The Wiz himself has shown the ability to approach 60% when he is allowed to use 'stale' lines, which of course no real bookmaker would permit.

If it behooves you, Joel, I am interested in the 'why' you select a certain team. An example like.... "When Florida teams play in any state bordering Canada they only beat the spread 30% of the time, so I am taking Minnesota over Florida this Saturday". Or..... teams that lost the national championship game the previous year are 12-2 if they open on the road, so I'm taking Georgia over Tennessee...."

Quote:SOOPOO

If it behooves you, Joel, I am interested in the 'why' you select a certain team. An example like.... "When Florida teams play in any state bordering Canada they only beat the spread 30% of the time, so I am taking Minnesota over Florida this Saturday". Or..... teams that lost the national championship game the previous year are 12-2 if they open on the road, so I'm taking Georgia over Tennessee...."

It would take a very large article to go over this in detail. I can try to provide a short summary but I tend to be long-winded. I use the following algorithms:

TSRS (True Statistical Rating of Strength is an algorithm I built back in 2007/2008). I determined that if I wanted to create a process that could be utilized across all stats, it would be easier to convert those raw stats into floating point values. I wrote a formula to handle stat inflation / deflation based on the competition value for the opposition category. If you have a strong passing offense but you faced many weak passing defenses, the overall strength rating will lower. Likewise, if you faced strong passing defenses, the overall strength rating will increase. The conversions can be used similar to a transforms table without affecting the actual statistics. I use standard deviation, covariance and correlation. Last year, before bowl season, the top 5 teams in TSRS for College were: 1) Alabama, 2) Clemson, 3) Ohio State, 4) Texas A&M and 5) LSU. Alabama and Clemson had quite a bit of separation from other teams. The goal of this algorithm was to determine the strongest statistical teams on paper.-

H2H (Head to Head Scoring Algorithm) is used to determine percentage of scoring for each team based on all positive scoring indicators, which are then mitigated by indicators that reduce scoring. Instead of saying a team has a 53% chance of winning versus 47%, this algorithm will show that a team "should" have 53% of all scoring in the game versus 47% for the opponent. I can then do a few things when doing historical regression research. I can measure the overall win/loss probability and I can measure the same win/loss probability for each percentage bracket. This allows me to determine what % scoring bracket offers the best probability. -

Riskmark (measures how much risk is in H2H) was built to identify any issues with the H2H scoring algorithm. I wanted to know how likely it was for the scoring percentage to be wrong. I can also measure this algorithm on win/loss probability over time. For instance, the probability curve is high on the win side when the Riskmark shows 19% or less. -

I use TSRSD (strength diffs between teams), Pass efficiency rating difference, Rush efficiency rating difference, scoring difference and a lot of other comparison factors. -

Historical Comparisons can be made so that I can find any games where Team A and Team B closely resemble the teams in a previously played game. I can then determine outcomes based on similarities. -

I've built true homefield advantage charts that take into account stadium capacity, avg. attendance, noise levels, win probabilities, turf type and other environmental factors. -

Recently, over the last two years, I started measuring ATS for every type of outcome in college football. Examples are home, away, neutral, vs. Power 5, vs. Group of 5, favored, underdog, etc. I also calculate the average (per scenario) for how many points over or under a cover in that scenario. As an example, the #1 team in college for ATS in 2018 was FIU. The #2 team was Army. #3 was Iowa. #4 was Syracuse. #5 was UCF.

Other than above, I dive into player personnel and match-ups, especially on the line and in the trenches. I look at player injuries, coaching tendencies , penalties, time of possession, turnover margin and line movements.

One of my biggest problems was trying to research too many games. It's actually more efficient to research games/teams I've watched throughout the season and concentrate on 5 or less games, rather than trying to focus on 20 for the week. That's my goal this year.

I'm always trying to improve upon research.

Quote:gordonm888Okay, so you're a big-time analytics geek. But how do you pick a game like Miami/Florida in the first week of the season when there are so many players starting for the first time?

More importantly, how can you know there will be an advantage against the lines at some point in the future?

Quote:gordonm888Okay, so you're a big-time analytics geek. But how do you pick a game like Miami/Florida in the first week of the season when there are so many players starting for the first time?

I look at the returning starters for each team and the playing time of reserves from the previous year. Miami for instance will be starting a QB that was third on the depth chart and a freshman. They are sitting Perry and Tate. Williams has only played in one game as a redshirt last year and he went 1 for 3. So, he’s essentially a freshman QB behind an inexperienced line going against a Grantham run defense that will blitz and send one extra man on every single play. His head will be on a swivel the whole game. I’ll go into it a bit next week.

Quote:RigondeauxMore importantly, how can you know there will be an advantage against the lines at some point in the future?

One example is when I forecast Team A to beat Team B and the Vegas oddsmakers have Team B as a 17 point favorite. I will be betting on Team A.

I’m not going to go much deeper than that. There are too many instances.

Lines are influenced by sharp action, bettor profiles, vig and action taken by the bookmaker just to name a few. It’s not infallible if you know what you are looking for.

Quote:JoelDezeOne example is when I forecast Team A to beat Team B and the Vegas oddsmakers have Team B as a 17 point favorite. I will be betting on Team A.

Has that situation ever actually occurred, or is that a silly hypothetical? Has there ever been a 17 point underdog that you thought had a better than even chance to win? That seems unbelievable, meaning NOT believable!

Others may chime in with an accurate percentage, but I'd guess 17 point dogs maybe win 7-13% of the time?

Quote:SOOPOOHas that situation ever actually occurred, or is that a silly hypothetical? Has there ever been a 17 point underdog that you thought had a better than even chance to win? That seems unbelievable, meaning NOT believable!

Others may chime in with an accurate percentage, but I'd guess 17 point dogs maybe win 7-13% of the time?

Yes. My forecast had LA Monroe bearing Arkansas in 2012 and I believe they over hyped them, ranking them #8 at the time. I knew it was wrong. Arkansas was favored by 21 and lost.

Quote:JoelDezeYes. My forecast had LA Monroe bearing Arkansas in 2012 and I believe they over hyped them, ranking them #8 at the time. I knew it was wrong. Arkansas was favored by 21 and lost.

In a scenario like that will you be betting them on the money line as opposed to the point spread?

Quote:DRichIn a scenario like that will you be betting them on the money line as opposed to the point spread?

No, ATS only in that scenario.