This is my 50th post on this space, which is kind of absurd. What started out as little more than recording my thoughts about a game I just started to play but had a passion for turned into 25,000 monthly page views across three properties, a podcast I co-host with an entertainment industry veteran, and regular access to fantasy guru Matthew Berry.
Yep, just like I planned 8).
Now, I realize I’m hardly curing cancer here, but this has been a fun ride that I’ve discussed previously in pieces and decided for this 50th post close to New Year’s Eve to write it all down in a single place. I don’t know where this FML journey is headed, but this documents where it’s been.
It started when Matthew appeared on the Rich Eisen Show on June 23 and I first learned about Fantasy Movie League. Here’s my dark secret: I don’t go to the movies that often. Maybe 4 times a year, actually. However, I watch a ton off of streaming services from the comfort of my couch and quickly became intrigued by the math around the FML rules.
Creating a Monte Carlo Simulator That Didn’t Work
My second week playing, Week 7 of the Summer Season, I started writing a recursion routine to help me figure out what all the possible legal combinations were, which didn’t help a whole lot to know that there were 250,000 possibilities when you could only choose one. A big fan of Nate Silver’s, my first attempt at writing a predictive model based on professional forecasts was to use the same Monte Carlo-style logic Nate applied to polling numbers when predicting both the 2008 and 2012 Presidential Elections. I decided to do so transparently and this blog, along with my nerdy bin-packing Monte Carlo simulator, was born.
That first attempt didn’t go so well, but I started to get comments from people who found this space on their own or through the little bit of Twitter promotion I did. Those early interactions with people like Sad Robot and Billion Dollar Cineplex really validated that I was onto something and I’m forever grateful for their encouragement. I knew I hadn’t cracked anything significant yet, but I knew I was at least on the right trail.
The FML Comparator is Born
On Friday, July 31 I was on a layover in Las Vegas when I realized that what would be more useful than blindly randomizing variance on individual movies, as the Monte Carlo simulator had done, would be to have a set of sliders where I could manually adjust variance and automatically calculate the impact those adjustments had on what the best combinations were. That weekend, I built what would be known as the FML Comparator.
It was hardly perfect at first. In fact, in an attempt to speed page loading time it used only a few thousand combinations and often missed fairly obvious plays. Despite that, I promoted it on Twitter and got great feedback.
Then I received a follow notification on Twitter from Matthew and this happened:
It’s hard to capture in typed text the giddiness I felt after this exchange. Validation looms large in my life and to have positive feedback like that from not only the founder of the game I grew to enjoy so much but from someone who is the most recognized fantasy sports expert in the country was stunning.
Then, a week and a half later, I was running a sales offsite for the cloud start up company I work for and while sitting at drinks I got this Twitter direct message from Matthew:
At first I thought I had a legal issue on my hands because I was reusing the images the main FML site hosts in the Comparator (the lesson: programmers are lazy). But, no, once I had conversations with multiple FML leaders, they wanted to re-run my columns and encouraged me to explore data analytics more. I had been on a weekly cadence where I’d review the previous week on Mondays and put out a picks column on Wednesdays, but the main FML site already had an excellent review article by Zachary Knight. This was my chance to spend time exploring the data more deeply and transparently in front of a wider audience.
Official Fantasy Movie League Columnist
It’s fair to say that when I became official for Week 1 of the Fall Season not everybody was digging what I was doing with my picks articles. Accusations that I was ruining the game by telling people what to pick, thereby lowering the barrier to play compared to those who did their own spreadsheet gymnastics, were common on the Chatter. I had my defenders, though, that grew into online friendships with folks like Ari, M37, Phil’s Phun Phlicks, Angry Geek, and others. Some would correctly point out that I was rarely making good picks, which is kind of a problem if you’re touting yourself as an expert since being right occasionally is sort of important. Summer Season runner up Sleestak very publicly quit FML, citing me as among the reasons for doing so.
But the data deep dive articles were starting to reveal some nuance about what information mattered and what didn’t. I got some really good ideas from comments in the articles. At the encouragement of FML management, I stopped telling people what to pick and shifted into a narrative that gave people educated choices based on what set of facts they chose to believe, which softened the cries from my critics.
The Comparator got a lot better too, as people started making suggestions for input alternatives and stressed how page load time wasn’t nearly as big a deal as having more combinations to consider. And I started to develop my current predictive model, which is similar to the Monte Carlo simulator but a lot smarter about what it places importance on. It isn’t perfect either, but through regression testing I at least know its shortcomings and continue to discuss them in articles.
The data deep dives have exhausted themselves for awhile, but as I type this Patrick Reardon, The Box Office Guy, and I just recorded our third podcast together and that should be fun to develop in the coming year. The three part new player series I just wrote in conjunction with several vocal players seems to have received good reviews as well. I continue to have a blast playing this game and appreciate everyone who has given me feedback, positive or negative, along the way. Thank you so much!
It seems contrary to my reputation to not conclude this with some statistics, so here’s a page view graph from the three places where my FML content resides:
The Captain Obvious within me points out that I had one article go live on the main FML site at the end of August and the growth of Comparator traffic in particular rose pretty substantially after that. If you read into it a little more, the Comparator gets about 2,500 page views a week for a player base of around 10,000 for most of the life of this graph, meaning that at best 1 in 4 FML players use the Comparator. In reality it’s probably closer to 1 in 8 if you figure an avid player might visit the Comparator twice a week. So if you use the Comparator, you’ve got an edge (assuming it gives you one) over around 88% of the players and possibly more.