It’s been 20 years since I started watching the UFC, 15 years since I wrote my first optimization algorithm, 10 years since my first publication in a peer-reviewed scientific journal, 5 years since I learned about DraftKings and daily fantasy sports (DFS), and 3 months since I reached out to Angelo here at We Want Picks to offer my services. So what exactly are my credentials, and why on earth should you give a damn about what I have to say? Well, it would be hard to list all of my credentials and remain anonymous as I plan to do, but to put it simply, I am a Ph.D. nuclear physicist working a day job at a prestigious national laboratory. I have 2 bachelor's degrees, one in physics, and one in nuclear engineering, and while my PhD is in nuclear engineering, my job title says computational physicist. I graduated with a perfect 4.0 GPA from one of the best engineering schools in the country.
When my brother introduced me to DraftKings 5 years ago, and told me people can make a lot of money at DFS, I said “I’m pretty sure I’m smarter than those people, so let’s give it a try.” I developed a software tool that I called Megalodon, to build thousands of unique lineups based on statistics compiled from the internet. The name was meant to signify that while there are plenty of sharks in the water, there was a bigger predator entering the DFS environment. And so after winning a bunch of tournaments and many thousands of dollars, me, my brother, and a close friend of ours were kicked off DraftKings for a technicality (apparently you can’t give lineups to other people), and for the past year, this little project of mine has been dormant. To make matters worse, direct sports betting is illegal where I live, and the other DFS sites like FanDuel and Yahoo have really shitty tournaments for MMA.
That’s when my brother told me about We Want Picks, and told me to watch them on YouTube. I heard Angelo say something to the effect of “We are hiring,” so I thought, if I can’t bet anymore, maybe I can help other people make money in this space. Ideally, I would have started my own site, but that didn’t really sound very fun to me, and besides, I have literally zero social media footprint to promote it. I’ve never had a Facebook, Instagram, or Twitter account. The closest I ever came to that was a MySpace page that my band had when I was a teenager. So starting from the ground up seemed like a lot of work for a part time gig. But Angelo and Jacob seemed sincere and knowledgeable, and an excellent team to hitch my wagon to.
While the tools I had pitched to Angelo were all DFS based, he proposed that I build an AI to predict fight outcomes, and 2 months later came PickGPT, a neural network trained on historic fight data to predict fight winners. It has been operational for 2 events thus far, with a success rate of 7/10 and 8/11 (those would have been 8/11 and 9/12 if it weren’t for the no contests in the Imavov Vs Curtis and Argueta Vs Lawrence fights). In addition, it has only taken 5 underdogs in those 2 events, and 4 of them won, and the other one was Dan Argueta (#AllNonsenseKeithPeterson). For the record, the other 4, were Charles Oliveira, Kyle Nelson, Carlos Hernandez, and Nicolas Dalby. Hey Jacob, maybe you should consult PickGPT next time you are looking for a dog to make your lock of the week! And for any doubters, I’ll include the pictures that I sent Angelo time stamped below.
While these preliminary results are impressive, when we test it against every fight going back to 2012, it’s success rate is currently only 65%, and hence the early success may be an outlier. This is the reason it is still technically in BETA, and should be used with caution, like any new technology. Each week, input is added, and the historical success rate creeps up, and it’s only a matter of time till PickGPT has a higher success rate than any human could aspire to. However, while it is still in its infancy, if you had placed a unit on each fighter it picked over these 2 events, you would’ve made a significant return on investment based on the odds at the time each fight went live.
Despite how it may seem, the purpose of this article is not to brag, nor to assert dominance over Jacob, although both of those have been fun. The real purpose of this article is to gauge interest among current and potential Premium Members in the tools I have to offer. While posting picks from PickGPT is easy, providing access to the DFS tools mentioned earlier would require some serious investment in compute power and careful reading of DraftKings terms of service to make sure none of our members run afoul of their rules (although as I have been told by them, they can kick anyone off their platform for any or no reason at all, completely at their discretion). So if you are a DraftKings DFS user, and want to know more about these tools, make sure to message Angelo so he knows you are out there! And oh yeah, the DFS tools are available for NFL, NBA, and MLB as well!
