The Fun in Making Predictions

Over the last couple of weeks, I’ve written two pretty similar pieces on third basemen. One of which has made me look like a genius — the other has made me look like a total idiot. That’s one of the main issues with making predictions here on RotoGraphs. You try to come to the best conclusions based on stats, player history, etc. — but sometimes you still end up with egg on your face. Hindsight being what it is, I would like to look back at my methods in each instance and try to determine where I went wrong and what aspects of the process I had correct. I think this exercise will help the readers understand my methods and will let me know the areas where I can improve in the future.

Aramis Ramirez is ruining my life. Ok, he’s not really ruining my life, but since writing my Bearish on Aramis Ramirez article he’s been on absolute fire. There were a few issues I ran into when researching this piece. First off, this was part of a series we were running in which two writers took opposite sides. Since I took the negative side of Ramirez, I had to find any and all potential holes in his game. Problem was, there wasn’t all that much in his stat profile — besides his complete lack of power — that convinced me he was completely ready for a downfall.

Once I looked at his age and recent injury history, I knew I could create an argument against Ramirez. By the time I had finished the article, I had convinced myself that I had done it. Ramirez is getting old, he’s going to get hurt — no need to worry. Since I wrote the article Ramirez has hit .321/.345/.641 with seven home runs.

My main problem here was ignoring Ramirez’s current stat line. There was little reason to be worried about his performance this season with the exception of his power — which I should have expected to bounce back. I relied too much on player history and other factors when I wrote this piece and it came back to bite me.

On the other hand, it looks like I may have nailed my recent Mark Reynolds piece. Since writing that article, Reynolds has hit .303/.467/.697 with four home runs in just eleven games. Not the largest sample, but Reynolds has really been on fire the entire month of June.

The interesting thing about this piece was that I did the exact opposite of the Ramirez article. Instead of focusing on the Reynolds’ history, and the stigma that comes with owning him — I focused on the stats. The stats told me there was nothing to be worried about with Reynolds. Actually, they seemed to indicate that Reynolds had improved quite a bit this season, and could be on his way to a nice second half.

At the same time, Reynolds does come with a stigma of sorts. He’s burned many owners in the past, and some owners quiver just hearing his name. Hell, I own Reynolds this season — I’ve lived through the frustrations. In this case, however, you have to ignore the that baggage and focus on the stats. Thankfully, Reynolds’ numbers were so strong that I couldn’t ignore them. Even though it was tough to endorse him, it looks like this was the right choice.

Two separate articles, two completely different results. Sometimes, even using the same predictive methods can lead to laughably conclusions. It just goes to show that making fantasy predictions will always be an incomplete science. But by learning from our mistakes, and altering our methods, we can always try to get better. I may not always come to the proper conclusion, but I will always look to improve my methods.




Print This Post

Chris is a blogger for CBSSports.com. He has also contributed to Sports on Earth, the 2013 Hard Ball Times Baseball Annual, ESPN, FanGraphs and RotoGraphs. He tries to be funny on twitter @Chris_Cwik.


8 Responses to “The Fun in Making Predictions”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Jim says:

    Seems like the problem with the Ramirez article isn’t that you did the same thing as the Reynolds article and happened to be wrong–it’s that rather than looking at the player, history, stats, etc., and trying to reach a conclusion based on that evidence, as you did with Reynolds, you already knew what conclusion you you wanted, and started looking for evidence to support it while downplaying anything that would lead to another conclusion. You say yourself that you were having trouble making the case because there wasn’t much in the stat profile to support it, but had to make the negative case anyway because that was the assignment. Which is kind of a problem with the bullish/bearish series as a whole.

    Vote -1 Vote +1

    • Mr. Thell says:

      Selection bias and confirmation bias.

      Vote -1 Vote +1

    • David says:

      The series isn’t a problem if you read both sides of the argument. Because both writers are doing their due diligence to elucidate the factors that would contribute to a rise or fall in the player’s performance, the reader is left a wealth of information to make his own conclusions. The process reminds me of our justice system with attorneys on both sides making opposite arguments from the same pool of facts, while the jurors weigh the evidence that seems reasonable and relevant. Relevant evidence can often be omitted from one writer’s work on a particular player if it doesn’t corroborate his thesis; Bearish/Bullish helps give the reader a more complete overall picture of a player.

      Vote -1 Vote +1

  2. justin says:

    2 more HR from Reynolds last night…what a turn around!

    Vote -1 Vote +1

  3. OaktownSteve says:

    You were neither right nor wrong in either case. Just lucky and unlucky. Nobody knows nothin’ as they say. Baseball is immesely complex and fundamentally unpredictable. What you are doing is constructing the most plausible narrative possible using an amalgamation of quantitative and qualitative evidence. These narratives give the illusion of predictive power because they appeal to our reason as grounded in the notion of cause and effect (see here for some further thoughts along these lines (http://www.edge.org/conversation/the-argumentative-theory). But all you are really doing is creating fictions that can be seen as right or wrong only afterward.

    Vote -1 Vote +1

  4. Matt says:

    I don’t think your Aramis article was even really that off base. Your article was predicated on three main points: Lack of power, age, and injury concerns. Just because he’s heated up doesn’t mean that the second two aren’t just as important as they were two weeks ago. He is still getting older, meaning he will be prone to more prolonged slumps, and he is very much an injury risk, something his owners have to be aware of whether he’s raking or not. And power comes and power goes.

    Vote -1 Vote +1

  5. Joe says:

    Aramis was having a very characteristic slow start for him. It was a bit exaggerated this year. That was mostly caused by intense wind blowing in on so many games. Also lots of drizzly days.

    The recent injury history angle is good though, maybe he will miss time later, but I hope not.

    Vote -1 Vote +1

  6. ScottyM says:

    Please write an Aaron Hill article. K thx, bye.

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *