How The Betting Line Is Created

The below article was originally published in a TSP Live Insider issue back in January (2023). Given I am always asked about how the line is created, I thought it would be good to post it to the website for future reference. I hope you enjoy!

I was asked yesterday what the process is like to set a line at a sportsbook. So, let me walk you through how it works. Let’s use an upcoming NCAAB game for this weekend and see how close I can get. Louisville got destroyed by Pitt (91-57) on Tuesday as we all know. On Saturday, the Pitt Panthers will play Florida St. What should the line be?

The first thing I do, and what a sportsbook would do, is put both teams into a statistics algorithm. The algorithm basically takes key statistics from both teams and uses those stats to figure out…based only on stats…what the price should be for the game. It usually assumes a neutral court for this first part of the calculation.

The algorithm says that on a neutral court, Pittsburgh should be -6.

Next up, I have to see if these teams already played each other and what was the result. As luck would have it…they have played one another earlier this month (January) and FSU actually won the game by 7 points as a +7 point underdog!

The fact that FSU already beat Pitt…in Pittsburgh…works out to a 2 point advantage for FSU. Some might say this should be like 5 or 7 points, but point spread adjustments are minor…and past performance is not indicative of future results. So, you can’t give it too much weight. Think of when a star QB is out for an NFL team, that’s a big change and often the spread moves 2-5 points. So, the Pittsburgh -6 I calculated in step one above now becomes Pittsburgh -4.

Well, we know the game is going to be played at Florida St. FSU is 5-7 at home and Pittsburgh is 6-2 on the road. So, FSU doesn’t have much of a home court advantage. I’ll give it 1 point.

Pittsburgh is now -3 when adjusted for FSU’s 1 point home court.

Now we have to create a public perception variable. The public perception here is everyone saw Pitt beat UNC and then destroy Louisville. Prior to those two, Pitt beat #20 Miami FL. At the same time, FSU lost to Miami FL (in a blowout), lost to Clemson, lost to NC State (in blowout) and just lost to Syracuse by 9. FSU did beat Louisville, but only by 3. So, the public perception is Pittsburgh is HOT and FSU is not! So, what am I going to add to Pitt given they will be attractive to the public no matter what price I hang?? Why not give FSU 6 extra points if the public will bet anything Pitt is hanging? I can’t give too much because then I make it too easy for sharps to see the value on FSU. So, I will go with 2 points.

I now have the price at Pittsburgh -5 when adjusted for public perception.

So, I used an algorithm to take the stats of both teams to give me an expected point differential. I then adjusted for the outcome of any previous meetings this season. I took into account who had home court advantage for the current game (and how the teams perform at home and on the road), and then I calculated the public’s perception of these two teams and what that means for the point spread (i.e. how much public shade do I need given how hot or cold the respective teams appear). After all that, I came up with the price being Pittsburgh -5 to Florida State (at Florida St)

Some books, especially back in the day would use power ratings to compute the price. Power ratings are simply a means of taking all the stats for a team and quantifying them into a single number. You would then take Team A’s power rating and subtract Team B’s power rating to get the line. Simple! Computers allow us to be much more accurate now and work in a lot of minutiae.

So, for that Saturday in February, I had Pittsburgh -5 on the road at Florida St. The line ended up opening at Bookmaker at -4.5, moved up to -5.5 and closed at -4.5. The result, Pittsburgh won 83-75 and covered the spread by 3.5 points.

The one key to remember is the goal of line creation is actually not to predict the outcome, but to predict a price whereby you are not lopsided in action for one side or the other. Some people will say that Pittsburgh won by 8, so why wasn’t the line Pittsburgh -8?? Well, the book makes no money with pushes…that’s the first part. The second part is sharp money actually took Florida State in the game. So, if sharps liked FSU mildly at +4.5, why would you give them +8 so they could load up the truck?

The line at Pittsburgh -4.5 over FSU (or perhaps even my calculated -5…patting my back) was perfect because it caused action on both sides of the market. Pittsburgh opened -4.5, was pushed up to -5.5 and then settled back down to -4.5 after sharp money grabbed the +5.5. At the end of the day, the price rotated around my calculated -5, and the book took action on both sides…and got a price which resulted in an outcome and not a push. It was great for the book…action was not heavy to one side or the other, and the game did not end perfectly on the point spread.

I love it when some sports announcer will say “the oddsmakers really nailed this one”, when a game lands perfectly on the point spread or total. Every time an announcer says that I want to say…”you fucking moron, you know nothing!” The book does not want to nail the actual point disparity between the teams…they make no money. Instead they spent time, manpower, and had costs booking all that action for the game…only to give ALL the money back to customers. The book wants a price that ends in a conclusion for one side or the other…not a push. Sure, the book would rather push than lose on a game, but the book isn’t in the business of making money on pushes! The game fell on the point spread, not because the oddsmaker wanted it, but because of LUCK!

Why luck? Let’s say the average NBA point differential (the difference in points at the end of the game between two NBA teams playing) is 10 points. Obviously, some teams will win by more than 10 points, some less than 10 points. However, if most NBA games end with a team winning by 1 to 20 points, there’s only 20 possible round number point spreads (-1, -2, -3, -18, -20, etc.). So, if you randomly set a round number point spread on an NBA game of -8, there is a 1 in 20 chance you will select the correct point disparity. Are you a genius, or did you just get lucky on a 1 in 20 probability. Even if you say 100% NBA games will finish with a disparity of between 1 and 50 points, that means you have a 1 in 50 chance of nailing the point spread for a specific game. With their being 2,460 NBA games in a season, there is a good chance, even at a 1 to 50 point disparity, that you will correctly select the point spread of 49 games (2,460 games/50 = 49 games). Are you a genius, or was it just statistics and probabilities?!? Reality is closer to a point disparity of 1 in 20 which means 123 games each season would have you selecting the correct point spread simply picking random numbers. So, no magic, just the law of large numbers.

Hope you enjoyed the article and GOOD LUCK!!