Sabermetrics | Introduction and Discussion

[BigCecil]

Pounds/AB?

 

Thanks. That works. I need the help on the 'metrics. I failed "Stats" @ Acadia University in 1986.

 

Playing football...working out...and partying every night were not courses. Who knew?

[o2cui2i]

we'll let this stay sticky as a reference for users to be able to go into as long as it's an open dialog and kept at a good level of respect between community members.

 

 

thanks

[jays4life19]

we'll let this stay sticky as a reference for users to be able to go into as long as it's an open dialog and kept at a good level of respect between community members.

 

 

thanks

 

You're welcome.

[Spanky99]

You're welcome.

 

lol

[Abomination]

Can we get the title corrected? The word is 'sabermetrics.'

 

I believe both forms are actually correct.

[Jimcanuck]

no question sabermetrics will be reasonably accurate for players with at x number of years of data. using past performance to predict future performance in this case, barring injury, works.

 

for prospects, the data is sparse, and MiLB statistics can be misleading. for prospects, scouting > sabermetrics.

 

my question is, at what point does the crossover occur and sabermetrics exceed scouting reports in value?

[GD]

go away

[Jimcanuck]

Sabermetrics is all about using all the available information to make better decisions by removing bias and not relying on unfounded assumptions. The most accurate approach will blend a player's statistics with his scouting reports. Teams generally have enough scouts and looks at prospects that they can use statistical methods to better extract information from their scouting reports.

 

Scouts have statistics on players and use them to influence their opinions. In recent years, we've seen top prospects lists being increasingly driven by statistics. So it is difficult to separate them into two distinct camps as you're supposing.

 

Not separating them. the question is when does sabermetrics exceed scouting in value. in other words both have value, but depending on the stage of the player's career, one has more relative value than the other. for example, the stats of a high school player probably has little value compared to a scout's assessment of potential

[Boxcar]

Not separating them. the question is when does sabermetrics exceed scouting in value. in other words both have value, but depending on the stage of the player's career, one has more relative value than the other. for example, the stats of a high school player probably has little value compared to a scout's assessment of potential

 

Yeah, of course no one would argue there's much to be gleamed from a high schooler's stats. However, and you see it with top 100 lists all the time, scouts react more to stats than stats guys do. A pitcher can be pitching well with some bad ERA luck and that will influence a scout negatively more than a real sabery guy.

[GD]

Okay, I understand what you're getting at. Double-A is commonly cited as the level at which stats really matter. If you can hit/pitch at an appropriate age in AA, you're set. Below that, tools are generally more important so having good scouts in the low minors can be very valuable.

 

I believe that JFAS has produced some research on this concept before. Good question.

 

Not beyond his Age/Level post.

[Jimcanuck]

Okay, I understand what you're getting at. Double-A is commonly cited as the level at which stats really matter. If you can hit/pitch at an appropriate age in AA, you're set. Below that, tools are generally more important so having good scouts in the low minors can be very valuable.

 

I believe that JFAS has produced some research on this concept before. Good question.

 

I can buy that, the jump from A+ to AA is the most difficult. Suggesting the talent at AA is 'major league'. So stats below AA shouldn't be given much weight......

[Jimcanuck]

Not beyond his Age/Level post.

 

Be quiet, the adults are talking

[GD]

Be quiet, the adults are talking

 

Thanks for the contribution to the thread. Glad you're in here. Maybe you'll pick up a thing or two.

 

probably not, though

[Jimcanuck]

Thanks for the contribution to the thread. Glad you're in here. Maybe you'll pick up a thing or two.

 

probably not, though

 

you might be interested to know that from 2012 to 2014 i compiled and analysed over 8000 records of pavement performance, consisting of both paired discrete and continuous data, and used advanced statistics to generate expected performance levels with time for 12 performance categories. it is pioneering work. will be presented at ottawa in the fall to national experts and again in washington early 2016 to international experts.

 

i get that you are passionate about stats applied to baseball but keep in mind there are flaws to every model..... the limitations of the statistical modelling i carried out were recognized early on so the results corroborated with detailed field evaluations of in-service pavements. kinda similar to sabermetrics corroborated with scouting i suppose

[GD]

you might be interested to know that from 2012 to 2014 i compiled and analysed over 8000 records of pavement performance, consisting of both paired discrete and continuous data, and used advanced statistics to generate expected performance levels with time for 12 performance categories. it is pioneering work. will be presented at ottawa in the fall to national experts and again in washington early 2016 to international experts.

 

i get that you are passionate about stats applied to baseball but keep in mind there are flaws to every model..... the limitations of the statistical modelling i carried out were recognized early on so the results corroborated with detailed field evaluations of in-service pavements. kinda similar to sabermetrics corroborated with scouting i suppose

 

Congratulations, but that still doesn't give you the right to come into this thread and act like an asshat towards me when I literally hadn't said anything to you, now does it? I was answering North's question and you told me to shut up cause the grown ups were talking.

 

I know barely anything about statistical modelling, and I just learned R this summer. No s***, I'm still learning. I've never claimed that models are perfect. I've never claimed that I've created the perfect model, or that there's one end all be all perspective. But just cause you're a smart engineer doesn't mean I can't disagree with you, it doesn't make my points invalid, and it really doesn't give you the right to tell me to shut up for answering someone's question.

[Jimcanuck]

Yep, AA is a big jump.

 

Here's the article from JFAS that contains relevant information: http://www.breakingblue.ca/2014/02/04/projecting-hitting-prospects-mlb-primes/

 

Nox posted this great table in February:

 

Thanks, this is good. Would be useful for JFAS to verify his model by using it on established MLB'ers ie. determine the coefficient of correlation using MiLB and MLB stats for these.

 

Nox' table is subjective(?), not agreeing that game events in Rk and A- ball is useful.... prospects at those levels are still generally quite raw. Would use "borderline useful".

[Jimcanuck]

Not beyond his Age/Level post.

 

My apologies, i misinterpreted the above as directed to me and being derogatory......

[King]

My apologies, i misinterpreted the above as directed to me and being derogatory......

 

how do you misinterpert this as derogatory

[GD]

My apologies, i misinterpreted the above as directed to me and being derogatory......

 

I'm a little confused but apology accepted I think (confused, though)

 

I just hope you don't discredit my views because of my inexperience/differing perspective from yours, I don't think that's fair at all

[Jimcanuck]

how do you misinterpert this as derogatory

 

maybe from being old (turn 50 on Mon) and its late

[Jimcanuck]

I'm a little confused but apology accepted I think (confused, though)

 

I just hope you don't discredit my views because of my inexperience/differing perspective from yours, I don't think that's fair at all

 

no such thing as improper views.... even profs learn from their students from time to time

[GD]

Do you guys think there's anything to the notion that defensive players can take chances away from one another? Like Longoria/Assdribble in TB. Assdribble has been great in the field this year, and so has Longoria. Since they play on the same side of the field, purely speculatively, idk, maybe Longo takes away some chances Assdribble could have made? Just a thought.

[Jimcanuck]

Do you guys think there's anything to the notion that defensive players can take chances away from one another? Like Longoria/Assdribble in TB. Assdribble has been great in the field this year, and so has Longoria. Since they play on the same side of the field, purely speculatively, idk, maybe Longo takes away some chances Assdribble could have made? Just a thought.

 

Sure.... also a bad SS can mishandle or miss the tag on a bullet from the catcher on a steal attempt (Reyes/Martin). LF chasing after a ball the CF has called, who drops with error charged as a consequence (Davis/Rasmus last year).

 

I would expect the impact of crap defenders on the stats of good defenders is greater than impact of good defenders on each other (Assdribble/Longo).

[Spanky99]

You're like the Bill James of pavement!

 

Pretty cool, Jim.

 

lol

[Abomination]

Congratulations, but that still doesn't give you the right to come into this thread and act like an asshat towards me when I literally hadn't said anything to you, now does it? I was answering North's question and you told me to shut up cause the grown ups were talking.

 

I know barely anything about statistical modelling, and I just learned R this summer. No s***, I'm still learning. I've never claimed that models are perfect. I've never claimed that I've created the perfect model, or that there's one end all be all perspective. But just cause you're a smart engineer doesn't mean I can't disagree with you, it doesn't make my points invalid, and it really doesn't give you the right to tell me to shut up for answering someone's question.

 

You should look into the Julia language as well as R. (http://julialang.org/) I haven't had time to develop my stuff yet (my other project is almost finished though I think, so I should be able to soon). I'm planning on trying out Julia when I do my pitchfx database and analysis programs.

[BigCecil]

GD, any thoughts on how Lawrie can be putting up -4.5 WAR on D 2015?

 

Unless we missed something he was + 3B as Jay.

[Jimcanuck]

do the saber calculations weight all data equally, or is more recent data weighted more heavily? do the projections detect non-linear changes in the data or account for disturbances such as TJS?

 

baseball is a complex system with a great many variables. Just questioning how accurately sabermetrics accounts for this.

[GD]

GD, any thoughts on how Lawrie can be putting up -4.5 WAR on D 2015?

 

Unless we missed something he was + 3B as Jay.

 

He hasn't put up -4.5 WAR, he's put up -4.5 runs above (well, below) average. -5.7, actually, going by fangraphs. That equates to costing the team about a half a win. In 3/4 of his years in Toronto, his RngR (Range Runs) were above average, and now they're significantly below average; unless it's something about Oakland's stadium, he might not be the same defender anymore. Or it's just noise in the defensive metrics; they're really not reliable until you've got a good sample.

[GD]

do the saber calculations weight all data equally, or is more recent data weighted more heavily? do the projections detect non-linear changes in the data or account for disturbances such as TJS?

 

baseball is a complex system with a great many variables. Just questioning how accurately sabermetrics accounts for this.

 

In all projection systems, more recent data is weighted more heavily and I believe certain injuries are accounted for

[BigCecil]

He hasn't put up -4.5 WAR, he's put up -4.5 runs above (well, below) average. -5.7, actually, going by fangraphs. That equates to costing the team about a half a win. In 3/4 of his years in Toronto, his RngR (Range Runs) were above average, and now they're significantly below average; unless it's something about Oakland's stadium, he might not be the same defender anymore. Or it's just noise in the defensive metrics; they're really not reliable until you've got a good sample.

 

Thanks for clarifying. One more if you don't mind. I've been looked but haven't gotten clear on it.

 

Statistically evaluating it, how bad was Jeter at SS on D over his career?