[briedward]

I found a method that is 80% or more accurate.

I go to teamrankings.com and I look up the scoring offense and scoring defense rankings for each team. Then I add them together for the "briedward sum"

Let's take an example of this past week:

Team **********Scoring Defense ******Scoring Offense ******* Briedward sum
Vanderbilt *********22 *****************84 *****************106
Wake Forest *******76 *****************113 ****************189

So who do I bet? I only bet on the team that is ranked better for BOTH scoring offense and defense. The higher the difference in the briedward sum, the more I like it. I've found that my occasional losses typically occur for briedward sum differences of less than 60. Also I don't like teams as much with a briedward sum of greater than 150ish, since the higher the briedward sum, the more garbage a team is.

So in my example, I would bet on Vanderbilt moneyline, which didn't pay much at -475. I only bet moneyline, since this technique tells me who will win, not by how much.

I make money by parlaying a bunch of highly likely winners, say 4 teams whose briedward sum difference is greater than 100. This method only works from about week 7 and forward (not enough stats prior to make it as reliable)

Also using this method, I can sniff out a potential upset.

Again the simplicity (the exception are significant injuries):
1) choose a team that is better ranked in BOTH scoring offense and defense
2) greater confidence is with teams that have a higher briedward sum difference (I like 60 or more)
3) consider upsets for any briedward sum difference (assuming the team is better in BOTH scoring offense and defense rankings)

Apply my example to any 2 opponents, and you will see how freaking easy this is.

I haven't lost any money since I started week 7. Haven't made a killing either. For instance this past week, I woke up late, didn't prepare at all, and put in a ML parlay of Texas A&M, UNC, Utah state, BYU - I bet 4K to make around 700. Easy day (though I was worried with UNC). Previous weeks I've tried all sorts of parlays - grouping them by gametimes, or I would simply bet one team ML.

[kj8210]

where are you getting the numbers for scoring defense and scoring offense? there are so many options....

[guy Fawkes]

I respect your thinking and applaud you for trying to do something a little different, but I don't think you've really thought this through. Firstly, you are claiming 80% success - and I have no doubt you are correct and possibly even underestimating it. You are also claiming a 4000 to win 700 parlay. If you are claiming that 80% of parlays will hit, you're losing an average of 240 every parlay or if you're claiming 80% of individual plays will hit you are losing every other parlay with a 1 in 5 exception.

Okay, so let's say that's an underestimation. Given the 4000 to win 700 on 4 ML parlay, you're assuming about -2500 each...you could give just about any stat on any number of teams around this number and they will all support a strong winning correlation. What I mean by that is you could say anything in this range "Any team that has/does ___________ rating playing against a team that has/does ___________ rating that is being offered at -2500 or higher will win at an 80% rate". In reality, you could probably claim this same statement with a (anything lower than 96%) winning rate and you'd still probably be right.

For your particular statistic, you are also taking numbers that the books will be strongly biased towards or are skewed for various reasons. Using your figure and trying to come up with a 100 rating-point difference (where a team is favored on both offense and defense), you arrive at teams that look like this:

Team A: Offense 10th Defense 10th - Team B: Offense 60th Defense 60th
Team A: Offense 10th Defense 90th - Team B: Offense 109th Defense 91st
Team A: Offense 90th Defense 10th - Team B: Offense 91st Defense 109th

No matter how you slice it you are taking a team with either a huge advantage on one-side of the ball or greatly better on both sides of the ball. Any line offered on any of the three above teams would be a very large favorite, unless the line is skewed due to division, FCS opponents, etc. If you are meaning to indicate that teams with skewed statistics are where your value lies, I'd say you are working backwards.

Again, I applaud you for thinking outside of the box on this one. Unfortunately, I think you may need to fine tune your numbers or clarify exactly how this proves profitable.

Edit: If on the other hand, you could find a correlation between your statistic and something else that provides +EV on plays (for instance a 60-40 record on -110 games) then you would have a very valuable tool.

[easysaid]

Stanford vs oregon i got buried by this method +21.5 favourite loses, why? because their is alot more to capping than just some dumb ranks, motivation, weather, letdown spots.. list goes on and on.

[briedward]

where are you getting the numbers for scoring defense and scoring offense? there are so many options....

I'm sorry - for clarification - points per game

http://www.teamrankings.com/college-...oints-per-game

The number I assign to a team is simply what their ranking is (not their actual points per game)

[briedward]

Stanford vs oregon i got buried by this method +21.5 favourite loses, why? because their is alot more to capping than just some dumb ranks, motivation, weather, letdown spots.. list goes on and on.

Using my method, you would not have bet on Stanford vs Oregon. Neither team is ranked better in BOTH scoring offense points per game and scoring defense points per game.

Better yet, using my method, you would have realized that an upset was possible to consider the underdog.

[briedward]

I respect your thinking and applaud you for trying to do something a little different, but I don't think you've really thought this through. Firstly, you are claiming 80% success - and I have no doubt you are correct and possibly even underestimating it. You are also claiming a 4000 to win 700 parlay. If you are claiming that 80% of parlays will hit, you're losing an average of 240 every parlay or if you're claiming 80% of individual plays will hit you are losing every other parlay with a 1 in 5 exception.

Okay, so let's say that's an underestimation. Given the 4000 to win 700 on 4 ML parlay, you're assuming about -2500 each...you could give just about any stat on any number of teams around this number and they will all support a strong winning correlation. What I mean by that is you could say anything in this range "Any team that has/does ___________ rating playing against a team that has/does ___________ rating that is being offered at -2500 or higher will win at an 80% rate". In reality, you could probably claim this same statement with a (anything lower than 96%) winning rate and you'd still probably be right.

For your particular statistic, you are also taking numbers that the books will be strongly biased towards or are skewed for various reasons. Using your figure and trying to come up with a 100 rating-point difference (where a team is favored on both offense and defense), you arrive at teams that look like this:

Team A: Offense 10th Defense 10th - Team B: Offense 60th Defense 60th
Team A: Offense 10th Defense 90th - Team B: Offense 109th Defense 91st
Team A: Offense 90th Defense 10th - Team B: Offense 91st Defense 109th

No matter how you slice it you are taking a team with either a huge advantage on one-side of the ball or greatly better on both sides of the ball. Any line offered on any of the three above teams would be a very large favorite, unless the line is skewed due to division, FCS opponents, etc. If you are meaning to indicate that teams with skewed statistics are where your value lies, I'd say you are working backwards.

Again, I applaud you for thinking outside of the box on this one. Unfortunately, I think you may need to fine tune your numbers or clarify exactly how this proves profitable.

Edit: If on the other hand, you could find a correlation between your statistic and something else that provides +EV on plays (for instance a 60-40 record on -110 games) then you would have a very valuable tool.

I am not a statistician by any stretch. I try to find the simplest big picture view on accurately picking a winner. Are my results statistically significant? Of course not since my sample size is too small, but so far it's working.

Occasionally there are teams that are better in both statistics where the spread is minimal or underdog status - these can be profitable.

Other times I like to have a good assurance that a team should easily win when the spread is large. It helps me weed out potential upsets.

All I'm saying is try it out - see what you think. I believe you will be pleasantly surprised. Takes all of 20 seconds to pick a winner.

This is how i do it:
I copy/paste scoring defense points per game stats into excel. Then beside it I post the scoring offense points per game. Then I cut out all the columns except for the name of the school and the ranking. Then I make another column to sum up the 2 ranking columns. Then I print.

Then I print the lines from where I'm betting.

Beside each team I write the three numbers as mentioned. I circle the better number between the 2 teams. If all 3 circles line up, I consider a bet. If the briedward sum is greater than 60, I'm more willing to bet. If it's less than 60, I'm more cautious and weigh the risk vs the reward.

Again i like to look at big picture. The bottom line in sports is scoring offense and scoring defense. In college, the talent disparities are magnified enough to where this method works. I haven't tried it in the NFL, but I doubt one would have success evaluating professional teams since the teams are much closer in talent.

[SamDiamond]

I have a question.

How many games a weekend are you getting that match your criteria?

Can you give us the games of this most recent weekend?

[Retrospect]

thanks for sharing.

[turbozed]

Might be a good method to narrow down to potential spots, after which you can do traditional analysis.

[thfootball]

after week 7 this works better you say, not that many weeks after that left. How does this work in the bowls?

[Blazermaniac]

Son.....you raised my eyebrow. Interesting reading. Always good to see new angles. Do more demos...add more filters and keep everyone posted.

[sitzlejd]

That method is similar to the method i use for nfl, which i got from the book "Changing the Game: How to profit from your passion for sports"

What youre talking about is only the first part though. Using rankings to identify your own spreads (as you do) and then once youve identified some spots, perform deeper analysis of those select games. Ive had success with it but it is VERY time consuming. The book is worth a read though.

[Stack-N-Chips]

if capping football games were easy, our wives would be doing it

[kj8210]

Ok so I just happned to pick the first game on the list for this Saturday and it happened to be Cincinnati vs Connecticut. So based on your theory it would look something like this:

Scoring Offense Scoring Defense Total
Connecticut 51 15 66
Cincinnati 115 18 133

So since the differene is 67 which is >60 then this play would fit into your system? Is that right?