Pirateboy's NCAAF ML Dogs Experiment - W9

pirateboy · #1 10-25-2007, 02:03 PM

Alright friends, I'm doing it. Diving in headfirst. Added bankroll to the ol' account to make it possible. Over at sbrforum, they had a breakdown of moneyline underdogs and their records broken out in each spread class and compared winning percentages. The data is from 1985-2007, and I will be updating it every week. They also included the "BE," or break even, odds you'd need to ... yes, break eve.

So, after talking to MT2R via PM, I decided that for every game where I find even the smallest edge, I'm putting a unit on it. He advised I wait until the latest possible time to place the bet, due to the theory that we'd get the best information throughout the week, and a more correct ML. He advised I do kelly betting for this, but because I do happen to like action, I'll play every edge. So, the schedule each week should be:

For Thursday games: make wager at 4pm PST on Thu
For Friday games: make wager at 4pm PST on Fri
For Saturday games: make wager either late Friday night or early Saturday morning

I happen to drink a bit almost every Friday night, so I'll have flexibility here. I'll post the plays for Week 9 in this thread, and start a new one for Week 10 next week. If a mod would rather me keep them all in one thread, let me know. Oh, and all odds from Bookmaker.

So, even though I'm not placing the wager for another 4 hours, here's an example for tonight.

<u>Dog (Spread) - Win % - ML (BE) Edge</u>
Boston College (+3) - 37.54% - +135 (+166) -31

Air Force (+6) - 37.54% - +205 (+166) +39

So, by using this procedure, I'd place 1 unit on the Air Force moneyline, and nothing on the BC moneyline. As was pointed out before, there can be a flaw here, because, is a 3 pt dog the same as a 6 pt dog? No. It's a flaw. However, I truly believe this is going to be profitable longterm, but I'm doing the experiment to find out.

I'll check back at 4pm PST to make the official wager.

knicknut · #2 10-25-2007, 02:40 PM

So your system is just to bet every ML > +166?

I'm confused.

Jazzy3113 · #3 10-25-2007, 02:54 PM

[ QUOTE ]

I'm confused.

[/ QUOTE ]

Austiger · #4 10-25-2007, 03:03 PM

Pirate- I think you should either compile the data yourself somehow (I know that's a tall order) or send a message to the guy who wrote the article to get his data. If you consider +166 as BE for all the teams between +3 and +6.5, you're going to be betting all of the +6.5 games and none of the +3 games. Or...at the very least estimate what the BE points are for each spread.

Austiger · #5 10-25-2007, 03:07 PM

Also, I mentioned this before but... Why are you restricting the experiment to dogs only? (I would pose the same question to the author of the article.) If +166 is the BE point, you should be betting any favorite that is below -166. Right?

FWIW, I have found that dog MLs get worse as the week goes on. That may just be a sample size thing with the ones I have looked at, but it is something I have noticed. You should really track that as well.

CaptainHook · #6 10-25-2007, 03:19 PM

4th(?)ing the shock of acting as if +6.5 dogs have the same chance of winning as +3 dogs. I really hope this is some sort of error in my understanding/comprehension and not your actual betting method, for your sake.

NajdorfDefense · #7 10-25-2007, 03:53 PM

[ QUOTE ]
[ QUOTE ]

I'm confused.

[/ QUOTE ]

[/ QUOTE ]

pirateboy · #8 10-25-2007, 04:02 PM

[ QUOTE ]
So your system is just to bet every ML > +166?

I'm confused.

[/ QUOTE ]

No, it's just coincidence that the 2 dogs tonight fall in the same point spread range in the table created by sbrforum.

For instance, a 10 point home dog has a different win % historically than a 3-6.5 point away dog.

pirateboy · #9 10-25-2007, 04:08 PM

Here is an example to help:

Utah State is a 2.5 point underdog at home. Historically, home teams getting up to 2.5 points win 48.2% of the time, thus you'd need +108 odds to break even on the ML. You can get Utah State at +120 right now on Bookmaker, giving you a positive EV on the play.

SMU is a 13.5 road dog, and road dogs of 10-13.5 have a 22.2% win rate, so you'd need +351 on the ML. SMU is at +425 at Bookmaker, so nice edge there.

An example of a negative EV play would be UTEP at home as a 4 point dog. They win at 37.1%, meaning you need +170 to break even. UTEP is at +150, so it's a negative expectation.

Hope that helps.

CTrayne · #10 10-25-2007, 04:20 PM

I started something similar last week. Probably missed a few games. Went 3-5, +3.9U thanks to Pitt and Stanford winning. Missed one by 2 pts and one by 3 pts. 8 games is nothing to judge by but I'm going to stay on it this week at least. I'm intrigued so keep us updated [img]/images/graemlins/tongue.gif[/img]