THE BOOK--Playing The Percentages In Baseball

It might be chance, but it’s also been a long period of time for a skill (MLB-level hitter) that has a shelf-life of 25 years max for the very best. I’m not sure it is meaningful to try and distinguish between “Markakis lost his power and hit for less power” and “Markakis didn’t lose his power and hit for less power”. If there is nothing to be gained in predicting either his future performance, or other player’s performance, then what is the difference?

I do agree that Bayes has some applicability to this sort of thing, but you can’t use it in retrospect - who right now it hitting more HR than you think they will in the future? And who is hitting less?

People make a defacto prediction based on the “pattern” that he’s lost his power. 

The reality is that the size of the sample doesn’t give us anything remotely a pattern.  And the reality is that, absent INDEPENDENT (*) scouting, your expectation of Markakis should be whatever Marcel says it should be.

(*) What does independent mean?  This is what I would like to see, just once: I’d like for a scout to see a guy who hits 11 HR per 600 PA over the first 4 seasons, who then hits 17 HR per 600 PA over the next 2 seasons to then make this declaration:

“Mick Narkakis has lost his power.  He’s getting weak swings, he’s late catching up to fastballs, and is not turning his wrists fast enough.”

You see?  I want him to say what he sees without relying on the actual outcomes.  Now, obviously, it’s pretty hard to diminish physically, but actually see substantial performance numbers.  But it’s not impossible.  But no one… and I mean NO ONE… will ever ever ever say anything approaching the above given anything close to the context as I laid it out.

That’s because just as people are far too scared to make a scouting observation that goes against what the empirical data might suggest, people are far too willing to make a scouting observation that can be supported by cherry-picked numbers.

I think “how much change in true talent takes place year to year among players” is one of the more interesting unanswered questions in sabermetrics. Can you envision a way to get closer to the answer? Maybe simulate the league (sounds super easy) and observe the variation in players year to year, then compare it to our actual data?

Do you think it’s a question that can be answered with the data and knowledge that we have at this moment in time? Or is it something that might be answered once we ultimately refine our ability to nail down true talent levels?

There is exactly one way to improve the model: break down the mechanics of the player.  The solution can only come down to scouting.  Sportvision and Trackman is what you need, and scouts (the subject matter experts) is who you need heavily involved to create the model.

Think of Mechanic + Mechanical engineer as the scout + saberist.  You need both to solve the problem.

For what it’s worth, Chris Singleton, a man who truly lost his power (17 HR his first season, 28 his other four seasons) last night on Sunday Night Baseball said it was going from a batting coach who taught “top hand” to ones who taught “bottom hand.” Why on earth someone would coach someone to hit with less power is beyond me.  But that’s what he said.

But what happens if 90% of the players actually do get more HR by doing that shift?

Singleton’s insight is definitely valuable.  However, it is n=1, and it was based specifically because we are looking at his change in HR.

Top hand hitting leads to a more level (or choppy when overdone) swing and thus more contact since the bat passes through the path of the ball for a longer period of time, while focusing on the bottom hand promotes a more uppercut/Griffey-esque swing, leading to more powerful flyballs (since what were once line drives with our level swing now have a homerun trajectory) and more flies in general - but the tradeoff is less contact, in general. (Sorry, you may have already known all this.)

The thing is, teaching the baseball swing is like teaching the golf swing… no drill or swing thought works the same way for, or should be applied to everyone. You take a look at what is wrong with the swing, and try to fix it - sometimes directly, sometimes indirectly - with simple swing thoughts. When something clicks for a player, they practice it over and over until something else about the swing needs to be fixed. All this is meant to say it… there could be a very good reason for a coach to provide a hitter with swing thoughts that on face value may actually decrease his ability to hit homeruns.

As for Singleton and his batting coach...without hearing his actual anecdote… it sounds like he is trying to blame his lack of production on his hitting coach.

If we’re Bayesians here, we should also account for the fact that he now chokes up on the bat much more than he used to.

Well, I watched a replay of an interview with Reggie Jackson recently where he described how he choked up on the 1971 All Star Game homer he hit off the transformer on the RF roof at Tiger Stadium.  Video shown along with it confirmed that he did indeed choke up on that swing…

Just sayin’…

Tom, I’ve read your desire for scouts to grade players only on what they see, purposefully ignoring the data. In the current system, scouts are paid, promoted, and hired, at the end of the day, largely on their ability to make accurate assessments of players. As everyone reading this blog is surely aware, stats are a great tool to help in predicting future performance.

I’m sure you’re aware of this, but the reason I bring it up is that I’m wondering, if you were a GM or in a position to implement such a decision, how you incentivize scouts to ignore the data? How would you evaluate who your best scouts are?

I don’t really know if I have anything intelligent to say on the matter at this moment.

Using MLB batted ball data (I know all the limitations, but it is the only data available) here is the average distance in HR and Fly Balls for Nick M. for the last 4 years:

301 ft
298 ft
290 ft
282 ft

Are these values significant? I am not sure (one of my off season projects). I do know that he is hitting fly balls on average 20 ft less now than a few years ago. I would bet a player doesn’t hit has many home runs averaging such a drop.

Great stuff Jeff.

Rather than all flyballs, can you give us the average distance of all batted balls (regardless of FB, LD, GB designation), if you limit it to the 33.33% longest batted balls?

Orioles fan here, and this is something that interests my friends and I.  For the comment about him choking up, he started that like a month and a half ago.  These are the observations we’ve made about Nick from watching him the last few years:

1.  He doesn’t pull the ball as much as before.  If true, this would lead to a drop in HR for him because he doesn’t have HR power to the opposite field.

2.  He seems to swing more defensively than in the past.  What we’re saying here is that he seems to be more concerned with making contact with the ball instead of trying to turn on the ball and hit it hard.

Those two are probably connected, in that swinging more defensively, leads to less balls pulled.  We’re not scouts, but we’ve been passionate baseball fans for more than 20 years.  So take what we say with a grain of salt I guess.

How many PAs does it take for homerun rates to stabelize? I think, that is a relevant question in this context.

This was posted just a few weeks ago by Derek Carty, that I linked to in this blog. 

http://www.insidethebook.com/ee/index.php/site/when_is_the_observed_data_half_real_and_half_noise/

About half a year for hitters and three years for pitchers.

I wrote about this for baltimoresportsreport.com this week. In particular used HitTracker to graph all of Nick’s MLB home runs.
The graph is here : http://baltimoresportsreport.com/wp-content/uploads/2011/07/Markakis_Nick_Career_HR_scatter.png

Two factors stand out:

1) Nick hit 3 opposite field HRs in ~1950 PAs from 2006-2008. Since then he has hit 3 in about 1950 PAs.

2) A lot of the opposite field HRs he hit in 2007 were “just enough” hrs. Meaning their “True Distance” was a lot less than his same field HRs. 2007 was also the year he posted a career high in HR rate and HRs.

For those interested the full piece can be found here: http://baltimoresportsreport.com/os-by-the-numbers-nick-markakis-power-outage-17269.html

Thanks to any and all for feedback.