Spaghetti Baseball
By John Sickels
(This is an edited and revised version of an article originally published in March of 2005)
Saturday, June 13, 1976. I was eight years old. My parents decided that it was time for me to spend some time away from home for the first time, so it was off to summer camp at the Episcopal Center for Camps and Conferences north of Des Moines.
The first thing I remember that Saturday morning was walking out of the house into what felt like a blast furnace: it was hot, humid, windy.
Mid-June is the heart of tornado season in central Iowa, and on the afternoon of June 13th the atmosphere would explode.
On the way to camp, we drove through a hellacious thunderstorm including large hail and fierce winds. After awhile, we broke out into a clear area of the storm. And off in the distance, we saw this:
Tornado at Jordan, Iowa, June 13, 1976 (Iowa State University photo)
It was an F5 tornado, and one of the most powerful ever recorded in modern times. Dr. Theodore Fujita, renowned tornado researcher and developer of the famous F-scale for tornado damage, once remarked that this particular tornado was the strongest he had ever studied. The tornado hit a small town called Jordan, annihilating it. Remarkably, no one was killed. The tornado stayed in rural areas, which was most fortunate. A shift of just a few miles would have brought this monster through the heart of Ames, Iowa.
Witnessing this thing had quite an impact on my impressionable young mind. I decided that I wanted to be a "weatherman," a meteorologist. I read everything I could find about severe weather, thunderstorms, tornadoes. It was one of my biggest passions as a child and teenager, along with baseball.
Unfortunately, once I got into high school, I discovered that I was not very good at advanced math. And I was especially bad at physics. I could understand the general theory behind everything, but when it came down to pencil and paper and formulas and a scientific calculator, my mind would blank. I eventually came to the realization that I wasn't cut out to be a real meteorologist, so it just became another hobby.
The internet is a boon to severe weather nuts like myself. When I got on line in 1996, I discovered a wealth of information available, things I could only have dreamed of having access to previously: model outputs, mesoscale discussions, convective forecasts, raw severe weather data, etc. No longer did I have to rely on local TV to let me know when a tornado watch was up: I could read the watch prediction itself right off the net, including the detailed reasons why the forecasters felt the watch was needed.
The second thing I do every morning during the spring and summer, after checking the baseball scores, is to log onto the Storm Prediction Center website and check out the risks for severe weather in Kansas. Over the last few years, I've learned more and more about the various computer models and forecasting tools.
Now, understand, I wouldn't even call myself an "amateur meteorologist." I understand this stuff a lot more than I did five or ten years ago, but when the discussions start getting too technical my eyes still bleed. I can look at the SPC Mesoscale Analysis page and figure out where the big danger spots are. . .I can tell you to watch for the spots with 4000 CAPE, but I'm still can't make sense of a sounding or a hodograph on my own.
OK, enough jargon dropping. So what does any of this have to do with baseball?
Weather is a natural system. It is somewhat predictable, if you have enough data. Baseball players (and human beings in general) are also natural systems, and with enough data they are also somewhat predictable.
Meteorologists use computer models when making their forecasts. Each model uses a different set of assumptions in taking a data set and projecting it out into the future. If you poke around the internet, you will find charts like this one:
This is a chart of "model output ensembles," commonly called a "spaghetti diagram." Each line on the chart represents the output of a different computer model, in this case projecting the flow of the jet stream. A meteorologist making a forecast will consult different models that use different assumptions, to get an idea of the possible outcomes of the current situation.
There is more to it than just that, however. No good meteorologist will rely only on the computer data: there is a place for intuition, instinct, "gut feeling" if you will. I personally believe that what we call "intuition" is often an expression of subconscious pattern recognition on the part of the human mind. No good meterologist will make a forecast based on one computer model alone. She will sift through all of the data and model outputs and make her own judgment.
Baseball player prediction systems like PECOTA or ZIPS operate on the same basic assumption as weather models. They are much less complex, of course, since there are fewer variables to consider. The parallels can only be drawn so far between weather and baseball, but they are there.
Of course, even the best model and the best human forecaster screws up sometimes. "High Risk" severe weather days sometimes result in nothing but a bit of wind and lightning. . .all the known parameters come together, but something just isn't quite right. . .perhaps the wind shear was less than forecast, or a cirrus shield prevented the atmosphere from destabilizing.
In baseball, this is the "can't miss" prospect who misses, sometimes due to injury, and sometimes for no obvious reason at all. Forecasting a player's career is easier once a player gets into professional baseball obviously, and we have minor league data to work with, just as forecasting the weather is easier from one day out than seven. When looking at players in the draft, we are basically making a long-term forecast with a limited set of data.
Prepare to break out your intuition and observational skills for an upcoming draft discussion thread at noon CDT.
Loading comments...