The Hedge and The Stand

A recent piece by Baseball HQ’s Ron Shandler brought to light something I have been thinking a lot about this spring. Not so ironically, I have done so directly because of a clever exercise Ron included on this year’s First Pitch Forum series. Ron suggested one way to orchestrate a projection is to come up with a set of possible outcomes and attach a probability each will occur. The final projection is then a weighted average of the varying possibilities. We did this using Ian Snell and Andruw Jones. Each had three possible scenarios – a recent solid season, a recent poor season and something in between.

As I expected (your mileage may vary), the final result in all 7 cities for both players can best be described as a hedge. The vast majority of attendees voted for the middle projection with some talking a stand good or bad. The good and bad basically cancelled each other out so the final results were pretty similar, with some variance because the good and bad were not exactly equal in each venue.
Here is what I have been thinking about. Let us say I fall on either extreme. What do you want, the extreme to be the published projection? A weighted average where the extreme is the majority of the calculation, but there is enough there to hedge the final number a bit closer to the middle? In a perfect world, my personal choice would be to take a stand, not hedge and publish what I think will happen, and then verbalize any hedge in an associated profile or analysis piece. But I am not naïve, I realize some customers utilize solely the projections and really do not incorporate the corresponding supporting material. So the projection has to be as useful as possible, as a standalone entity.

So what did we do? I would like to think we took a stand, at least in terms of playing time. Our projection model and philosophy does not lend itself to going out on a limb with a significant increase or decrease in skills. We allow for some subjectivity but rigorously build as much of what others consider personal touches into our foundational basis. Playing time, however, is a good deal subjective, with history serving as a strong guide.

Ron’s piece is entitled Pundits on Parade and is his annual comparison of projections for players chose by his subscribers. Mastersball was included in the group of sites surveyed.

Brandon Wood was one of the subjects, and we came in with a lowly projection of .237 with 4 HR in 93 AB. The next lowest total of at bats was 247 with 409 leading the pack. Ron’s comment was the following:

Perhaps the most fascinating subject on this list, it's almost all about playing time. And clearly, the most gutsy projection is from Mastersball's Todd Zola. That says to us, "yes, there's a playing time crunch in the Angels infield, but it's going to take more than a .237 BA to squeeze into it."


Well, that was EXACTLY my thought process. Might a hedge have been better, to at least get him on the radar, as he does have some potential, especially when it comes to power? Perhaps, but I kept looking at that .237. In order to project more playing time, I would need to justify an improved level of performance, but I could just not do so, even with a strong spring. I just didn’t see it, so I left the average at .237 and thus concluded that barring an injury to an infielder, and perhaps two as Erick Aybar and Robb Quinlan are ahead in Angels’ pecking order, Wood’s performance is not going to warrant regular playing time. So I took a stand.

Another interesting comparison was David Purcey. We have him sporting a rather ugly 4.91/1.40 line, but over a relatively high innings total of 165. Ron’s comment was

Mastersball gives a lot of innings to a near-5.00 ERA

Yes, we do. But when assigning innings to the Toronto staff, I asked myself if not Purcey, then who? With McGowan and Marcum on the shelf, and no confidence that Matt Clement would complete his comeback, we needed to account for someone eating those innings. So Purcey was projected for what I see as a reasonable total, think of it as 33 starts with an average of 5 innings per outing. Perhaps not as strong as a stand as with Wood, but an outlying opinion nonetheless.

As a site, we are admittedly growing when it comes to projections. I am personally most comfortable with taking a stand, presenting what I honestly feel will occur, even if it means a whole lot of site disciples perhaps miss a Brandon Wood breakout or maybe suffer through another middling campaign from Delmon Young. That said, the other thing I have been thinking about a lot this spring is how frustrating it is to work so diligently on the objective projection base, flavor it with some justifiable subjective adjustment, then throw darts at a playing time board. But we do have some ideas in mind to help mitigate this, at least partially anyway.