Every summer, I begin the arduous process of updating my college football and NFL models to prepare for the upcoming seasons. Invariably, I have to go through a bunch of code I’ve written years ago and try to remember what the hell I was doing. It would all be so much easier if I had just thoroughly commented my code when I was writing it, yet while I’ve gotten somewhat better about this, I still fall woefully short in this area. The problem is that when I’m actually writing code, I don’t need to comment it, because I know exactly what is going on. Why do I need to needlessly jot down notes about things that I already know? It’s amazing how much I forget. I can be in the zone, deep into a topic, and a year later have no recollection of what I was thinking.
Adding Notes To Your Handicapping Process
Many bettors are guilty of the same mistake I make in a very different area: record-keeping. Whether you’re a qualitative (narrative based) or quantitative (numbers based) handicapper, your records are an invaluable resource that can help you grow as a bettor. But it requires an investment in time now to pay dividends in the future. My biggest regret as a bettor — aside from betting on the Jets too often the last few years — is not consistently keeping more detailed records in my handicapping process.
A quantitative handicapper can (and should!) log their expected return on a bet. Logging CLV is very important for everyone. But if you have a model, you should log your model’s fair price and your theoretical edge. For quants, a leap of faith is always required to bet using a model that you built and thoroughly back-tested. There are two primary concerns:
- Is the model overfit? Are the edges actually real?
- The production code doesn’t work the same way as the model-build code. Are there bugs in the production code? Are you using a different data source for production than you did to build the model? If you log your fair prices you can identify if there’s a flaw somewhere in the production process.
The Cooper Rush Trick Play
I’ve bet NFL and college football second halves since 2010. My 2H models have changed over the years, but they have always been built based on data that I paid for, that is as close to error-free as possible. But that data isn’t available during the game. To productionize the model, I wrote scrapers to pull play-by-play data from ESPN.com (and later, NFL.com). In 2017, when my CFB 2H model seemed way more biased towards Unders than it had been in the past, I was able to compare the productionized model prices to the prices I got using the more official data available after the game. It turns out, there were some missing drives showing up from time-to-time in the ESPN play-by-play data.
Then there was also the time that my play-by-play parsing code thought Central Michigan never passed the ball. It turns out, I was labeling every pass by quarterback Cooper Rush as a rush.
Greater Benefits To The Qualitative Handicapper
Paradoxically, a really detailed bet log has more value to a qualitative, intuitive bettor. As a quantitative bettor, I have systems (models) that I build. I don’t have to make lots of little decisions on a day-to-day basis. The decisions were made in designing the system. Evaluating the system is fairly straightforward.
But qualitative bettors have systems as well. Maybe your system is “follow picks Dave Portnoy tweets out”. Maybe it’s “fade picks Dave Portnoy tweets out”. Being able to categorize the system you used to place a bet is critical to being able to determine whether that system has an edge in your handicapping process.
Someone may have a system of making lines on the following week’s NFL games without looking at the market, and then betting the games where their line differs substantially from the market. In your bet log, note that that was your system, and how far your line differed from the market line. I would assume that most qualitative bettors have many different systems for placing bets. Some of these systems may have an edge, while others may be unprofitable. If you don’t identify the system you used to place the bet in your bet log, how will you know?
Logging Your Confidence As Part Of Your Handicapping Process
While you may not have an estimate of your edge on a bet, sometimes you have a higher degree of conviction than other times. Log that shit. Classify your degree of conviction as high, medium, or low. Being successful is about building testable hypotheses. Logging your degree of conviction will help you test how strong your intuition is. One of the biggest mistakes bettors make is playing too many games. Are you getting good closing line value when you have a higher degree of conviction, but not when you don’t? You probably are betting too many games.
Sometimes there are factors outside of your control that affect line movement. Two starting wide receivers are ruled out. Create a “notes” column in your bet log. If a line moved against me 6.5 points because the starting QB got hurt after I placed my bet, that could have a huge impact on the conclusions I draw about how good my system is at generating CLV — unless I explicitly noted that in my bet log. A good bet log helps you identify the signal and weed out the noise.
Learning From Drunk Card Counting
Recently, I found an old bet log from 2008, when I was working at Las Vegas Sports Consultants and betting in my free time. I logged not just sports bets, but blackjack, poker, and video poker sessions. Not surprisingly, returns on sessions with the note “drunk card counting” were significantly worse than ones without that designation.
The one skill that I recommend every bettor invest in is a basic knowledge of spreadsheets (Microsoft Excel/Google Sheets). A good bet log does you no good if you can’t interpret your results. It doesn’t take a lot to summarize your results and CLV, filtering by betting system, but if sumif() and vlookup() sound like a foreign language to you, I highly highly recommend investing the few hours it will take to build a basic spreadsheet proficiency. The internet is a wonderful place if you need to learn a skill.
Logging bets is a tedious process that few enjoy. But a good bet log is more like a bet journal. It’s a record not just of what bets you placed, but of why you placed them; your journal where entries can be read, organized, and analyzed later, which should have a direct impact on your future betting decisions. It’s an investment of time that – much like rigorously commenting code – can feel unnecessary in the moment, but can be a hidden source of edge in your handicapping process, helping you improve your process and your ROI long term.
Whether you are Qualitative or Quantitative in your approach to sports betting, Unabated has tools and resources for you to use. From lightning-fast line shopping with our Odds Screen to simulator tools for all your player prop projections, we want to be your betting companion. Take us for a spin today with our trial subscription.