Jim Leyland on high-velocity pitchers

[DrungoHazewood]

You may jest but answer me this, how in heavens name can a player be properly evaluated or even accurately evaluated defensively soley by stats? Say you have two players that play third base. PlayerOne has 125 put outs out of 128 chances with three errors two of which were throwing errors. This same thirdbaseman also made at least 4 out of the world highlight film plays out of the 128 chances that no other thirdbaseman in the game would have likely made saving his pitcher from giving up untold runs and helping his team win several tight one run games.

Player two has 130 put outs out of 132 chances with two (non-throwing errors) where he booted routine grounders that were tailor made double plays. His errors led to big innings and specifically put his team behind in the game where they never recovered. He made one highlight film play during this stretch, that was a great play but nothing close to the plays made by the first player.

Now tell me, which player is better? If going by stats you would certainly say the second player based on his stats but guess what, you would clearly be wrong. This is why I think stats (especially defensive stats) are way over-valued. It is a representative illustration as to how a guy like Brooks Robinson who routinely made plays that I doubt many who never saw him play ever saw another thirdbaseman make could fathom. Yet these plays just show up in the stats books as a put out.

I never said stats were infallible, that they were the only method of evaluating players, and I certainly never said that basic, unadjusted numbers like putouts and total chances were any way to analyze someone.

All I've said is that relying on only your own subjective observations is more flawed than even a decent numbers-based system.

[mefogus]

What does Wakefield do to your stats? He has probably the best "movement" of any pitcher's pitches yet throws with the least velocity? Just goest to show that there are flaws when you view only numbers and not the actual game and its players right?

You mention Wakefield, but if you look harder you will find perhaps hundreds of pitchers that are counter to Drungo's post. However, there will conversely be thousands who will fall right in line with it. I still think you are not quite understanding that Sabermetric systems are not designed to assert truisms, but instead are used to illustrate tendencies and likelihoods. Pulling out Wakefield does nothing to dispel this.

-m

[Lucky Jim]

You mention Wakefield, but if you look harder you will find perhaps hundreds of pitchers that are counter to Drungo's post. However, there will conversely be thousands who will fall right in line with it. I still think you are not quite understanding that Sabermetric systems are not designed to assert truisms, but instead are used to illustrate tendencies and likelihoods. Pulling out Wakefield does nothing to dispel this.

-m

Yes, O5F has been mistaking "exceptions" to probabilistic predictions as refutations of their overall correctness for months now.

[Old#5fan]

You mention Wakefield, but if you look harder you will find perhaps hundreds of pitchers that are counter to Drungo's post. However, there will conversely be thousands who will fall right in line with it. I still think you are not quite understanding that Sabermetric systems are not designed to assert truisms, but instead are used to illustrate tendencies and likelihoods. Pulling out Wakefield does nothing to dispel this.

-m

To me "tendancies and likelihoods" are not all that valuable or important when there are so many exceptions to these tendancies. Ergo, you cannot draw much value out of "tendancies" for specific individual players in many instances, as in Wakefield for example. So again, we are back to looking and observing individual players and making a judgment.

I think one of my problems is I work for an organization that analyzes all kinds of aspects of my job via statistical measurements, which essentially are meaningless in many cases, yet moronic managers who don't know how the job actually works think they find flaws or areas that need to be improved by the worker. Ergo, I think those who rely overly so on looking at reports, stats and surveys generally are some of the more clueless people in the organization as to what is actually going on. So to me "tendancies or likelihoods" are just stats with little practical application or meaning as generalizing about anything is pretty much worthless in value.

[Old#5fan]

I never said stats were infallible, that they were the only method of evaluating players, and I certainly never said that basic, unadjusted numbers like putouts and total chances were any way to analyze someone.

All I've said is that relying on only your own subjective observations is more flawed than even a decent numbers-based system.

Let me simplify it for you then. If you see a 90 MPH fastball coming at your head do you need to analyze the speed, trajectory, pitching mound slope, height and weight of the pitcher, whether it is during the day or night, at home or away LH or RH pitcher, reliever or starter, or would you just duck, and not worry about all these periphal things that are really of very little importance? (Now admittedly, this is kind of off the wall, but you have to get what I am saying.) I mean, in this case as in the majority of instances a subjective observation is superior over anything else.

[DrungoHazewood]

Let me simplify it for you then. If you see a 90 MPH fastball coming at your head do you need to analyze the speed, trajectory, pitching mound slope, height and weight of the pitcher, whether it is during the day or night, at home or away LH or RH pitcher, reliever or starter, or would you just duck, and not worry about all these periphal things that are really of very little importance? (Now admittedly, this is kind of off the wall, but you have to get what I am saying.) I mean, in this case as in the majority of instances a subjective observation is superior over anything else.

:SuN024:

........................

[DrungoHazewood]

To me "tendancies and likelihoods" are not all that valuable or important when there are so many exceptions to these tendancies. Ergo, you cannot draw much value out of "tendancies" for specific individual players in many instances, as in Wakefield for example. So again, we are back to looking and observing individual players and making a judgment.

I think one of my problems is I work for an organization that analyzes all kinds of aspects of my job via statistical measurements, which essentially are meaningless in many cases, yet moronic managers who don't know how the job actually works think they find flaws or areas that need to be improved by the worker. Ergo, I think those who rely overly so on looking at reports, stats and surveys generally are some of the more clueless people in the organization as to what is actually going on. So to me "tendancies or likelihoods" are just stats with little practical application or meaning as generalizing about anything is pretty much worthless in value.

I'm guessing you're the reason your organization can't get its ISO9001 or CMMI certifications.

[mefogus]

I think one of my problems is I work for an organization that analyzes all kinds of aspects of my job via statistical measurements, which essentially are meaningless in many cases, yet moronic managers who don't know how the job actually works think they find flaws or areas that need to be improved by the worker. Ergo, I think those who rely overly so on looking at reports, stats and surveys generally are some of the more clueless people in the organization as to what is actually going on. So to me "tendancies or likelihoods" are just stats with little practical application or meaning as generalizing about anything is pretty much worthless in value.

So in other words: stats screwed you, so screw stats.

-m

[Old#5fan]

I'm guessing you're the reason your organization can't get its ISO9001 or CMMI certifications.

I have no idea what those two items are, but in my organization they want work completed fast, yet thorough and accurately. While this is a valid goal that I certainly have no problem with it is the way they track how it is achieved that is simply poorly done. They evaluate every job as it is should be optimally completed in a certain time frame as a standard, no matter how complicated or easy, complex or simple, or a myriad of other variations that can and do quite often happen. Thus, they fail to take into account important aspects that impact each job and the performer.

This to me is simply a stupid and ineffective way to try to evaulate work performance and actually hinders it. If you must measure something statistically because you don't have a clue how to do the job, at least try to learn the job first and experience or observe it first hand (in the trenches - so to speak) before blaming workers for simply properly handlling a job in the best way possible.

A perfect example of this is just yesterday folks in another department got a call from their boss that they were the slowest in processing customers in the Baltimore area. He wanted to speak to them immediately about the report he just received, which meant shutting down the customer processing and further hurting the stats. This just made no sense at all. Also, he failed to take into account that the ticket processing system they were using had malfunctioned several times distorting the time shown in processing. Again, he was relying on stats that were skewed and jumping the gun without actually observing what practically was going on.

[Old#5fan]

So in other words: stats screwed you, so screw stats.

-m

No, moronic managers try to screw good employees by utilizing misleading or misusing stats. Either way, both are wrong. I was a manager in a former job and I was a hands on manager who used common sense instead of looking at stats and not observing what is going on like some less enlightened folks in management seem fond of doing today.

[DrungoHazewood]

I have no idea what those two items are, but in my organization they want work completed fast, yet thorough and accurately.

They are systems designed to optimize your business by instituting repeatable processes, developing metrics to analyze those processes, then using the analysis of those metrics to increase the quality and efficiency of your business.

[Migrant Redbird]

They are systems designed to optimize your business by instituting repeatable processes, developing metrics to analyze those processes, then using the analysis of those metrics to increase the quality and efficiency of your business.

The objective is certainly laudable, but old timers like me have been through similar processes before, which turned out to have so many embedded inefficiencies that they depressed productivity and drove up operating costs. Does anyone else recall the "Zero Defects" campaign of the late seventies?

[RShack]

The objective is certainly laudable, but old timers like me have been through similar processes before, which turned out to have so many embedded inefficiencies that they depressed productivity and drove up operating costs. Does anyone else recall the "Zero Defects" campaign of the late seventies?

AFAIK, the ones that actually work are based on tracking subjective human judgment, not supposedly-objective stats... it was about systematizing human subjectivity, not removing it.

I know that was true in the great quality strides the Japanese made in the auto-industry, not sure about other industries. In the auto industry, every significant effort to remove the subjective and inconsistent human factors led to major and costly disasters. Same thing happened with automating the control systems for the Brit rail system: they had to re-engineer their zillion-dollar system to put the fuzzy human factors back in...

[DrungoHazewood]

The objective is certainly laudable, but old timers like me have been through similar processes before, which turned out to have so many embedded inefficiencies that they depressed productivity and drove up operating costs. Does anyone else recall the "Zero Defects" campaign of the late seventies?

Maybe lessons learned from earlier systems have been fed back into the new ones to make them more efficient and workable.

[Migrant Redbird]

Maybe lessons learned from earlier systems have been fed back into the new ones to make them more efficient and workable.

I'm sure that's true!