By now, you've seen various baseball previews scattered throughout the print and electronic media, and almost all of them have a line like this:
"Player A is due for a regression."
Sometimes this comes about because last year's numbers were several notches above the expected level of performances. The statement could come because of age-based concerns, or injury history. Or it could just be a partisan's way of proclaiming that no opposing player could possibly be that good.
This is why I hate doing predictions based on component players. This guy will be worse, this guy will be better, the team finishes second, blah blah blah. I could easily paint a picture where the Indians finish anywhere from first to last, and it would be at least half-way plausable. (In fact, I might just do that later on this month.) That's the fun of previews, and also the drawback. Some people are really good at looking at a roster of players and deciding the season's outcome. Some just like to play it safe. Some like to be contrarian for contrarian's sake. Predictions sell copy and it's mainly for entertainment purposes, but I know some people who freak out if the Indians aren't in the right spot on somebody's pre-season rankings.
It's extremely easy to fall into the trap of just modifying last year's numbers to fit what you think or believe will happen. You have no basis for these numbers; often it's the first thing that pops into your mind. And of course it's real easy to let your rooting interests inflate or deflate predictions because of familiarity or even emnity.
This is why I like prediction systems like PECOTA, because not only does it remove the subjectivity from the process, but it also gives the reader a range of probabilities as to the kind of season a player will have. But any prediction system, whether it be for weather or baseball, isn't going to be perfect; if it was, why watch the games?
The intoxicating mix of continuity and unpredictability is what make professional sports so enticing to hard-core fans. We have a basic idea of what's going to happen during the upcoming season, but there's always going to be an event or a performance that throws the most sophisticated prediction systems for a loop. This by no means invalidates them, but merely tells the fan that anything is possible.
I don't really know where else this rant is heading, but I'd better stop it here before it becomes completely incoherant. But not before saying that my next post will be much better thankfully because of regression to the mean. Or perhaps it will be worse because of an odd brain lesion affecting sentence construction. Or perhaps I've already reached my peak blogging age and in my decline phase. Or...