.amazon.com/Betting-Baseball-Richard-Nichols/dp/0977878732

Peta's book doesn't do anything like that, was hard to take what he said seriously.

[Big Bear]

Bullpens will always appear inconsistent because relievers throw so few innings. When throwing just 60-70 innings per year the samples are so small that an average pitcher can look fantastic one year and terrible the next. Only a few bullpens carry more than one or two high quality relievers, the rest are average arms that can swing from looking great to terrible from month to month and year to year.

most good bullpens have 2-3 quality arms.

Yankees may have more.

Pirates have 2.

Royals have 3.

Reds have 1.

Giants have 2.

Nationals have 3.

Rockies actually have 2 this year.

[TheLock]

Pitchers end up relievers and not starters for a reason, boys.