Give Me The Rock

this is fantasy basketball 

The Top 100 Fantasy Players – A Visual Representation

Author Icon for Patrick

Categorized as: Author: Patrick, Fantasy Basketball, Fantasy Basketball Strategy
Posted on: June 15th, 2008

Fantasy basketball is all about lists: the top 20 point guards, the top 10 rookies, the 99 reasons why I hate Bonzi Wells. The game is driven by stats and these stats give us ample opportunity to calculate, sort, rank, and evaluate players in pretty much any way that we can think of, such as with the GMTR player rater. But instead of giving you another list, what if there was a way to visually display players to see how they relate to one another?

I’m sure you’ll be ecstatic to find out there is such a way. It’s called Multidimensional Scaling, but I promise that’s the last time I’ll mention those two words. Basically, the procedure takes information for each player (in this case, the 9 most commonly used statistical categories in rotisserie leagues) and places each player on a two-dimensional map* so that the more dissimilar a pair of players, the further away they are from each other on the map. Got all that? Similar players = close together. Dissimilar players = far apart. And as the distance grows, the more and more dissimilar players become to each other.

Here is the perceptual map of the top 100 fantasy players for 07-08 (Click on the picture to see a version that can actually be read)

Perceptual Map - Fantasy Basketball

First off, does this map pass the smell test? Does it make sense? Well, on one side of the map we have Marcus Camby. On the other side we have Steve Nash. They seem like pretty different players. So far, so good.

Second, this map takes 9 pieces of information on each player and condenses into two dimensions. However, it leaves it up to us to figure out what these dimensions actually are. The first, horizontal, dimension is the easy one. It clearly goes from big men (or guys who play like it) on one side to little men on the other, with a bunch of forwards in the middle. The second, vertical, dimension is a little tougher. I have a guess myself, but I’d like to get your opinion. What’s your take on the vertical dimension?

A couple of other interesting things to be gleamed from the map:

* The map doesn’t necessarily have to be in two dimensions, it can be in 1, 3, 25, it’s just the easiest to interpret using two.

Other Stuff Like This:

  • Henry

    Nice. definitely bookmarking it for next year.

  • http://www.givemetherock.com Nels

    The vertical axis is clearly the size of their manhood. The only outlier I see is Andrew Bogut. Who knew?

  • http://thebuzzerbeater.wordpress.com/ Stavros

    Good job!

    In the end it all comes down to the categories each league uses. For example Howard is a great asset but you’ll be have a serious disadvantage in FT% and maybe TOs.

  • http://www.BallHype.com/profile/JWJeff/ Jeff W

    Yes, definitely a bookmark-worthy map. Kudos, Patrick.

    Here’s a fun exercise: Looking for little guys who are most like big guys (Anthony Carter?!) and bigs who are littlest (Sheed?).

  • http://www.givemetherock.com Nels

    Wait wait wait wait… So, you’re saying… if I pick all the guys in the middle of the chart, I’ll have the perfect Midball team?????????????????? ??? ???

    !!!!!!!

  • http://www.givemetherock.com Patrick

    Mike Dunleavy, Josh Howard, Rudy Gay, Brad Miller? All that is missing from that ultimate mid-ball team is KVH at the 4.

    And thanks guys for all the complements. It was a fun little exercise and the fact that the results make a little bit of sense is gravy. Seeing Brad Miller in the middle of the map makes me smile.

    I may try it again including each guy’s rank from this past season. That should make it a more useful drafting tool than it currently is.

  • dyeyk

    i’m gonna take a crack at what the vertical axis means.

    well looking at it, it seems it has something to do with fg, ft and the amount that they take.

    at the top are the high volume guys, who will really affect either your fg or ft due to the amount that these guys take

    at the bottom are the low volume but efficient guys (that’s why they still crack top 100).

  • http://www.givemetherock.com Patrick

    dyeyk – that’s pretty much my thought as well. It looks like the second dimension has something to do with percentages/shot attempts.

    Also, and I never mentioned this in the post, but in MDS the vertical scale doesn’t necessarily have to be perfectly vertical. It can have a slope in either direction depending on the best fit model. If you draw a line from Dwight Howard to Peja, it looks like the vertical dimension could also incorporate threes as well. I could run a regression to test that theory, but I’m lazy at the moment.

blog comments powered by Disqus