THE BOOK--Playing The Percentages In Baseball

Interesting.

After roughly 30 votes, the medians are 70% and 110% of league average.

That means you guys think his mean forecast is at around 90% of league average, which is fairly bleak for someone of his toolset.

Well, but I think that’s a pretty reasonable to good projection, given that a) the sample size is so small, so we expect a lot of regression, and b) he’s coming off injury. I also expect that a large bit is just that there’s a lot of uncertainty in what we can expect from him because of these two factors, and the larger the uncertainty, the more it looks like you’re being pulled to average. Probably if you somehow asked for a projection from everybody, allowing them to put in any continuous number, then took the median, we’d expect him to do better than this 90%.
But also, thinking about it more, I don’t think it’s really fair to average .7 and 1.1 to .9 arithmetically like that. ‘cause let’s say he’s twice as good as the league; that’s .5. If he’s half as good, that’s 2. You can never get better than 0, but theoretically, there is no maximum. So really, seems like a geometric mean would be more accurate, and .77 is not so bad.

Strasburg will either be awesome or injured. There ain’t no third direction.

He had TJ surgery, not a shoulder injury.

Seriously are people really projecting him to lose significant velocity, lose movement on his curveball, and lose control on his changeup?

He has 3 plus pitches and good control. That right there is impressive. Throw in that the fastball is 96+ ... and basically you have Verlander with his (Verlander’s) improved control ... only SS K’s more batters.

How’s Verlander doing this year?

There just isn;t any way for a guy with Strasburg’s “stuff” to be that close to league average. He’d have to have the control of Farnsworth. He doesn’t.

Again, great or injured. Given his age and potential, they won’t be pitching him when he’s hurt or needs rest.

After a couple injuries, then you start moving him toward average is he loses his stuff ... but after one injury, especially TJ surgery, there shouldn’t be this much “uncertainty” or concern.

I could be wrong, but aren’t there a host of pitchers, or at least a few, that come back from TJ surgery and lose a significant amount of their stuff and/or command, and/or velocity? 

The idea that he is going to be awesome or injured (and I have no idea what that means - the injured part) seems ridiculous to me.

It reminds me of a throwaway comment that my friend made the other day about the movie Cowboys and Aliens, before we went to see it.  He said, “This movie is going to be either awesome or great.” That was a ridiculous comment on so many levels.  We saw the movie, and I thought it was, “OK” - exactly the opposite of what he said.  Of course that was my opinion.  If we polled 100 people, we probably would get the entire gamut of opinions from great to terrible.  I have no idea what that distribution would be centered on, although we can guess from the RT viewer ratings I guess.

Sure, we can actually measure how great or terrible SS will be.  Still, I expect the uncertainty is a pretty smooth continuum (from great to terrible).  The idea that any pitcher’s uncertainty distribution can be extremely bi-modal (great or injured), again, seems silly to me.  Again, I don’t know what “injured” means. He was already injured and had surgery to repair that injury.

Now, if by “injured,” CC means everything on the continuum from “not great” to “terrible” then I guess I can’t disagree with him, although that characterization seems obvious and tautological - “He is going to be great like he was, or not great because he is not fully healed and recovered?” OK....

For the poll about his downside, I might have voted better than league average, but that wasn’t an option.  That might have an effect for his 25th percentile forecast.  The 75th percentile poll does seem wrong to me though.  I was expecting .6 to be the median for that one (it’s also what I voted).

Maybe some people are thinking about Liriano and what has happened after he had TJ surgery.  Even after finally looking like he did before the surgery, Liriano has struggled this season.  But Strasburg does have better stuff than him and should have a better outcome after the surgery.

And to echo what CircleChange said, this isn’t like Kerry Wood or Mark Prior with their shoulder injuries that were career changing.

Remember that we are looking at observations, not talent.  So, Felix, Lincecum, Verlander, etc, will all have seasons where they “appear” to struggle, but could end being close to league average just by luck.

***

Bob/5: yes, excellent point about the downside being better than league average.  After I posted the poll, I did a test to figure out what it “should” be.  And, I should have had a downside of 0.80 or worse, and 0.90 or worse.  Nonetheless, that only 25% of the voters after 80 votes selected the most optimistic of the downside forecast tells me that that still would have been overkill (in this case).

All I’m interested in is the median.

For the record, I selected 0.60 and (would have selected) 0.90.

***

As for the geometric mean: If he’s at 25% for 0.70 and better, and 25% at 1.10 and worse, then that means that he’s at 50% at 0.70 to 1.10. 

Why would you argue that the midpoint of that0.70 to 1.10 would be less than 0.90?

Now, what you can argue is to convert the runs scale to a wOBA scale, and then make that additive (which you can).  If that is the argument, then that is an excellent argument.

But, no, geometric mean is not the correct thing to do.

This is the test I did: I took all pitchers aged 23-27, born 1952-1979, and found the 10 who performed the best.  You have Pedro, Clemens, Rijo, Santana, Maddux, Stieb, etc.

Their runs allowed index averaged 0.72 (i.e., runs allowed at 72% of league average).

Then I asked: what was their RA index over each year from age 28-31.  25% of their RA index was under 0.60, and 25% was above 0.84.

So, you get that kind of wide range simply because of the vagaries of sample size, the sequencing of events, and the impact of the fielders.

Even if you believe Strasburg is better than the 0.72 level, you still have to expect some wide range in performance that has nothing at all to do with his talent.

The voters here definitely have the right idea that the gap between the 75th and 25th percentile should be about 0.40 in runs index (higher than our test group, mostly because of injury and lack of sample data).  I find the voters to be fairly pessimistic in that they have 0.70 and 1.10 as those threshholds, rather than 0.60 and 1.00.

Anyway, my main point was to show the range in outcomes, and the voters passed the test.  Good job!

With the 25th percentile poll, I wasn’t just trying to say that some people who voted league average would have voted better than average.  I was also thinking that some people who voted 1.1 or 1.2 would also have voted better.  I think most readers might defer to you because you are quite knowledgeable, and the boundaries that you set might seem like the absolute highest/lowest they realistically should vote for. And if you’re making the poll, they’re going to expect that the “real” answer shouldn’t be at the extremes.

With that said, I do find it a little surprising that people are as pessimistic about his downside.  And your point about their being a range of outcomes is important.  In fact, that’s one of the better insights into the game that I’ve gotten out of reading this blog.  That one season isn’t a large enough sample that random variation can’t a impact on a player’s statistics.

Yes, I do worry that making the “right” answer as the midpoint of choices biases the results.

Then again, the actual results in both polls were not in the middle.

I wonder how much opinions will change based on the reports back from his first rehab start in A ball this Sunday.

Don’t worry… I’ll do another poll in the offseason.

Wow, results are far worse than I expected. This kid is a once in a generation talent. I have to assume folks think TJ is scarier than it really is.... or it’s another case of too many Mariners fans. Anyone who voted .8 or .9 can’t really understand the question or is biased.

I would think Francisco Liriano--whose 2006 was every bit as dominant as Strasburg’s 2010--has already definitively shown that a pitcher can come back from TJ and remain healthy but ordinary.