Pitching Stat

[Nox]

2nd point: Again that's over a huge sample and the larger the sample size the less the luck or "real" part of baseball effects the data.

 

Nobody is suggesting using any numerical metric in an insufficient sample size.

 

Conversely though, there is a required sample size for any scouting based evaluation too. If you went by a scout's first look on a player, you'd do worse than a random number generator.

[theblujay]

Nobody is suggesting using any numerical metric in an insufficient sample size.

 

Conversely though, there is a required sample size for any scouting based evaluation too. If you went by a scout's first look on a player, you'd do worse than a random number generator.

 

I'm just arguing with the people on this board who take all SABR stats as gospel and seem to completely ignore all traditional stats, visual observations and the human elements of the game.

[theblujay]

Are you jaybooster?

 

No I've only had one account

[Nox]

I'm just arguing with the people on this board who take all SABR stats as gospel

 

That's not really how it seems when you say things like "FIP isn't based on real things". It seems like you're out to completely discredit analytics instead of providing thoughtful counterpoints.

[Angrioter]

I'm just arguing with the people on this board who take all SABR stats as gospel and seem to completely ignore all traditional stats, visual observations and the human elements of the game.

 

The Baseball is the same. Now we have better methods to evaluate the performance.

 

An iPhone 5 is better than Qualcomm; We call this evolution.

[theblujay]

That's not really how it seems when you say things like "FIP isn't based on real things". It seems like you're out to completely discredit analytics instead of providing thoughtful counterpoints.

 

That's kind of what the discussion degraded into after I was bashed for disagreeing with some of the regulars.

[Save Dtieb]

(H + BB + HBP + ER) * 9 / IP = X ||| X * BABIP = stat. Am I getting closer?

 

Also, I don't like how xFIP uses an "average" HR rate, the pitchers actual HR rate would be better. Dickey's home run rate for the Jays is higher than the average home run rate around the league, so I'd rather know what his ERA "should" look like playing for Toronto. Not as if he was playing for all 30 clubs and playing the same amount of time in every stadium. He's plays for Toronto, therefore he will give up more home runs due to the park.

 

I think every stadiums home run/fly ball ratio should be taken into account, and not just a league average. SkyDome ratio * 81 + Yankee Stadium ratio * (number of games) + Fenway ratio * (number of games) + etc, etc, / number of projected starts (or something like that). I don't have time to do the math here, but I think if you take into account the games on a particular schedule, and use that home run / fly ball ratio, you'll see a better projection.

[GD]

That's kind of what the discussion degraded into after I was bashed for disagreeing with some of the regulars.

 

I like your signature.

[TheHurl]

Conversations like these are great for everyone. Five years ago I had such a very basic understanding of analytics. Now, thanks to posters like Nox, NJH and sites like fangraphs I have a just slightly better than basic understanding of these numbers but I do understand what the numbers hope to tell us. That's really all I'd hope from everyone who is employed in baseball, sadly it doesn't happen. I think that there a large number of newer posters on this board that quote numbers without an understanding of them (and what sample size is) it's good to occasionally educate and maybe others will have a 20% understanding like me.

[theblujay]

I like your signature.

 

Thanks

[z3r0s]

Conversations like these are great for everyone. Five years ago I had such a very basic understanding of analytics. Now, thanks to posters like Nox, NJH and sites like fangraphs I have a just slightly better than basic understanding of these numbers but I do understand what the numbers hope to tell us. That's really all I'd hope from everyone who is employed in baseball, sadly it doesn't happen. I think that there a large number of newer posters on this board that quote numbers without an understanding of them (and what sample size is) it's good to occasionally educate and maybe others will have a 20% understanding like me.

 

+1 Hurl. I love hearing from these guys.

[Nox]

(H + BB + HBP + ER) * 9 / IP = X ||| X * BABIP = stat. Am I getting closer?

 

While I appreciate an interest in tinkering with models and the attempt to create something useful, I would discourage you from the current path you're on.

 

Why are hits, BB, HBP and ER all equal value?

Why would you ever multiple the resulting sum by BABIP? Also, as soon as you do that, the contextual /9 innings you attempted to created before gets obfuscated entirely and becomes a useless multiplicative factor.

What, in English, is the result? Why do we want to know this specific result?

 

You should really go read the SIERA discussions (comment sections on posts) on the Book blog. The FIP ones too.

[Save Dtieb]

Occam's razor. It's ok to sacrifice insignificant increases in predictive power if they cost a lot in terms of increased complexity.

 

Wouldn't Occam's razor object to sabremetrics?

[Angrioter]

(H + BB + HBP + ER) * 9 / IP = X ||| X * BABIP = stat. Am I getting closer?

 

Also, I don't like how xFIP uses an "average" HR rate, the pitchers actual HR rate would be better. Dickey's home run rate for the Jays is higher than the average home run rate around the league, so I'd rather know what his ERA "should" look like playing for Toronto. Not as if he was playing for all 30 clubs and playing the same amount of time in every stadium. He's plays for Toronto, therefore he will give up more home runs due to the park.

 

I think every stadiums home run/fly ball ratio should be taken into account, and not just a league average. SkyDome ratio * 81 + Yankee Stadium ratio * (number of games) + Fenway ratio * (number of games) + etc, etc, / number of projected starts (or something like that). I don't have time to do the math here, but I think if you take into account the games on a particular schedule, and use that home run / fly ball ratio, you'll see a better projection.

 

While I appreciate an interest in tinkering with models and the attempt to create something useful, I would discourage you from the current path you're on.

 

Why are hits, BB, HBP and ER all equal value?

Why would you ever multiple the resulting sum by BABIP? Also, as soon as you do that, the contextual /9 innings you attempted to created before gets obfuscated entirely and becomes a useless multiplicative factor.

What, in English, is the result? Why do we want to know this specific result?

 

You should really go read the SIERA discussions (comment sections on posts) on the Book blog. The FIP ones too.

 

5/10 points for the effort.

0/10 points for results.

[Nox]

Wouldn't Occam's razor object to sabremetrics?

 

Because I'm lazy and had wiki open, here's a good quote:

 

"Occam's razor is used to adjudicate between theories that have already passed "theoretical scrutiny" tests, and which are equally well-supported by the evidence"

Dino stats are not equally well-supported by evidence.

[Save Dtieb]

While I appreciate an interest in tinkering with models and the attempt to create something useful, I would discourage you from the current path you're on.

 

I am just trying to create something that we can all agree on. Something that takes out all assumptions and projections, and simply shows results. Something that encourages hit prevention and better control. I make ER the same value because they are the result of hits and poor control. I don't know how more of a value an ER has over a H and a BB, so maybe that could be determined.

 

(H + BB) + (ER * X) * 9 / IP.

[Save Dtieb]

5/10 points for the effort.

0/10 points for results.

 

Come on, the first post showed that Jeff Locke is not elite and doesn't deserve to be among the leaders. That's worth at least a 1 for results.

[Nox]

Something that takes out all assumptions and projections, and simply shows results.

 

You're not taking assumptions out though. If anything you're adding them (defensive performance remains constant throughout time and across teams, ER are a better indicator of pitcher talent than RA for a couple of quick ones).

 

FIP simply shows results too (BB, K, HR). Just because the result is scaled to ERA doesn't change that fact.

[Angrioter]

(H + BB + HBP + ER) * 9 / IP = X ||| X * BABIP = stat. Am I getting closer?

 

Also, I don't like how xFIP uses an "average" HR rate, the pitchers actual HR rate would be better. Dickey's home run rate for the Jays is higher than the average home run rate around the league, so I'd rather know what his ERA "should" look like playing for Toronto. Not as if he was playing for all 30 clubs and playing the same amount of time in every stadium. He's plays for Toronto, therefore he will give up more home runs due to the park.

 

I think every stadiums home run/fly ball ratio should be taken into account, and not just a league average. SkyDome ratio * 81 + Yankee Stadium ratio * (number of games) + Fenway ratio * (number of games) + etc, etc, / number of projected starts (or something like that). I don't have time to do the math here, but I think if you take into account the games on a particular schedule, and use that home run / fly ball ratio, you'll see a better projection.

 

While I appreciate an interest in tinkering with models and the attempt to create something useful, I would discourage you from the current path you're on.

 

Why are hits, BB, HBP and ER all equal value?

Why would you ever multiple the resulting sum by BABIP? Also, as soon as you do that, the contextual /9 innings you attempted to created before gets obfuscated entirely and becomes a useless multiplicative factor.

What, in English, is the result? Why do we want to know this specific result?

 

You should really go read the SIERA discussions (comment sections on posts) on the Book blog. The FIP ones too.

 

I am just trying to create something that we can all agree on. Something that takes out all assumptions and projections, and simply shows results. Something that encourages hit prevention and better control. I make ER the same value because they are the result of hits and poor control. I don't know how more of a value an ER has over a H and a BB, so maybe that could be determined.

 

(H + BB) + (ER * X) * 9 / IP.

 

Not all hits are created equal

uBB = 0.691

HBP = 0.722

1B = 0.884

2B = 1.257

3B = 1.593

HR = 2.058

[Angrioter]

Btw, you can find normalized versions of FIP and xFIP on Fangraphs in the form of FIP- and xFIP-. Such stats are fairly easy to come up with, it's not really complexity that makes people avoid them.

 

The problem isn't the equation, is that stubborn individuals are reluctant to accept that someone who did not play baseball can create a better evaluation method.

[Nox]

Not all hits are created equal

uBB = 0.691

HBP = 0.722

1B = 0.884

2B = 1.257

3B = 1.593

HR = 2.058

 

Please have this available in your clipboard the next time a mouth breather insists everyone here thinks BB = 1B

[Slot Machine]

Some lazy trolling going on in here. Troll harder!

 

[edit]

 

As per the topic. Visual observations and the human element are very flawed. People see what they want to see a lot of the time, its just human nature.

[Abomination]

(H + BB + HBP + ER) * 9 / IP = X ||| X * BABIP = stat. Am I getting closer?

 

Also, I don't like how xFIP uses an "average" HR rate, the pitchers actual HR rate would be better. Dickey's home run rate for the Jays is higher than the average home run rate around the league, so I'd rather know what his ERA "should" look like playing for Toronto. Not as if he was playing for all 30 clubs and playing the same amount of time in every stadium. He's plays for Toronto, therefore he will give up more home runs due to the park.

 

I think every stadiums home run/fly ball ratio should be taken into account, and not just a league average. SkyDome ratio * 81 + Yankee Stadium ratio * (number of games) + Fenway ratio * (number of games) + etc, etc, / number of projected starts (or something like that). I don't have time to do the math here, but I think if you take into account the games on a particular schedule, and use that home run / fly ball ratio, you'll see a better projection.

 

Stats like xFIP aren't really designed to suggest how well the pitcher has pitched for a particular team, stadium, or league. They're designed to attempt to place all pitchers on a level playing field (by only taking into account data that the pitcher has somewhat direct control over) and evaluate how well each would have performed relative to each other if they faced the exact same batters, in the exact same parks and leagues, and with the same defense behind them.

 

We can't use xFIP to tell us what Dickey's numbers SHOULD look like this year pitching for us. To do so is a misapplication of the statistic. You would have to take something like xFIP as a base and scale it based on the ratios for the teams he's faced along with park factors and team defense factors or something.

[Save Dtieb]

Not all hits are created equal

uBB = 0.691

HBP = 0.722

1B = 0.884

2B = 1.257

3B = 1.593

HR = 2.058

 

Where did you get these numbers? Cause it appears they are the values of hits for the hitter. and not the pitcher. if a pitcher walks the lead off man, its equal to giving up a 1B. No? They both add to OB and a potential ER.

[GD]

Where did you get these numbers?

 

Guts! Section of fangraphs. I'll post a link when I'm home if you can't find it.

 

Seriously, you should go read Fangraphs. Some of the stuff on there is fascinating.

[Save Dtieb]

Stats like xFIP aren't really designed to suggest how well the pitcher has pitched for a particular team, stadium, or league. They're designed to attempt to place all pitchers on a level playing field (by only taking into account data that the pitcher has somewhat direct control over) and evaluate how well each would have performed relative to each other if they faced the exact same batters, in the exact same parks and leagues, and with the same defense behind them.

 

Love this answer. Level playing field. I do think if, let's say, the Jays pick up a new pitcher this off-season. A better way to predict his FIP or xFIP or whatever, would be to factor in the Dome's homerun/flyball ratio and the pitchers groundball rate. Cause after all, half his starts will be there.

[Angrioter]

Where did you get these numbers? Cause it appears they are the values of hits for the hitter. and not the pitcher. if a pitcher walks the lead off man, its equal to giving up a 1B. No? They both add to OB and a potential ER.

 

Those numbers seem high. I usually refer to these: http://www.tangotiger.net/bsrexpl.html

 

Edit: Although it seems to be a contextual difference. Those make sense for wOBA.

 

Hitters

wOBA = (0.691×uBB + 0.722×HBP + 0.884×1B + 1.257×2B + 1.593×3B +

2.058×HR) / PA

[Save Dtieb]

Ok so those are the wOBA numbers. I get it. In regards to the pitcher, I would think those numbers should fluctuate depending on the situation they are in. A single would be worth more with a guy on third, than a walk. But a single with no one on base is worth the same as a walk. But you could factor in the hitter's numbers. For example a walk to Elliott Johnson is worth a lot more than a walk to Miguel Cabrera because you are limiting his slugging potential. A walk to Rajai Davis is worth more, because of his stealing percentage.

 

So that 0.691 number can turn into 1.256 more often than Cabrera.

 

My head hurts.

[Save Dtieb]

Kind of off topic but not really: Is the pitcher's ability to get more called strikes on pitches outside the zone because of umpire favouritism a skill? Or is it luck?

 

Wouldn't that have more to do with the catcher's ability?

[Save Dtieb]

Yeah I understand. Also, a stat showing swings and misses outside the zone would be fun. Hitter Foolery Percentage.