APBRmetrics

<admin delete if this post is inappropriate>

Greetings.

Although this is a first post for me here, I come to you after several weeks of unsuccessful reddit posts, covers posts, support emails to data providers, YouTube clips, etc. This seemed like the appropriate place to ask this question, but I'll let you guys be the judge.

I am building a 4-leg martingale system based on the four quarters available in a matchup. The model looks for confluence in several areas that provide an advantage/disadvantage to a team and picks high probability setups for that team. It has been quite successful, but the problem is... I am not sure if I am just lucky or if this system can withstand the test of time. That's why I want to back-test the strategy as far back as the data will allow me.

The issue is that data providers can't seem to provide me with historical 'In-game' spread data. Is it possible that nobody has a record of this? It seems unlikely that this data is being displayed and then simply destroyed afterwards.

Data Needed:

• At the end of the 1st quarter, what was the spread for the 2nd quarter?

• At the end of the 2nd quarter, what was the spread for the 3rd quarter?

• At the end of the 3rd quarter, what was the spread for the 4th quarter?

...also, is there a way to calculate 1st quarter pre-match spreads based on full-game spreads?

There seems to be only two ways to approach this dilemma.

(1) Continue searching, continue emailing, and continue posting in hopes of someone who has been hoarding this data decides to sell it. Meanwhile, I will gather what I can from my sportsbook.

(2) Post here and hope that someone with knowledge on how these odds are being calculated can enlighten me with a mathematical solution that can be incorporated into the model.

I would be grateful for any type of direction.

At the end of the 3rd quarter, what was the spread for the 4th quarter?
...also, is there a way to calculate 1st quarter pre-match spreads based on full-game spreads?

Can you explain what is meant by "spread"?

Sure thing.

A "spread" in sports-betting terms is the handicap between two teams based on the probability of one team beating the other. Sometimes teams are completely even in strength and a handicap isn't required, but most matchups require a handicap where one team is given points, and the other team has a certain number of points handicapped.

For example in last night's matchup between Golden State Warriors vs. Los Angeles Lakers, the "spread" for Warriors was +6, which automatically means that Lakers were handicapped -6. The odds-makers believed that Lakers were the favorite for the matchup.

In pre-match wagering terms, the formula for coming up with the "spread" is based on several factors: Power-ranking, home court advantage, line-up changes, to name a few. The spread is then subject to market influence based on how much money comes in on either side. If there's too much money coming in on Lakers, then that can push the Lakers odds towards a higher spread. The odds-makers actually opened the lines at Lakers -4.5, but so much money poured into Lakers that the line was pushed up to -6 by tip-off. Inversely, Warriors line become more favorable and provided better odds for bettors which enticed a good amount of money flow to Warriors. Book makers are not interested in who wins or loses. All they're interested in doing is finding the equilibrium in terms of the "spread" so that they can balance out both sides of the wager.

Last night, the score ended at Warriors 100 - 103 Lakers. This means that Warriors +6 won against the spread (aka "ATS")

Now, live odds are a different animal altogether, but the algorithms are almost purely mathematical in nature and have less to do with market influence because there's very little time to move these markets in between quarters or during the match. My quest is to find a "good enough" equation to apply it to data that is already available so that I can back-test a sports wagering model.

I hope this helps, but I get the feeling that I may be informing you on a topic you're not familiar with, and it may be a while before I can recoup some of these investments. I don't mind it, I'm just frustrated with how difficult it is to find data on this topic especially regarding a multi-billion dollar industry.

Sometimes we've had people here discussing gambling, but not very regularly.

Mike G wrote: ↑Thu May 20, 2021 10:43 pm Sometimes we've had people here discussing gambling, but not very regularly.

I completely understand.

If I'm reading this correctly. It sounds like you are using what happens in Q1 to predict the Q2 spread etc.

How is that possible? Q1 is played mostly by the starting lineup. Q2 is played mostly by the bench. How is data from the starting lineup going to help predict how the bench performs in Q2?

I can see an argument for using Q1 data to predict Q3 or late Q4. Since these periods tend to be where the starting lineups matchup. For example, I personally, have an algorithm that uses the first 3 quarters to predict the O/U of Q4's.

But I don't see how you can predict Q2's without looking at previous games. (And thus, trying to be better than the sportsbooks at it. As that is their general strategy at setting the Q2 lines).