In a recent column, I made the comment that I feel the fantasy industry as a whole is doing a poor job educating the fantasy public in some areas. I must have tweaked a nerve as I received a rather nasty anonymous e-mail, challenging me to come up with three examples. I say anonymous because the e-mail address did not contain a name, it was not registered in our database and it was unsigned. This is speculation, but it may have come from someone in the industry, which is fine. Instead of replying back to the mystery person, they obviously read the material on the site so I thought I would share five examples publicly.
Before I do, I want to preface this by saying my comment was on the whole. There are obviously some people offering solid advice – a lot of people. But the advice is not getting through to everyone. While some of the blame goes to the fantasy public for being stubborn or lazy (whoops, now I’ve also pissed off the public, too) more blame lies with those charged with disseminating the information, not doing their homework, preaching clichés they learned five years ago from an outdated version of the “So-Called Experts handbook.”
1. The notion that trades are won or lost
Trades are not won or lost. When we receive a question like “Jones for Smith, who wins?”, our response should be an explanation that trades are not about winning and losing the deal – they are about improving your team. The currency used to evaluate the deal should not be the rest-of-season dollar value we offer in Platinum, where they rank on the Yahoo big board or the ESPN Player Rater. The measuring stick should be the intrinsic value to your team. The best trades are when both teams get the help they need. If that means both win, I can accept that in a semantic sense. But the point is, “Jones for Smith, who wins/” is the incorrect way of approaching the process and if the answer is either “Jones” or “Smith”, the advisor is doing a disservice to the advisee, regardless of the names involved. It should be explained that every move involves a balancing move, either a player is replaced or added to the active lineup.
Who does Jones replace? Who replaces Smith? It is the difference in these players that matters, not the raw value of the player in the proverbial vacuum. This is especially true when an uneven number of players is exchanged. The equation is points potential of your roster before the trade versus the potential after, not Jones is projected for $25, Smith is projected for $15, you need to get back a better player than Smith if you trade Jones. If Smith earns your team 5 standings places while Jones does not help you, Smith has more intrinsic value to your team – case closed.
2. Playing streaks and matchups
This is a topic that has come front and center as more fantasy leagues allow daily moves and the emergence of the one-day fantasy pay for play sites. The bottom line is no matter how well we think we know baseball, no matter how many games we have watched, regardless of what our intuition tells us, streaks are usually just clusters of good (or bad) things happening together and historical hitter-pitcher matchups are non-predictive of future matchups.
The Internet is full of “proof”, if you are so inclined, employ your favorites search engine and check it out. The pioneer of this sort of thing is Tom Tango and the work is published online and in books. More recently, Derek Carty has done some simple yet elegant studies on the subject and our friends at Baseball HQ are now also dabbling in this for the benefit of their subscribers.
With respect to streaks, sure, some players are legitimately hot or cold. But more often than not, it is just a series of events falling on the good or bad side of the probability curve. If you flip a coin 500 times, there will be a stretch where there are more heads than tails and vice versa. But once you get to 500 (or whatever), the heads and tails are almost the same. That’s all a streak is, five heads in a row. I will personally look at a player’s strikeout rate when he is in the midst of a streak, since making contact is the entity in the player’s control. If a player is striking out at a depressed rate during a streak, I’ll make the subjective assumption he could be legitimately hot and manage accordingly. If he is fanning at the same or elevated rate, he’s just lucky. Similarly, if a cold player is whiffing more than normal, I consider him truly cold. But if his strikeouts are normal, he’s just unlucky.
Early in the season, Bryan LaHair was as hot as a pistol, or was he? His K-rate was even higher than normal, so while there was some likelihood of “being in the zone”, he was also quite lucky. You don’t strike out that much if you are truly in the zone. His monthly splits are:
In April, he was quite fortunate with BABIP but struck out a ton, in May he was snake bit but made better contact and so far in June, his luck is back but so are the whiifs. LaHair is about a .280 hitter. Batting average can fluctuate 15-20 points just due to randomness and probability and have nothing to do with skills. By season’s end, Lahair will be right where he is supposed to be, .280 +/15 points. Trying to time his streaks is a crapshoot.
The hot commodity today is Trevor Plouffe. Some are saying we should have known because his second half numbers last season were promising. But, so were a lot of others. Second half numbers are non-predictive. Plouffe’s splits are as follows:
His strikeouts have been fairly consistent; so Plouffe is still the same player. He is probably making harder contact now, but there is no evidence this is an increase in skills and is sustainable, it is just the randomness of when it happened to manifest. Picking Plouffe up and dropping a better player is likely going to cost you in the long term and perhaps short term as well as we have no idea when the pixie dust runs out and Plouffe returns to being Plouffe. The instances of an emergence like Jose Bautista are rare, and are usually accompanied by a discernable skill improving. In Bautista's case, it was more walks and fewer whiffs.
The mistake being made by fantasy pundits is recommending that better players are dropped for Plouffe, with the recommendation to “ride the streak”. This is misleading and shows some giving advice are uneducated as to what is really going on.
In general, the fantasy public has a misperception of what a projection is and expects it to be something it isn’t. By definition, a projection is an objective view. It is the most likely scenario based on what has happened in the past, using history as a guide. Will it always come to fruition? No, of course it won’t. The most likely outcome of rolling a pair of dice is seven, but that only occurs 17 percent of the time! But the proper answer to “what will the dice roll be” is seven. Anything else is speculation, and once you add speculation to a projection, by definition it is no longer a true projection.
You’ll see complaints that so-and-so’s projections are too conservative, they never go out on a limb. Well, if they did, they would no longer be projections.
You’ll see someone providing projections pride themselves in not relying on a stupid spreadsheet to do the work. They do each one by hand and may cite a success based on a player having a great September. Well, unless you can show that every player having a similar September gets the same boost and you give that boost to everyone that had a great September, you are using a subjective bias and are no longer providing true projections.
That said, I am not trying to imply that I or anyone else provide the best projections. I may feel I have a better understanding of the process in a global sense but will admit that the execution may not be completely correct. I do the best I can, determining translations, but I am not going to claim they are the best. Maybe my MLE’s are not complete enough. Maybe I should be regressing BABIP differently. I look into this every off-season and attempt to improve methods, but I am sure I am missing something. All I am saying is I am providing what can be defined as a projection, not my opinion of how a player will perform, mainly because my opinion is just that, an opinion, and I feel my job is to provide as much information as possible for you to formulate your own opinion.
Now with that said, I also think those providing projections do not emphasize they are not static, but each player actually has a range of expected performance, which we represent by a single, static line. I think we need to do a better job of sharing that objective range, especially since the projection commonly provided is not always the midpoint of the high and low ends. In fact, it usually is not. This is an area I would personally like to improve and will strive to do just that.