bazooooka
bazooooka
Joined: Nov 21, 2016
  • Threads: 3
  • Posts: 182
May 9th, 2017 at 3:25:19 PM permalink
Joel,

I'd assume your system and the bookmakers choose the same favorite over 80% (90%?) the time?

Given that; how is it possible your seeing 78% (even 60% would seem improbable) in the -1 to -9.5 games when this range covers half the D1 games (around 800)? Not to mention favorites only covered around 47% for the 2016 across all D1 games.

Your seemingly saying had you just bet your favorites (or even the Vegas favs since there will be huge overlap with your picks) in all games between +16 to -16 you'd have scored huge.

But surely as a seasoned gambler you know that this is false? What am I missing?





Quote: JoelDeze

I finished the calculations for the ATS results over the past two years. I analyzed and extrapolated the data, trying to find a correlation between my system picks and Vegas oddsmaker data. I currently pull oddsmaker data from every Vegas and every offshore book for every week of seasonal play. The goal is not to prove or disprove anything. The overall goal is to find an edge within the results that provides inherent value.

I personally did not do well last year. During previous years, I focused on ML values and was doing fine. However, the last two years I've been trying to shift to an ATS only format as the return value is greater. The results below show a great deal of promise for 2017. The data below is accurate.



In the topmost case, if the spread was an oddsmaker favorite between -31 and -37.5 and it was a also a system favorite on my site, picking the opponent to cover produced a 73.33% winning pct.

In the second and third cases, if the spread was an oddsmaker favorite between the ranges posted above and it was also a system favorite on my site, picking the favorite to cover produced 72.38% and 78.31% respectively. The majority of games were posted in the -1 to -9.5 bracket.

In the final case, if the spread was an oddsmaker underdog between +24 and +35 and the underdog was considered a system favorite on my site (contradicting the oddsmaker), picking the underdog to cover produced a 77.78% result.

The overall number of games for the past two years that fit the criteria was 548 games. 412 won, 125 lost, 11 pushed. The overall winpct was 76.72%. If using a $100 game wager per game as a flat bet per game, discounting the pushes (since no money would be won or lost), $53,700 would have been wagered. The total gains would be $37,080 and the total losses would be $12,500. The net gains would be $24,580.

The goal this year is to use a similar wagering structure and ATS criteria. However, I'm still working through all of the losses to find any correlating trends that worked against the system. My goal is to produce a 3-5% additive value to the current results.

The ATS calculations for all other markers showed little to no promise as providing an edge when comparing them against both systems. Therefore, those ATS markers are not posted.

These are for College Only. I'll post the NFL results by next weekend.

JoelDeze
JoelDeze
Joined: Apr 20, 2016
  • Threads: 47
  • Posts: 386
May 10th, 2017 at 3:00:38 AM permalink
I was describing the breakdown example of the ML AI, not the ATS. 78% is the base probability for all ML scenarios the AI resolves.

ATS is much more complicated. What I want to see when I look at ATS is the following:

Breakdown comparison for each team by scenario.

If Team A is an power 5 team and an oddsmaker favorite playing at home against another power 5 team. If the ATS is -19.5 as an example -

What is the probability for Team A as a favorite?
What is the probability for Team A as a home team?
What is the probability for Team A against a P5 opponent?
What is the probability for Team A where a cover scenario is -19.5 or greater?
What is the historical cover scenario probability for a P5 vs P5 playing at home as a -19.5 favorite?

The same probability is shown for the opponent as well.

Additionally, I want to see the points avg for each scenario above or below the cover margin for each team.

Now I have a basic understanding of which team has shown a true advantage in this match-up.

Lastly, line movement has to be factored in as well.

Will this relationship help me win more on the ATS? It will be tested and adjusted. I won't assume anything at this point. Neither should you.
“Know where to find information and how to use it; that is the secret of success.” – Albert Einstein
bazooooka
bazooooka
Joined: Nov 21, 2016
  • Threads: 3
  • Posts: 182
May 10th, 2017 at 3:49:28 AM permalink
Joel,

I get your logic on the ML betting scenarios and if you can really pick near 80%, and get lines better than -350, then you can make a few nickles.

However, you clearly talked about ATS "cover" in the post (see below). Have you found correlations between your system picks and line covers?

I ask since picking winners in NCAA football is often easy. But picking ATS winners and beating the vig is a whole different story.

What has your backtests shown you? Seems like all the data work you do wouldn't be worth it just for betting ML favs whenever you find line value. Now, if you can pick ATS winners at 55-60% then you'd really have something. Historically how well have you done and are you going to post/track your picks going forward?


>>
>>
In the topmost case, if the spread was an oddsmaker favorite between -31 and -37.5 and it was a also a system favorite on my site, picking the opponent to ""cover"" produced a 73.33% winning pct.

In the second and third cases, if the spread was an oddsmaker favorite between the ranges posted above and it was also a system favorite on my site, picking the favorite to ""cover"" produced 72.38% and 78.31% respectively. The majority of games were posted in the -1 to -9.5 bracket.

In the final case, if the spread was an oddsmaker underdog between +24 and +35 and the underdog was considered a system favorite on my site (contradicting the oddsmaker), picking the underdog to ""cover"" produced a 77.78% result.

  • Jump to: