Buy tomorrow, sell yesterday

The mantra of “buy low, sell high” (BLSH) is as sacrosanct an axiom as there exists in the fantasy baseball universe. However, I think that a number of people adhere to its orthodoxy either improperly, or for the wrong reasons. So, today, I’d like to unpack the true meaning of this adage elevated to veritable scripture status.

The first thing that’s important to realize about BLSH is that it inherently assumes that prospective trading partners don’t have as good a handle on a player’s true value as you do. Fundamentally, BLSH is partly about perception, specifically that the market value of a player is lower than his actual value when he’s slumping and higher when he’s sailing. For that be true, there must be a gap between a player’s market value and actual value beyond the expected margin of variance that exists for all players.

If, by some miracle of forecasting, the exact value of all players going forward was known, BLSH could never be implemented. Therefore, the savvier your league, the less applicable BLSH is. A fellow owner in my leagues wouldn’t have any problem at all moving Carl Crawford to my team, while there is no conceivable way any owner would be able to move Willie Bloomquist to me, no matter what the respective stats say.

To savvy owners, the values of Carl Crawford and Willie Bloomquist are known, at least approximately so. In any case, no 100-at-bat sample is going to change those values; only new information will. Is Crawford actually hurt? Does Boston foresee his demotion in the order as permanents? In the absence of this kind of news, it’s actually rather difficult to sell high or buy low on known commodities. So, it doesn’t hurt anybody if you want to try to dangle Bloomquist and see if another owner will bite, but don’t assume that if you are considering trading Crawford or Jayson Werth you’re going to have to take a loss on the deal.

In my mind, BLSH is really a statistical mantra. The reason I don’t want to sell Crawford now isn’t because the market value for him is too low, but because I have faith in his talent and realize that the distribution of a player’s production is not uniform. Things tend to even out, or at least bear that way, over the course of a 162-game season.

When projecting a player’s seasonal output, we aren’t really saying anything about how those totals will be reached. Sure, some players have production patterns that relate to the calendar, but I tend to chalk more of that up as noise than signal. The point is that Crawford is going to have many bad games throughout the season, while also having a lot of good ones. We know nothing as to when those games will come. If those games happen to come in close proximity to one another at the beginning of the season, and I trade Crawford, I’m basically accepting a disproportionate amount of his bad games and trading a disproportionate amount of his good ones. And, that is the point of adhering to BLSH.

Think of a player’s season as your favorite musical group’s discography playing on random. You wouldn’t want to switch to playing another group’s collection simply because the first few cuts were weak. That just means your favorite jams are yet to come, and more likely to be packed tightly going forward.

BLSH really applies only to quantities about which there is significant unknown—rookies, unproven relievers thrust into closer roles, players doing things well beyond reasonable expectation, etc. In theory, Crawford’s value shouldn’t be at a low right now. It should actually be higher than it was at the beginning of the season, if we assume he is healthy and hasn’t substantially lost skill. If you have faith that the “numbers will be there” at the end of the season, then it just means that 95 percent of them will be packed into the 90 percent of the season’s remaining games. Buying low isn’t just about making a profit off a fellow owner’s emotional state, but about having faith in the ultimate symmetry and justice of baseball: that for every slump, there will be an equal and opposite hot streak.


16 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Tom
12 years ago

Whoa! Take it easy there with the Gambler’s Fallacy. There is no reason to believe that if Crawford had started the season on a hot streak his numbers from this point forward would be any different than they will be after his current cold stretch. I did, however, enjoy your random playlist comparison. If I were to throw on Vitalogy and Stupidmop happened to come on first I’d be tempted to change discs even though I’d be missing out on some greatness to come.

Mitch
12 years ago

Some serious non-THT thought going on here. First of all, the album analogy only makes sense if, like a well-liked album, you already know how many hits are on it. But we do not know what Crawford’s numbers will be at the end of the year. As a Crawford owner I’m quite skeptical that his numbers will align with his career norms at the end of this year.

Admittedly I’m ignorant of that basketball study but it doesn’t pass the sniff test. If a player was increasingly more likely to miss a shot as he got hotter, how would there ever be a hot streak? Is an 0-10 shooter less likely to make his next shot than an 0-15 shooter, who is less likely to make his next than an 0-20 shooter? I know the human element is at play here but this just can’t be right. Right?

Brad Johnson
12 years ago

The study was done by Tversky and Gilovich. Here’s an anecdote from it. Andrew Toney made 46% of this shots for the 76ers (82-83 season). When he made 3 shots in a row, his shooting percentage dropped to 34%. When he missed 3 in a row, it rose to 52%. These results were witness almost uniformly in all available data.

Clearly it’s an old study, I would like to see a newer updated version of it. The basic conclusion is that it’s no different than a random process like flipping a coin, but our brains are overeager to attach meaning to meaningless static.

Tom
12 years ago

What I remember from a study about “the hot hand” in basketball is that shooters who have made several baskets in a row tend to start forcing shots which are commonly referred to as heat checks. More often than not the defense locks down on the so-called hot hand which causes ill-advised, low percentage shots. A streaky shooter can’t continue their success unless the defense is oblivious or incapable of defending them. I don’t think we can make the same comparison with pitchers vs batters quite as easily. And again, I’d like to reiterate the flawed logic in saying that Crawford will outperform his rest of season projections simply due to the fact that he is off to a poor start.

MangoLiger
12 years ago

If a basketball player is on a cold shooting streak, the defense might begin to cheat away from him, and so he begins to receive more favorable shooting opportunities. Conversely, when a player is “hot” the defense may play tighter or even double team him, and so he receives less favorable shooting opportunities.

Thus, the defender’s irrational expectations could be what generates the statistical result. I suspect that free-throw shooting would be a more “pure” data source, and this effect would disappear into the noise.

Brad Johnson
12 years ago

“More often than not the defense locks down on the so-called hot hand which causes ill-advised, low percentage shots. A streaky shooter can’t continue their success unless the defense is oblivious or incapable of defending them.”

This is correct, the shooter, his teammates, and his coaches become overconfident in his ability or streak and force the issue. The salient point here is why they do this and the answer has to do with how the brain functions. Basically it attempts to enforce a pattern on a random occurrence. I don’t think it’s unreasonable to think that hitting is completely dissimilar. Doesn’t The Book demonstrate that streaks aren’t predictive (clearly I need to re-read The Book if I have to make that a question)?

varmintito
12 years ago

Sorry, but a slow start should not increase a player’s value, no way, no how.  It’s not like there’s a fixed, pre-ordained number of steals, homers, hits, etc. that a player will accumulate in a year, and the less that’s come so far, the richer the return going forward.  In other words, it is nothing at all like listening to your favorite band’s discography in random order and having all the duds packed into the first 8 cuts you hear.

At best, it’s all random noise, and the player is just as good as you thought he was.  Going forward, he’ll do over the remaining 90% of the season about 90% of what you thought he would do over the course of a whole season.

On the other hand, maybe there’s a reason the production is down.  Maybe it’s injury.  Maybe it’s decline.  Maybe it’s off-field issues.  I had the misfortune to pay $40 for Carl Crawford three years ago.  An established roto star entering his age 27 season.  He started slow.  I kept the faith (I wasn’t going to sell the cornerstone of my team for half value).  His average lingered in the 260s, he wasn’t attempting many steals.  I almost made a trade but he had a 2-week stretch of production and I decided to wait it out.  Finally, I traded him in July for somebody who probably went for less than $20 at auction (I forget who it was), but it was too late.

In short, Crawford is probably the same guy he was 3 weeks ago, but if anything, the severe underperformance to date would make me discount his price to account for the non-negligible possibility that there is a reason for his numbers to date beyond fun with random end points.

Brad Johnson
12 years ago

” If you have faith that the “numbers will be there” at the end of the season, then it just means that 95 percent of them will be packed into the 90 percent of the season’s remaining games.”

This doesn’t seem like a statistically sound philosophy. Shouldn’t we be expecting Crawford’s RoS production to basically be 99.5% of his preseason projection on a pro-rated and lineup adjusted basis? In other words, if 90% of the games remain, shouldn’t we expect 90% of his preseason projection going forward (again adjusting for lineup)?

Andrew
12 years ago

“If you have faith that the “numbers will be there” at the end of the season, then it just means that 95 percent of them will be packed into the 90 percent of the season’s remaining games.”

Actually, no. You shouldn’t expect a player’s production to overcompensate for what has already happened. The past is the past. Everyday is independent. You should only expect a player to perform to his True Talent going forward.

Let’s say Adam Dunn is a True Talent .260 hitter. If he’s hitting .300 at the All-Star Break, it would be incorrect to expect him to hit .220 after the break. No, you should expect him to hit .260 because that’s his True Talent.

Brad Johnson
12 years ago

I know I basically just made the same argument, but there is some reason to believe that if Dunn hits .300 for 20 games, he will have a another streak of 20 games where he hits .220. Sometimes players get hot and stay hot or vice versa, but generally they bounce around all season long.

Neuroeconomists have found that when basketball players have the “hot hand” they’re more likely to miss a shot relative to that player’s shooting percentage. Conversely, when a shooter is ice cold, his shooting percentage sky rockets in his next shot.

The parallel to baseball is a little murky since a shot is simply throwing at a static target subject to defenders while hitting a baseball is literally a 2 person battle with physics. Nevertheless, it seems reasonable to hypothesize that a hot hitter might be less careful at the plate (and pitchers would probably try harder too) while a cold hitter might be more focused. Of course, we also know that being cold can lead to overthinking which is a sure way not to hit a baseball while being hot could give a player the confidence to NOT think, extending the hot streak in the process.

Basically, I think we’ve got the theory all wrong.

Pat
12 years ago

I think this one of the better debates in fantasy baseball and I think there is no correct answer but to say Crawford’s value is higher today than 3 weeks ago is absurd.

I understand where you are coming from because early season starts seem to be overanalyzed but when a player starts the year off bad you have to discount his value somewhat because of the chance that he has a down year/something is wrong. Think david Wright 09.

These are the possible outcomes ( I don’t think I missed any):
1.Down year/decline(he has alot of miles on him)
2.Just a random cold streak that happened in the beginning of the year and he will “make up” stats
3.Having a bad strech and will continue as projected

He could also be hiding an injury but from our perspective we can’t know and this is including in part 1.

Depending upon the player in the 3 options above I will determine how much, if at all I lower a player’s value.

In Crawford’s case I would say 25% chance down year decline and 50% he will ‘make up” and 25% he will go on as projected.

jeffrey gross
12 years ago

I’ve recently come to love buying high. You can get a guy like Bautista if you believe in him at a discount because the owner wants to sell high. It’s an unexploited market inefficiency in and of itself.

Brad Johnson
12 years ago

Buying high, aka buying at the peak of the player’s perceived value, aka making a fair offer is by definition not a market inefficiency. It’s just a normal trade.

Blair Wendell
12 years ago

I won’t repeat Brad’s excellent rebutal to the inane premise of this article, but I will say this…

I’m stunned that you think getting 5 Tails in a row means your are “owed” more Heads in the back half.  That level of thinking is WAY below what I would expect from this site.

Brian Steinhauer
12 years ago

I’m surprised whoever edits this website even allowed this to be posted. The Gambler’s Fallacy is the kind of stone age thinking you guys are supposed to be refuting.

Mark F
12 years ago

Tell me again why it is called a batting average?
Are we to assume that Crawford was going to gather 3 hits over every two games (10 at bats) for the entire season?

0.337
0.297
0.296
0.301
0.245
0.360

These are Crawfrod’s batting averages for each month of last year.  A guy starts off hitting .337 and yet finsihes at .307 for the year.  If we scramble the order of these numbers, he still hits .307 for the year.

He is a .258 career hitter in Fenway and this fact was hardly ever mentioned since he signed with Boston.