Introduction by Jack Andrews: Finding your own unique path is how you can achieve your goals in sports betting. Sometimes it can be very useful to read about another bettor’s journey to help shape your own. I was impressed by the journey Tomas outlined on his Twitter account. I’ve asked him to expand on it in this article. I think it’s useful to see how his process evolved with NBA betting.
In 2020, sports betting became legalized in my home state of Illinois. That, combined with an amount of free time unparalleled at any other point of my life, provided the perfect confluence of motivation and opportunity for diving deep into my favorite pastime, the National Basketball Association. I always appreciated the use of analytics in sports. So the challenge of actually analyzing the data myself was exhilarating. It proved more difficult than I’d imagined. At this point two years ago, I was at ground zero when it came to using machine learning for data analysis with regards to NBA betting.
Studying The Experts
I started digging deeper, sifting through the internet, finding people doing interesting quantitative analysis. Eventually I found Captain Jack Andrews who was making these surprisingly informative YouTube videos on how to build your own models.
I went quickly from not knowing what scraping data meant to doing my own work crunching numbers in big Microsoft Excel files. Andrew Mack’s book Statistical Sports Models in Excel helped accelerate the process.
Then, I revisited articles from some of my favorite hoops writers over the years like Kevin Pelton and Seth Partnow to see if I could piece together how they did some of their research. I tweeted at or messaged people who I thought were doing similar work and hoped that they’d share some insight. I even went to Partnow’s book signing. Sat in the front row, and asked him during his Q&A about how to best use data to analyze defense in the NBA. Spoiler alert: it’s very hard!
I finally reached the point where the math/computer science got a little too complicated so I hired someone to help put the finishing touches on my model.
Along the quest to become my own mini version of Bob Voulagris, I was surprised to find that while a strong model will always help on the margins, the best way to beat the book comes from a more simple, logical, information driven place. Let’s start on that process before exploring some NBA specific information to beat the books.
The Path To Profitability In NBA Betting
Respecting The Market
If you’re familiar with Unabated, you’ve probably heard countless times about the importance of line shopping and respecting the market. That being said it bears repeating.
The market, particularly if you’re betting sides or totals for a game, is almost always close to what the “true number” should be. Having access to as many different numbers to get the best price can ultimately be the difference between doing this for a living and merely treading water.
The closing line also works as a great test for any models you develop. If your model is constantly drastically different from how the market values specific teams/games you almost assuredly know your model is bad.
As time passes, you’ll realize which books are accepting large wagers from sharp money and making the market. Using that information allows you to attack slow moving books that struggle finding the right number.
If you’re newer to sports betting, or unfamiliar with who has the sharpest lines with the NBA, the Unabated Line is a great tool for Premium members to use as a starting point.
Player Availability Information Is Gold
If you want the highest return on investment beating spreads/totals on the NBA, this is the way to do it. If you know that a Giannis, Steph, or Jokic is out for the game before the sportsbook that you’re betting against figures it out, then you’ve given yourself a roughly 25 percent projected return on investment on that bet.
The undisputed champ in breaking news fast right now is Justin Phan and his team on Twitter account @underdog__nba. Simply following them will get you ahead of the slowest books. But if you took some time, you might be able to compile a list of beat reporters who will get you ahead of even them by a couple of seconds.
This also works as a great test for which sportsbooks actually know what they’re doing. I almost never have the injury info before the sharpest books, while routinely getting ahead of the least skilled books. Figuring out which books are on top of this and which books are slow can be extremely profitable. A second or two often makes a huge difference.
Detailed Tracking of Results
Figuring out what you’re successful at versus what you’re not is paramount to maximizing winnings.
For years I loved playing DFS. How to best utilize salary and put the puzzle together always struck a chord with me. After a more careful review of the results, however, I found that all I could really hope for was to breakeven. Plus, it was a time-suck. So, I reallocated more energy toward other areas like NBA player props that I found less interesting yet easier to beat when getting started.
Tracking your results in more detail may also help you find unknown biases in your approach or bugs in your model.
Toward the end of my first year I noticed that I was betting more underdogs/teams with bad records. I was still getting (barely) positive closing line value, but not nearly as much as with my bets on favorites/good teams. It turned out I was regressing a variable twice on accident instead of once while making some in-season changes to my model. It wasn’t moving a team in my power rankings by more than half a point, but it ate into my margins and caused systemic bias in favor of worse teams.
At last check there were a couple of OK tracking results apps out there, but having your own Excel spreadsheet to get as many variables as possible might help you cut down on similar mistakes.
Automation Helps Prevent Fatigue
The NBA season can be an absolute grind. There are games nearly every day. Information about injuries or roster transactions comes flooding in constantly. If you’re going to do this somewhat seriously/professionally a certain amount of fatigue and burnout is inevitable.
That’s why it’s vital to make every part of your process as easy as possible. Let the machine do the work. Consolidate your most important information into a smaller file or spreadsheet instead of clicking through a bunch of tabs. It’s far from the sexiest part of the job, but by tidying up loose ends as quickly as possible means you’ll have more time to research other areas. You’ll thank yourself for taking care of this as early as possible. Come mid-February there are few things that will annoy you as much having to click through a tun of information to try and figure out just how much a previously unannounced injury should move a line on some already boring Thunder/Spurs type of game.
Using Qualitative Reasoning To Determine Signal And Noise
A specific veteran’s career three point shooting percentage is 33 percent coming into a season. This year, through 30 games, he’s at 45 percent. What should we expect him to do in 3-point shooting going forward?
Lots of factors could be influencing the players’ fluctuation in three point shooting. Was it actual improvement caused by change of mechanics, increased repetitions in the offseason, or a coach helping a player gain more confidence in his ability? Or did it have to do with a change of roles or a new team that’s helping a player get easier looks from three point range? Or is it just random variance?
A team’s defense was projected to be average coming into the season, but they’ve played at a Top 5 level so far this year: How many games at this level do we need to see before we conclude they’re actually a Top 10 defense going forward?
Did the team change their scheme or coaching to put players in better positions to succeed defensively? Did a certain player make a large, unexpected leap in their defensive skillset that’s spearheading the improvement? Or have they just been fattening up on a relatively easy slate of opponents?
Use the Great Research Tools Available
Data analysis in the NBA can provide a strong starting point for answering questions like the ones above. Additionally, there is a lot of great publicly available research to get you started. There’s somewhat of a double edge sword to this however. Several brilliant thinkers have created publicly available research but it’s also available to every other originator.
Where I’ve found success and had my own breakthroughs was by taking some of this available research, and adding a step or two on top of it.
There’s a plethora of all-in-one player metrics: DARKO, EPM, LEBRON, RAPTOR to name just a few. Their accuracy varies from OK to excellent as far as analyzing specific player performance. Some assess past performance, some predict future performance and some do both.
These measures all take a combination of specific box score stats along with their respective RAPM (regularized adjusted plus-minus) to predict player performance.
There’s research that shows which is best. However, even taking a simple average will give you a pretty good idea of a player’s value. You can divvy up how many minutes and games each player is going to get. From that make a rough depth chart and you can project a simple rating/win total for each team.
A Unique Angle: My NBA Betting Model
I’ll start with a disclaimer that’ll be obvious to any pro bettors reading this. There are far easier ways to go about making money than betting sides on NBA games. Yet taking a bottom-up approach was the part I found most interesting and generally allows for the most account longevity.
The flaw with all of these amazing all-in-one metrics is that there are several variables that are out of their scope. Simply put, context is key. Certain players who thrive in certain roles may suddenly crater if they are asked to do more on the court. Other players will have stronger synergy with certain teammates and wind up elevating each other’s’ performances.
Some players will play for high-level coaches that know how to best utilize their talents. While some will have the misfortune of playing in schemes that don’t allow them to prosper. Some teams try to win as many regular season games as possible. Other teams are basically waiting for the playoffs to start, tank for better draft position, or focus more on developing players for the long term at the expense of the present. Asking an all-in-one player metric to figure out intent is impossible.
Adjusting for Team Performance in NBA Betting
So what did I do? Instead of taking all the players, giving each a plus/minus rating and certain amount of minutes then calling it a projection. I took the team’s performance as a whole. Then I looked at 27 different variables to distinguish what was signal vs. noise in their performance. I factored out injuries and tied it all together with a preseason prior.
One season is obviously a small sample size, but the results were pretty good. My Pietrus Model was really high on both Boston and Memphis after slow starts to the season. On the flipside Pietrus almost immediately realized poor coaching and roster construction led to overweight ratings on Brooklyn and the Lakers.
How to Make NBA Betting Work for You
When I first set out to become a professional sports bettor, I was motivated by the idea of having freedom in how my time was spent in my daily life. I could replace a boss’s demands and instead interweave betting with the rest of my life.
It didn’t work out like that. I’d spend a couple nights a week researching until 3 or 4 a.m. Then wake up four hours later to be ready before the line originators released their lines. Every afternoon I would grind through player props on as many games as I could. I’d also try to manage a handful of DFS rosters.
I wound up missing meals and skipping quality time with the people close to me. I had awful days where I’d put in 14 hours of work and still lose what at the time felt like a lot of money. It was too much.
Refine Your Approach to NBA Betting to Fit Your Schedule
As time went on, I made adjustments. I started by automating some stuff I had been doing by hand. Then, I quit DFS altogether.
I decided to focus on key times during the day. For instance, player availability information typically dropped around the same time each day. I would designate exercise time in the gaps between news. I decided to only toy with player props on slates with fewer games to ensure I had adequate free time. Finally, I developed enough confidence in betting sides/totals that it became the most profitable thing I could do considering the limits involved.
I accepted that by not working as much I was definitely leaving some “value” on the table. However, it was worth it to create the lifestyle I was interested in pursuing in the first place.
At the end of the day, I have respect for everyone. From 100 hour per week steam chasers to innovators and originators. From player prop grinders to the folks who focus on betting season-long props using sports betting as a side hustle. Figuring out which of these “avenues” works best for you is ultimately how you’ll be most successful in this field.
Key Takeaways
- Study the market to find the best approach.
- Player lineup information is key.
- Use automation to prevent fatigue.
- Lean on metrics, but add a qualitative approach to them.
- The best path for you might not be the most time-intensive path.
Unabated Premium Subscribers have access to advanced tools for their NBA betting such as our Alternate Lines Derivative Calculator, our In-Game Betting Tool, lightening fast Premium NBA Odds Screens, and our NBA Player Prop Tools.
