A question for discussion on rankings

Everybody prepares for drafts differently in the offseason. On one extreme there are the people who go all-out, creating models to predict player performance or using some other mathematically involved method to create their rankings. And on the other extreme there are the people who do very little to prepare, at most maybe purchase a magazine and have it open next to them while they draft.

In the middle of those two extremes are the people who spend some time creating personal rankings, probably from looking at last year’s stats—making adjustments based on age, playing time, and luck—and then creating a rough prediction for each player’s stats for the upcoming season. Today, I have a question for those people in the middle about how they come up with their rankings.

Do you think that when making your rankings if you looked solely at the players’ stats without any names attached, your rankings would look different than if you made them as you usually do, with names?

My feeling is that most people would answer yes to the above question. Some people at the beginning of 2009 just had this feeling about Matt Kemp and they knew that he would have a good year. Obviously these people were rewarded for bumping up Kemp in their rankings in this example, but had they felt the same way about Chris Iannetta then it would not have worked out so well. This leads me to my next question:

Do you think it is harmful to allow your instinctive feelings about certain players affect your player rankings?

Some will say “No, that is not such a bad thing” while others will argue vehemently against allowing irrational feelings on certain players take effect. Personally, I suggest that you use player names along with their stats to make rankings, despite knowing they would look different if the names were not attached.

Some people might call allowing a player’s name to affect your opinion of him an irrational bias, but I do not believe it necessarily is. There are many subtleties that are unique to each player’s situation—such as playing time or contract situation—that the numbers do not capture. By associating a player’s name with his situation and then adjusting your projections slightly based on feelings, I do not believe you are hurting your team’s chances of winning by any significant degree. You might even be helping.

So do not feel guilty about sliding James Loney up a few slots in your rankings if you feel he is in for a breakout 2010 campaign. For the most part you should have statistical backing to your rankings, but there is no problem in indulging in a few of those feelings of yours while making them.

Print Friendly
« Previous: And That Happened: LCS Edition
Next: THT Awards season finale, part two »

Comments

  1. Detroit Michael said...

    For pitchers, I’m one of those people in the middle.  I want my subjective impressions to be included in the rankings so then during the draft or auction, I am not tempted to make further subjective adjustments based on who I like.

  2. Paul Singman said...

    Peter D it does not surprise me that you are more “hit or miss” with the young guys since by nature, because of their inexperience in the major league level, they are harder to predict. Generally speaking I would agree with taking proven guys early but every year there are a few Garrett Atkins’s, so ignoring other player’s upsides in fear of those players not developing as expected will ultimately cost you.

    As I tried to show with the Garrett Atkins example even the most proven of players sometimes don’t perform, in most cases because of injury. What you might want to consider doing instead of creating one projection for each player, is come up with three: a best-case scenario, a middle, and a worst case. Weight them by chance of happening based on age, injury history, underlying skills and then average them together. That way you can capture the potential of those players with higher ceilings than others, without drafting directly for the ceiling.

    The trick is weighting each line correctly, and that will take some research.

  3. Peter D said...

    I’m generally one of those guys more in the middle, which works out fine with players who have a few seasons under their belt because I look at their stats season to season over their entire career and how it compares to the last season.  This usually gives me a fairly close projection on what the player will do for the upcoming season.

    However for those young players who have little major league experience I tend to look at their minor league stats.  My projections on these guys are generally hit or miss, while I was proud to say it did lead me to take Mark Reynolds last year, it also had me extremely over-rate guys like Alex Gordon, Jay Bruce, and Chris Davis.  So much so that I’m contemplating taking only proven guys in the first 10 or so rounds of the draft next year.

  4. obsessivegiantscompulsive said...

    I totally agree.

    When I play the Yahoo Fantasy League, I start out with their ranking, then I start removing players that I either don’t care to own or think they had an outlier plus season the year before or think that they are headed for a downslide (particularly for players I think are injury prone).

    That automatically moves up the players I do like, a la what you noted in your article, and it has worked for the most part.

    Also, for the Giants players I particularly want, I jump them up a little bit more, and I ended up with Lincecum, Sandoval, and Wilson this last season (missed out on Matt Cain, unfortunately).

  5. Peter D said...

    Lucky for me I didn’t rate Garrett Atkins very highly since I’m in a league that uses OBP, and his 08 numbers were not very good.  But I get your point. 

    That being said looking at my draft last season, I took Chris Davis over Kevin Youkillis in the 5th round, partially because (I may have read it here) that Youkillis would not repeat his power numbers due to a higher than expected home run to fly ball ratio.  This year, if it comes down a choice similar to this one, I’m going with the player with more experience.

    Of course it will get to a point where the remaining proven guys don’t offer much, then it will be worth the gamble on upside.  Justin Upton for example was high on my list last year, and despite the fact that I didn’t draft him, that pick turned out well for the guy that did. 

    Here are a few examples of proven guys that I would take over upside guys in next years draft

    Raul Ibanez over Carlos Gonzalez
    Nick Swisher over Kyle Blanks
    Shane Victorino over Andrew McCutchen
    Johnny Damon over Julio Borbon or Dexter Fowler

  6. Andrew said...

    Looking only at the numbers excludes tons of extra information.  Ignoring that information puts you at a disadvantage.  So long as the criteria you’re using to evaluate a player is rational, any extra information you can garner is an advantage over only looking at the numbers.

    The best evaluators are able to sift through both the statistical and the nonstatistical information to more accurately gauge a player’s expected production than their opponents.

    Let’s put a hypothetical example out there.

    Player A hit around .250 last year with 15 HRs in around 350 ABs.  The player in question didn’t offer much in the way of speed.  Let’s pretend that the underlying stats say that player A wasn’t particularly lucky or unlucky, and that the numbers accurately reflect the skill level that player A performed at during the previous season.

    Knowing nothing else… how can we deviate from last year’s stats in coming up with projections for next year?  Looking purely at the numbers, we can’t.

    Now let’s add in some more information.  Player A will most likely get more ABs in the coming season, as he is expected to be a full-time player.  This piece of non-statistical information obviously helps player A’s value.  Though not a completely reliable method of projection, we can quite easily extrapolate player A’s stats to how much we expect him to play.

    Now let’s add some more information.  Player A put up the aforementioned numbers in his age 21 season.  This small piece of information might suddenly be the more important than anything else that’s been mentioned.  After all, the only 21-year olds on major league ballclubs are typically the most elite prospects.  If player A is already in the majors, it’s highly likely that he was an elite prospect.  A little more information- namely that player A is a former #1 overall pick and scouts say that he’s got physical tools overflowing out of his ass- gives us reason to believe that he’s got a significantly higher ceiling than what he put up in the previous year.

    If you haven’t figure out by now, player A is Justin Upton.

    Going only by stats, we can’t expect anything other than what the numbers have told us.  If we attach the name Justin Upton to the numbers, though, we suddenly have an incredible wealth of new knowledge.

    Even intuition that we are unable to rationalize shouldn’t be wholly discarded, since intuition is pretty much our subconscious at work.  I wouldn’t too much stock into it, but I wouldn’t discard it wholly, either.

    Even our “CONSCIOUS” thoughts are typically a matter of heuristics that occur before we are even aware of them.  Our brains reach their conclusions before we can even articulate what those conclusions are.

  7. Derek Ambrosino said...

    One thing I don’t understand is when unproven younger players are taken ahead of archetypally similar, high quality veterans.

    Peter said he took Chris Davis in the fifth round. I saw Adam Dunn slip in a number of drafts, and thankfullly I was able to get him late in a number of leagues. A few times, Davis was taken for Dunn. What would be a rationalization for this decision.

    I mean, if all goes right for Davis, he’s Adam Dunn. So, why leave Dunn, the proven commodity, in favor of the guy who might be… Adam Dunn-like?

  8. Andrew said...

    @Derek

    In the Davis/Dunn case, I guess people just expected Davis to be a .285 hitter, while Dunn would hit near his career average (.249 after this year).

    In a more general sense, I guess people prefer shiny new toys to their old ones.

  9. Paul Singman said...

    Yup, Andrew, what you say in your first post is a good example of what I’m talking about in the article. Looking at more than the numbers may introduce some bias, but some of that is “good bias” if that makes sense.

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 day month ye@r *