Článek v češtině zde / click here for Czech translation.

Where I’m from, the NFL is called American football. That alone tells you that fans like me are in the minority. Naturally, the first time I saw the NFL on TV, I had no idea what was going on and nobody around to explain. Big men were grunting for a few seconds and then there were thirty or so seconds of waiting until the ball was snapped again. A strange thing happened several years later as I’ve actually played a quick game of football during baseball camp. I finally understood the point of the game, starting with the simple rule: you have 4 downs/attempts to gain 10 yards in order to get first down, renewing your set of 4 downs. Keep doing this until you reach the end zone, you score a touchdown. Fail to do it, the defense won this one. Once I grasped this simple concept, I was hooked. My relationship to this tough sport – in which seemingly most of the time nothing is really happening – quickly shifted from indifference to a shallow hobby to a source of countless sleepless nights to an extra income.

In 2014 I was publishing free betting picks on Czech betting site KolemDvou and had such good result that next season I was offered to publish only paid premium picks. On the site, I established myself as a credible figure and true connoisseur of the game. Last season I was on a roll and provided not only informative write-ups but I even wrote a 5-part series about the sport both for potential new viewers as well as experienced bettors. Shortly before the 2014 post-season I also created a rating model built on play-by-play details. Transforming the plays from ESPN to tidy excel spreadsheets kept my Mondays busy for the next couple seasons. The model helped me pull statistical comparisons for any two teams I wished and after my very first primitive spreadsheet sucessfully predicted Colts upsetting the heavily-favored Broncos in the divisional round, it was on.

First version of the model was called IDBOS, which was either a random cluster of letters, or a bad, unpronounceable abbreviation of some sort. I don’t even remember. Eventually I stopped trying to figure out cool name and just landed on Anthony, after Matthew McConaughey’s role of football tout in Two For The Money.

With so much preparation and with this new weapon in my arsenal, I was ready to go. Everything seemed perfect as my 2015 betting-advising venture started off really well. I went 2-0-1 on Thursday season opener and 4-1 the first Sunday. At this point, I felt unstoppable. False. Reality check. Next two weeks I went 3-7 and 5-7, erasing my winnings from the first week and more. Then week 4 came with a bang as I went 6-1. Well, I could go on and on, but basically, this was my season in a nutshell. Ups and downs, highs and lows, heavy swings, I didn’t have a single average week to settle down a bit and mentally that took a toll. I realized this game is not for me. In hindsight, it was my love for analyzing games and match-ups and writing all the reasoning down that got me here. Selling picks was my way to have my thoughts heard. But bettors don’t give a damn about your thoughts unless you win and as the season progressed that’s what I did less and less. Betting on the most efficient sports market out there is not an easy job and the stress of losing other people’s money made me almost hate NFL. Of course, once I gave betting up, I started enjoying watching the game again.

After this failure, I switched my attention to Anthony. I added more and more stuff into it and every off-season worked backward to add another two or three years in the database, to have more data and more reference points. Basically, the whole model runs on play-by-play information. I have details for every play that happens. That meant that if from time to time I went online and in someone’s blog post I’ve seen a stat I liked and didn’t use, ten minutes later I would be working new formulas into my model to have that stat at hand. Not all of them were kept around. Bunch of stats didn’t make it past the test ride; I would create a stat and two weeks or even years later I realized I don’t particularly use it, like it nor trust it.

Dennis Erny, the creator of Armchair Analysis, was collecting NFL play-by-play details as well. Only he did it better: he was running a successful business doing it and before the 2016 season, he was looking to expand into new territory, play charting. To chart a play, you have to actually watch it and mark whatever you’re interested in manually, there’s no way around it. It takes many eyeballs, a lot of time and some brains to chart a full NFL season, but this data is incredibly valuable because it provides context that normally isn’t there. For example, there are plays in which quarterback is under pressure, but the context tells you that the play call was a screen pass, in which case the pressure is invited and designed around by offense; or there’s an interception, but the context tells you that the ball was intercepted only after the receiver tipped the ball and that it’s hardly QB’s fault. That’s why it was very exciting for me to be accepted as a volunteer charter for the 2016 season.

In 2017, there were more data points to collect and sadly, more quitters, but in the end, that didn’t really matter. Dennis took a chance on me as an unproven European and I was ready to deliver. I took plenty of workload on my back to get the project going no matter what and that resulted in my promotion to the senior analyst; and that resulted in me watching so much football for couple weeks in a row, that I actually had to skip one football Sunday, which never happened before or since. It was a major case of NFL overdose.

In 2018 my responsibilities grew again. Now under contract, I was officially getting paid for watching NFL. This senior position not only required me to watch every pass of the season, but I also had a say in which direction the project should go. Before the 2019 season, I was named a “Head of game charting”, further growing the level of responsibility and number of hours I can spend doing what I love. It started making sense as a business, too. We rolled into 2020 with a new attitude and for the first time ever, we didn’t use the offseason to take a break and just kept on rolling. It paid off: on the back of the NFL covid-season, we’ve put together the biggest downloadable data collection on the web, for about third of the prize our competition asked. This even attracted investors and eventual buyers of the project.

One of the features we installed for 2020 season was tracking the field not only on the vertical, but also on the horizontal level. Among other things, this gave us an information about how many yards the ball really traveled through the air. Our numbers lined up with Nextgen Stats, who have the benefit of using automated tracking systems using microchips installed in players’ pads.