Finding an Edge and Keeping It By Harry Crane With the right combination of skill, discipline and mindset, it is possible to make a significant among of money betting on sports — maybe even enough to quit your day job. Whether you want to do it full-time or make a little extra money on the side, the key is to identify which approach works best for you, and to implement that approach systematically over an extended period of time. In this post, I discuss some important practical considerations for the aspiring serious, semi-professional, or professional sports bettor. Staying Within Yourself What do an expert data analyst with little or no betting expertise, an expert computer programmer with no understanding of data analysis, and a high school dropout who doesn’t know any math, statistics, or computer programming have in common? They can all be winning sports bettors. But each needs to play to their strengths, and minimize their weaknesses. As my baseball coach would say whenever I was up to bat, “Stay within yourself.” The same advice applies to sports betting. A singles hitter who tries to hit a home run every time he’s up to bat is going to fly out or strike out most of the time. In the same way, a sports bettor who tries to win using techniques he hasn’t mastered will end up depositing a lot more than he withdraws. If you don’t know any math beyond high school algebra, then implementing an advanced machine learning model probably isn’t where your edge in sports betting lies. If you’re really good at modeling soccer, but don’t know anything about baseball, then betting on baseball probably isn’t the first place you should look for an advantage. If you’re expert at web scraping, data processing, and automation but don’t know anything about data analysis, then a less statistical and more technical approach might be your best bet. Identifying the approach that works is a personal decision. Whatever approach we choose, it’s important to have a clear idea of why we expect it to give us an edge and how we plan to implement that approach in a real world betting situation. Even with clear answers to these questions, there are a number of potential pitfalls between the ideation stage and the execution stage that we can’t ignore. Before we can be confident that our approach will be profitable over the long-term, it’s worth asking: What is the source of our edge? Why are we able to execute this approach better than other bettors? How can the approach be refined to guard against common pitfalls and increase our edge? Sources of Edge There are 3 primary ways to get an edge as a sports bettor. We can have better information, better execution, or better models than the sportsbook. As a bettor, if you don’t know which of these applies to you, then you probably don’t have an edge. Back to the drawing board. Even if you do know where your edge comes from, there’s a good chance your edge is smaller than you think it is. It may not even be an edge at all. There are a number of reasons for this that I’ll discuss in more detail in another article, but some possibilities include: There may be others in the market who are leveraging the same edge (i.e., the same source of information, the same method of execution, or the same modeling approach). There are others in the market who have an advantage that you don’t have, and therefore are either taking advantage of the same opportunities for different reasons or avoiding spots that your approach identifies as profitable based on information they have that indicates otherwise. What was a profitable edge last season might no longer be available. The rest of the market may have caught up. Sportsbooks have built-in advantages in information, execution and modeling that are hard to overcome. Regarding the last point, sportsbooks have the following built-in advantages. Information Advantage: Sportsbooks have access to information that bettors don’t have. Most obviously, sportsbooks know which sides and for how much all of the other bettors in the market are betting. Because some of these bettors are using methods that are either the same as or correlated to our approach, the book can sometimes counteract our strategy based on betting patterns of competing bettors. Execution Advantage: In setting limits, taking markets down, reviewing bets before they are accepted, and punishing or banning winning players, sportsbooks enjoy a first- and last-mover advantage. Their first-mover advantage allows them to set low limits on markets they are most vulnerable (or to remove the markets entirely). Their last-mover advantage is the ability (in some jurisdictions) to review bets before accepting them, and also to ban winning players. All of these work to the advantage of the book and the disadvantage of the bettor. Modeling Advantage: Sportsbooks have their own internal data science teams developing their own models. So while it’s definitely possible to make money as a sports bettor, there’s a big difference between identifying a potential advantage and profiting from that advantage. It’s not enough to be good at what we do. We have to be better than most or all of the competition (books and bettors included). Because we know that there are other players in the market who know what we know and there are other players in the market who know things that we don’t know, every potential bet that our approach identifies as profitable is a possible aberration. Every profitable bet that we identify is a bet that the rest of the market apparently hasn’t identified as profitable, and therefore is an invitation for us to rethink why we might be wrong. Now, for the economics scholars out there, don’t worry. I’m not going down the usual Efficient Markets Hypothesis “If you see $100 on the sidewalk, you shouldn’t bend over to pick it up because if it was a real $100 bill then someone else would have already picked it up”-rabbit hole. We don’t want to second guess ourselves to the point of never betting at all. Edges do exist. But taking a moment to rethink things is an opportunity to either (i) reaffirm our edge and bet with more confidence and also (ii) correct deficiencies or blindspots in our approach that will increase our advantage moving forward. Below I’ll discuss how this mindset applies to each of the three situations highlighted above. Information Advantage An information advantage is knowing something that nobody else knows, e.g., injury news, lineup changes, rule changes, etc. (Note: access to public information, such as the Draymond Green incident, is not an information advantage because everyone had access to this information at the same time.) Having access to information that nobody else has is like being able to see the future. We can take action now, and watch as the rest of the market tries to catch up later. For this reason, an information advantage is probably the most profitable edge in sports betting. If our information is correct, then we know how the markets are going to react once the information goes public. We have as close to a sure thing as we’ll ever get. But even with such a big advantage, we have to ask ourselves: if a piece of information in a particular situation is so obviously being ignored by the market, is our information as valuable as we think is it? Do we really know something that nobody else in the market knows? Maybe. But in any given situation, the market disagreement could also indicate: Our information may be incorrect. Our information may not mean what we think it does. That is, it might not have as big of an impact on the probable outcome as we initially believed. There may be other important information that we don’t know about which nullifies or diminishes the advantage gained from our information. For example, we might have inside knowledge that half of the Lakers starting lineup is out with Covid, but if half of their opponent’s starting lineup is also out with Covid and we don’t know about it, then our partial information is much less valuable. These are just a few of the possibilities. Of course, if we have a trusted source that has served us well in the past, then we may be able to quickly rule these out and place the bet with confidence. But, especially if a particular source of information is new or rare, it pays to ask the question. Last thing we want is to bet a large amount of money on what we think is a sure-thing but turns out to be based on information that is either well-known, priced-in, or simply false. On a practical note, even though an informational advantage is a very profitable way to get an edge on any given play, it is also very easy for sportsbooks to detect, and therefore take countermeasures against. Before acting on a piece of information, it’s worth considering whether the potential payoff on a single bet is worth the risk of detection, and thus the risk of possibly losing profits from all future bets at the book. Execution Advantage An execution advantage is the ability to act on information faster and more effectively than others in the market. We don’t need access to non-public or little known information. As long as we’re able to execute faster on information everyone else has, we’re in a position to profit over the long-run. Arbitraging between books before one of them moves is a common example of an execution advantage. The Draymond Green prop mentioned above is another example of an execution advantage: despite the sportsbooks built-in advantages in information and execution, bettors were able to execute on this public information before certain sportsbooks. But again, even with an execution advantage, any time our bet is accepted is an indication that we might have missed something. Maybe: The market doesn’t believe our action is sharp enough to limit or deny outright. We think we’re acting early, but in fact the information we are acting on is already reflected in the line. If we’re using automated software to execute bets, there’s always the possibility of a technical failure. Even if we do have a real advantage in the vast majority of situations, a bug in our software could wipe that out in a single mistake. Of the three advantages we discuss here, a consistent edge in execution is probably the least susceptible to second-guessing. But it pays to be vigilant. Also, as mentioned in the above section on Information Advantage, being faster than the rest of the market may be a reliable way to profit on any given bet, but it is also easy for a sportsbook to detect, and therefore take countermeasures against. The art eventually overtakes the science: Once the system is in place to execute bets consistently, capitalizing on this advantage requires significant effort to avoid detection and to maintain the edge over a long enough period of time. Modeling Advantage A modeling advantage is an ability to more accurately price events than the market. On a per-bet basis, an edge in modeling is usually smaller than an edge in information or execution. But superior modeling can offer a bigger potential to scale with the number of markets offered on a given sport. If we’re really good at modeling a particular sport (e.g., soccer) then we can potentially leverage this advantage in every soccer betting market around the world. While there are many potential advantages to modeling, there are also major challenges to profitably implement such an approach. Aside from needing significant technical knowledge in statistics, analytics and data management, we still need to compete with the sports books and other sharp bettors in the market who are leveraging their own advantage. First, the sportsbooks have their own models, which our model has to be better than. Second, the sportsbooks have built-in information and execution advantages, as discussed above, which generally means our model needs to be a lot better than the sportsbooks’ model. Third, even if we overcome these first two challenges, we’re still competing with other modeling teams and bettors who have better information and execution than we do. To be a profitable model-based bettor, we still need a good amount of expertise in the information and execution departments. We need to understand how the market works, and develop the intuition to know when the market is telling us that our model is missing something. When our model finds an apparent edge in the market, it’s worth considering whether there is a key piece of information missing from the model, the perceived edge will disappear before we can act on it, or the model is simply wrong in assessing the situation. All of the above situations are common, and will happen from time to time, even with a profitable model. This is why the per-unit profit (ROI) tends to be lower and the volatility tends to be higher for model-based approaches than the others mentioned above. In fact, the situations that are easiest to act on tend to be those where we are in one of situations (1)-(3). I’ll discuss this in more detail in a future post on market support and market resistance. Ideas are easy, Execution is hard Becoming a winning sports bettor isn’t easy, but it is possible. Having a clear plan to take advantage of better information, execution, and/or modeling capability is an important first step toward success. It also helps to not only know what we don’t know but also know that there are others out there who do know what we don’t know. Without a solid foundation, there’s little hope of long-term success. Even with a strong foundation, there are practical challenges to avoid detection and maintain our edge in a competitive environment. For the aspiring sports bettor who hopes to make significant money through betting, these challenges should be welcome. The fact that it is difficult means that most people won’t be willing to put in the work to become winners – that includes sportsbooks! If you’ve found this blog, then you’re already in a good position to find actionable betting information and develop smart ideas. All that’s left now is to execute.