Who was actually good…
Last time, I used ADP data and player values to determine Kyle Lohse was the most under-drafted player of the last five years. Turns out, there are some assumptions in the calculation that could be tweaked, and the result could be a totally different most under-drafted player. Go figure! The methodology was to take the difference between a player’s preseason ADP and his end-of-season rank to determine “undervalued-ness”. This time we’re still going to take the difference, but it’ll be between the square root of his ADP and the square root of his EOS rank.
Why the square rooting? The reason is to give more weight to better players, which square rooting accomplishes.
For reference, here’s the list from last time (that won one lucky man a Razzball T-Shirt):
Now, let’s see what the new Top 10 list looks like using the square root of ADP and Rank:
Well, sh*t. Lohse is like “Can’t dethrone me!”. We’ll have to take more extreme measures. But first, let’s take note of some of the newcomers to the list. The major addition is Max Scherzer, who was propelled to No. 5 off the strength of his 2013 (drafted 78th, finished 6th), his 2012 (drafted 170th, finished 51st), and 2010 seasons (drafted 237th, finished 106th). Using different ways of getting there, Ortiz and Kluber are the other new faces that round out the Top 10.
So what’s more extreme than taking the square root? We could ask these guys or we could take the natural log, your choice. Not! Using natural logs, here’s your new Top 10, America:
Woo! An interesting result! Most dataists do ethically questionable things for one of these!
Posolutely one of the more polarizing preseason players, Clayton Kershaw forces his way to the top. And I like that it’s not off the back of one season, but rather by consistently beating his ADP a little bit each year. Take a look:
|Clayton Kershaw – SP – LAD|
Kershaw’s dominance over the past five years—and almost equally important his health—has challenged us to consider just exactly how high we will draft a pitcher. While I wouldn’t necessarily take away from this table that Kershaw should be taken first overall in drafts, it does illustrate well the skepticism tax we place on pitchers, and how maybe Kershaw should be the lone exception to the rule.
So where did we draft Kershaw in 2015 you ask? His ADP this year was 4.1, which ranks fourth highest behind only Trout, Andrew McCutchen, and Miguel Cabrera If you look at Average Auction Cost, he ranks second at $49, still way behind Trout at $60. However based on my auction experiences, I would have thought Kershaw would have gone for less. Now I’m intrigued about the relationship between AAC and ADP.
The Most Over-Drafted
Like anti-matter to matter, no Top 10 list is complete without its corresponding (and usually more entertaining) Not Top 10 list. Here are your Razzies:
By this methodology, Albert Pujols is the most over-drafted player of the past five years. To be honest though, that designation is a little harsh for him; he’s being punished a little more than he should for being a top 10 pick 4 of the 5 years we’re looking at.
The next three guys on the list though—Hanley, Longoria (sorry Grey), and Mauer—are truly players I feel we remained optimistic about for a few years too long and are more deserving of the ‘over-drafted’ label.
Tim Lincecum get the auspicious title of being the only pitcher to make the list. And unfortunately for him, he’s the only pitcher in the top 40 until you get to none other than Steven Strasburg, who certainly isn’t helping his case with his performance this year.
That’s all I got this week. Hopefully you found this information interesting and even a little useful.
For those who remain interested, I’m still working on the fantasy baseball bot. I’m actually in the middle of the “fun” part, which is coding the algorithm itself that figures out when to bench a player, or add a guy from free agency, etc.
One practical spin-off of the full-season algorithm that I will inevitably build in the process is a single day lineup maximizer. My thought right now is to offer that as a standalone product (maybe charge $1 per lineup maximization if I’m feeling greedy). If you’ve ever signed up for a site and had to ‘click to authorize’ to give an app permissions, what you were doing was generating an access token for that app to use another site’s API (like Facebook’s to get a list of your friends).
Essentially if I figure out how to do it, people could authorize me to use Yahoo’s API on their behalf and the algorithm can do it’s thing on their specific league and team data.