Editor’s Note: The article below originally appeared in Berryhorse’s (real name Kieran) free newsletter BetItUp, which you can (and should) subscribe to here to learn more about predictive sports modeling, betting, bankroll management and more. The article is published at Sports Handle with his permission.
It’s very easy to lose money betting on sports. Losing wagers may still provide good entertainment for a few hours, but people wanting to actually make money need discipline and at least a basic understanding of math and probability.
Some bettors consider themselves “Positive EV” or +EV bettors, referring to positive expected value. There’s a bunch of articles on the subject that are too complex, especially for those not mathematically inclined. So if you’re encountering EV principles for the first time or need a refresher, we’re pleased to share what’s below by Mr. Berryhorse, which should be digestible by sports bettors of all levels.
Positive Expected Value (+EV), Probability And Implied Probability
Probability In Sports Betting
Most of us went to high school. While at high school, most us took some math class dealing with probability, statistics or both. And while we were suffering through the lectures in this math class, most of us probably stopped paying attention 10 minutes into the first day, thinking “When will I ever ACTUALLY use this in real life?!”
I don’t blame anyone for their decision there, but as one of the few nerds who did pay attention, my answer is a resounding “every day.”
For anyone interested in predictive sports modeling – by far the most important prerequisite is a strong foundation in Probability Theory and Statistics. You by no means need a Ph.D in math, and honestly, the curriculum from the aforementioned high school class is probably even overkill. But there is almost no way to consistently beat a sports betting market without understanding the basic fundamentals of probability. And shockingly, so many sports bettors continue to implement methods that completely ignore these crucial principles.
I hope this hasn’t scared off the readers who are more mathematically-inclined, and who do have a strong understanding of stats and probability – there are still some key points in here that are specific to sports betting. Plus, I’ve found it’s always helpful to review fundamentals too, even if you are an expert.
Today I’ll just explain some really simple points about implied probability and expected value, but these newsletters will slowly build upon one another in a series of lessons on probability in sports betting.
Implied Probability And Expected Value
Sports are really crazy. As fans, we’ve all seen stuff that we know is statistically unlikely – 92-foot heaves in basketball, 28-3 comebacks in the Super Bowl, and pitchers hitting flying animals in mid-air. Events like these should immediately make it clear to everyone that there is no such thing as “a lock” in sports betting – even the most lopsided matchups do not have true probabilities of 0 or 1.
The entire premise of sports betting is to come up with some number between 0 and 1, that represents the chance of something happening – usually a team winning, or scoring a certain amount. If you can do that with some level of accuracy, you can compare it to the implied probability of the line or odds at your books and see if you have an edge. Calculating this implied probability is extremely straightforward – I’ll use American odds to illustrate but this calculator is useful if you prefer a different format.
With each line it sets, a sportsbook is implying some probability of a bet winning – in other words, it’s declaring that bettors would “break-even” if they won this bet a certain percentage of the time in the long-run. The obvious example is a flip of a fair coin – the true “line” for a coin flip should be +100, implying a probability of exactly 50%.
Here’s an extremely simple formula to convert American odds into an implied probability:
Once you grasp this simple concept and start converting lines into implied probabilities, you now have a way to see if a bet has positive expected value or is “+EV.”
Why I Bet on Teams I “Expect” to Lose
The end-goal of every sports betting model I have ever built, and hope to inspire and help others build, is to calculate the probability of something happening (again, this is usually a team winning a game, or scoring a certain amount of points).
This end-number the model spits out is what we use to compare to the implied probability in Las Vegas to see if placing the bet would be +EV – if we would expect to turn a profit on this bet if we could place it over-and-over-and-over again given the same circumstances. An easy example is if you could bet on a fair coin landing on “heads” at +110 odds. +110 has an implied probability of 47.6% – so it is +EV, because the true probability of landing on “heads” is 50% (we don’t need a model for this). We’ll lose the bet the same amount of times that we’ll win it, but when it wins, the reward is more than the amount we lose when it lands “tails.”
Shifting gears to a more useful sports betting example, let’s take a look at the DET@HOU baseball game on July 15 this year. The bottom-feeding Detroit Tigers sent the erratic, past-his-prime Francisco Liriano to the mound against the defending World Series Champion Houston Astros, who were pitching their Cy Young candidate and ace, Justin Verlander. Any reliable and predictive sports betting model that favored Detroit to win the game likely has a massive bug, or was created by someone with an intimate connection to a point-shaver.
My MLB model gave the Tigers just a 32.25% chance of winning the game – extremely low for a sport where the worst teams are consistently at or around .400 winning percentages each season. The sports betting markets, however, gave the Tigers even less of a shot. The line on the game closed with the Tigers moneyline at a price of +351 – implying that they would win the game just 22.17% of the time. Because this discrepancy in probabilities was so large – about 10.08% of what we call “edge” – I bet on the Tigers, fully expecting them to lose the game.
Fortunately, the Tigers did end up cashing a massive ticket for me and those following my model that day, but the result is not the important lesson here. It’s that I bet on a team my model thought had a below 50% chance of winning because it was +EV – NOT because it was a “winner” or a “lock.”
I know that many of my followers understand this point, but every day I am surprised to see some do not, and I cannot emphasize this point enough. Achieving long-run profitability in sports-betting is about accumulating small amounts of value over the course of a long season and career. This means profitable bettors must frequently bet on teams he or she “expects” to lose – teams that have a less than 50% chance of winning – because he or she expects bets like those to accumulate value over a large sample size. If you’ve made it this far, I appreciate you taking the time to read through all of this.
This is probably an overly simple refresher for most of you, but I think it’s important to start with the building-blocks before moving into anything else more advanced. Please reach out with any questions, comments, or suggestions for future newsletters and thanks again for your time and interest!
The article originally appeared in Berry Horse’s free newsletter BetItUp, which you can (and should) subscribe to here to learn more about predictive sports modeling, betting, bankroll management and more. The article is published here at Sports Handle with his permission.