Sabermetrics School and Analysis

[papipapsmanny]

Alright I have multiple questions, of course one relating too tRA which was asked before but not really answered because the thread died.

Alright so you now how it shows a pitcher's "true talent level". Well my questions are why does that true talent level change every year, or if in fact it takes a career to figure out the pitchers "true talent level" then would the stat not be as useful if it is after the fact? Again i just have a problem with the phrase true talent level. I think it should be like tru production level for a season.

And another is why does it seem like VORP is used so rarely? I really like the stat but it seems like its just never used.

[papipapsmanny]

and this is just something i kind of came up with and it may sound stupid.

But is there anyway to figure out how man runs a pitcher is expected to give up with specific defenses? like to actual team defenses?

Like this pitcher should give up this many runs with the Red Sox defense, and this is what he would give up with the yankees defense? So you could figure out if a pitcher would be more valuable to one team as opposed to another?

And dont worry about saying its a stupid question because i feel there is a good chance it is.

[Matt-the-great]

when did this become a forum? i am very excited to see this.

I have recently been getting into sabermetrics pretty hard. I now am a member on Fangraphs, I love their articles and analysis.

It is nice to get away from the mundane realm of the general MLB forum.

[C1Bman88]

And another is why does it seem like VORP is used so rarely? I really like the stat but it seems like its just never used.

I'm uncertain as to what the precise formula is for pitchers, but the formula for position players is downright terrible. VORP uses the most basic form of Bill James' Runs Created formula- which is nothing more than AB*OBP*SLG. That's all. Then it uses a replacement level of ~80%, when it should be more around 73-78%.

As for positional adjustments, VORP has a bad habit of overvaluing first basemen and undervaluing catchers, among other things. Just stay away from it. It'll undervalue the value of a walk just as much as OPS does.

Baseball Prospectus also has RARP, which is essentially the same thing, but they use a different run estimator- EqR. While it's a step up from basic RC, it still undervalues walks.

and this is just something i kind of came up with and it may sound stupid.

But is there anyway to figure out how man runs a pitcher is expected to give up with specific defenses? like to actual team defenses?

Like this pitcher should give up this many runs with the Red Sox defense, and this is what he would give up with the yankees defense? So you could figure out if a pitcher would be more valuable to one team as opposed to another?

And dont worry about saying its a stupid question because i feel there is a good chance it is.

It's not a stupid question. You'd need specific batted ball data, along with player data for each specific vector under their control. Then you use the player's probability of making the play and adjust accordingly, after putting in park effects.

What makes it even more complex is that you'd have to simulate the pitcher against the other team- since you're placing the pitcher on another team, that means he'll be facing different hitters at different points in the season. Also, since we're changing the lineup the pitcher is facing AND the defense, that means a simple single allowed changes the entire course of the game.

You'd essentially have to simulate the whole thing (a la Baseball Mogul), or use an odds ratio (like log5) exhaustively.

It's best to just place the pitcher in front of an "average" defense based on the run values and out frequencies of each event the pitcher surrenders. And that's what tRA does. But it's still an imperfect model.

[Matt-the-great]

i haven't looked through this thread to see if this has been asked/answered...

i am wondering how a ball is deemed to be a line drive?

[Gigantes4Life]

Alright I have multiple questions, of course one relating too tRA which was asked before but not really answered because the thread died.

Alright so you now how it shows a pitcher's "true talent level". Well my questions are why does that true talent level change every year, or if in fact it takes a career to figure out the pitchers "true talent level" then would the stat not be as useful if it is after the fact? Again i just have a problem with the phrase true talent level. I think it should be like tru production level for a season.

You're somewhat nitpicking, but the truth is a player's talent level is ALWAYS changing.

tRA doesn't necessarily show true talent level. As the primer explains, the rates can regress a lot so tRA* is the best measurement of true talent level (of a season).

tRA* is not a measurement of a pitcher's results per se, but should be seen as the system's best estimate of a pitcher's true talent level based on his stats in any given year at any given level. tRA* does not consider a pitcher's statistics from other years and leagues.

And another is why does it seem like VORP is used so rarely? I really like the stat but it seems like its just never used.

It's flawed. It also only considers offense for position players.

and this is just something i kind of came up with and it may sound stupid.

But is there anyway to figure out how man runs a pitcher is expected to give up with specific defenses? like to actual team defenses?

Like this pitcher should give up this many runs with the Red Sox defense, and this is what he would give up with the yankees defense? So you could figure out if a pitcher would be more valuable to one team as opposed to another?

And dont worry about saying its a stupid question because i feel there is a good chance it is.

Well say you're using a projection system. Perhaps ZiPS and FIP. Since FIP is defense independent, you can use fielding projections to add or subtract to the runs. Say a guy's FIP is 3.00 and he pitches 180 innings (60 runs allowed). You could see the fielder's projected +/- runs in 180 innings and then add or subtract it from 60. Obviously there's more to it since some pitchers get more groundballs, etc. so you'd have to look into it to be accurate.

i haven't looked through this thread to see if this has been asked/answered...

i am wondering how a ball is deemed to be a line drive?

MLB.com?

Whoever is watching the game data really. Not sure though.

[viktor06]

Regarding why players have big swings in performance from season to season - 700PA nad 200IP are really small samples when it comes to statistical analysis...just calculate for yourself what the difference for a hitter is if he has 10more singles fall in, or 10doubles, or if he flies out to warning track instead of hitting it out 5times....

That's the biggest problem and reason why looking at stuff like LD%, hit ball speed, plate discipline is important...because if you are unlucky, you will have bad numbers over 700PA even if you are Albert Pujols

[k_rock923]

I saw IsoP mentioned in the formula's thread. Can someone give me a quick description of what it measures and what it tells us?

[Gigantes4Life]

IsoP is the same thing as ISO as far as I know, and stands for Isolated Power. It measures raw power, and is the difference between SLG and BA.

Some variations account triples as the same as doubles (crediting them to speed more than power) making the calculation (2B+3B+3(HR))/AB

[Matt-the-great]

the other day i asked how a ball is deemed to be a line drive?

i am wondering if that is determined by some program/site that measures height, distance, and trajectory of each batted ball? and thereby classifies balls as line drives or not....

[Zep]

I was trying to read up on tRA, and found this, which was extremely helpful, however I also found this regarding FIP. Which metric do you all feel is better for determining a pitcher's value, and why?

[Gigantes4Life]

Definitely tRA, as it considers all part of a pitcher's game. FIP is limited, while tRA accounts for everything.

[papipapsmanny]

i just don't like tRA according to it Lester is having a "better" year than Beckett, but that isnt the case at all, and I wouldn't have to look at any stats to know that.

yea maybe lester's K rates are better but he hasn't been unlucky he has just been god awful up until the start of june.

[Gigantes4Life]

Sure, believe what you want. But Lester's .340 BABIP suggests that his BAA is too high, and he K's 9/9 IP, and only walks 3. He also has a very high GB rate, which is always a plus. Beckett's numbers are basically identical though, he just hasn't been lucky nor unlucky. And you're nitpicking, the difference between their tRA is .03.

[papipapsmanny]

ok? but beckett has done better.....

maybe tRA says lester is supposed to be better but he hasn't been