[Mudcat]

I don't know how much anyone besides me cares about this stuff but it's been an interesting season for my ncaaf moneyline plays.

As I've mentioned before, I bet all ncaaf ML's of +200 or more (up to about +750) from week 3 on. I also use similar rules in certain situations in other sports. (Line shopping is absolutely essential.)

The ncaaf bets were freakishly profitable last year. So anyway this year started out like this:

week 3.....2-15, -10.76 units
week 4.....3-13, -3.87 units
week 5.....3-17, -6.3 units

I was starting to question everything from my past record-keeping to my sanity to my purpose for living. And I strongly considered packing it in. Even though my past results were undeniable I just thought that something had somehow gone haywire in the universe and the old rules no longer applied.

Long story short:

week 6.....9-10, +20 units.

So I'm down a little less than a unit on this system for the year.

I just wish I had've bet SMU +1300 on Saturday (not to mention South Florida +1200 in week 4). But I'm still compiling research on those more lop-sided dogs.

Anyway the moral of the story is that, if you're going to bet a lot of moneyline dogs, you better have the fortitude to deal with some serious losing streaks. But I'm telling you, it can be remarkably profitable long-term (in certain specific situations).

The other moral of the story might be that I should wait longer into the season to start with these ncaaf bets. I'm not sure about that one though. I still need to dig up the old data.

I'd be interested in any thoughts (if not actual data) on that.

[Ganchrow]

Sounds interesting. You know I love those stats. At what years have you looked so far?

One word of caution: For fixed staking plans when dealing exclusively with long odds, higher return volatility is to be expected. Therefore, for any given success ratio, in the longer odds scenario one needs a larger sample size for significance than in the shorter odds scenario.

[pags11]

nice job mudcat...

[jjgold]

I believe the only way to win in sports betting is just playing trends and throw out trying to handicap as it is useless.

Follow a pattern and bet the pattern no matter what

Good Stuff Cat

[Ganchrow]

With regards to this I'd love to locate some historical NCAA Football money line data.

Any ideas?

[Mudcat]

With regards to this I'd love to locate some historical NCAA Football money line data.

Any ideas?

Not really. I'd be interested to see if anyone else has ideas on that.

What I do is in real time where I am comparing live lines at as close to an optimum time as possible (for ncaaf, this seems to be about half an hour before kickoff - although I'm a busy man and I usually just end up placing all my bets and gathering all additional data for the whole day at about 11:30 am - noon on Saturday.) Weeknight games I can be more rigorous about.

With some kind of database of historical odds there are always many questions. Are they opening lines or closing lines (or somewhere in between)? Are they just from a single book and if so, which one? Those things can make a staggering amount of difference to the results.

Don't get me wrong, there could still be a lot to be learned from something like that.

Another possibility which I have seen done is to translate pointspreads/totals to ML's. I know lots of historical pointspread/totals data is out there. Like if you saw a game that went off at +10 with a O/U of 45, you would just call it an average number like +500 (or whatever it works out to).

It would be a bit of work - although not too outrageous - to throw together a conversion chart with average values. This too would not be nearly as accurate as compiling the data the way I do but there could still be value to it.

[Santo]

The data is all available at Don Best...

I've long considered putting up a "Data Entry" project at rentacoder -- to be done either via scripting or manually to compile a database "on the cheap" in terms of my labour..

Of course if anyone's seen one I'd also be interested.

[Ganchrow]

The data is all available at Don Best...

Just sides and totals. No money lines as far as I can tell.

[Ganchrow]

Another possibility which I have seen done is to translate pointspreads/totals to ML's. I know lots of historical pointspread/totals data is out there. Like if you saw a game that went off at +10 with a O/U of 45, you would just call it an average number like +500 (or whatever it works out to).

It would be a bit of work - although not too outrageous - to throw together a conversion chart with average values. This too would not be nearly as accurate as compiling the data the way I do but there could still be value to it.

Only problem here is that if one puts together said chart based on current ML/Side equivalences in order to extrapolate the available historical sides data backwards, how does one know that the approximations held equally at the dawn of the data epoch? This potentially could be a large source of bias.

You know, there has got to be some source of historcial ML data out there and I aim to find it.

[Mudcat]

Only problem here is that if one puts together said chart based on current ML/Side equivalences in order to extrapolate the available historical sides data backwards, how does one know that the approximations held equally at the dawn of the data epoch? This potentially could be a large source of bias.

This is very true. I would only expect to get very general answers from this type of research and would not actually start pulling the trigger on real bets because of it.

But maybe it could at least provide some worthwhile information on if lop-sided upsets become more frequent later in the season and therefore maybe I should wait until later than week 3 to bet this ncaaf ML system.

[Spektre]

Reviving a very old thread. Any luck on this?