If you took the points in ALL of the Bowl Games, how would you fair in units?

[abacus30]

So the next logical question may be, for those of us betting blindly on all dogs before New Year's Day, - do we bet 'em all now or wait and bet them before kickoff of each game? Maybe wait and allow "JQP" more time to bet the chalk, but watch the lines and jump on any dog that starts losing points earlier?

[jackpot269]

After New Year's
38-34-1 (52.78%)

Before New Year's
97-56-1 (63.40%)

ok we now know about before 1/1 & after 1/1 how about the ones played on 1/1?I bet not much change !

[HedgeHog]

ok we now know about before 1/1 & after 1/1 how about the ones played on 1/1?I bet not much change !

I believe the "after" New Years includes Jan 1 games.

[jackpot269]

I believe the "after" New Years includes Jan 1 games.

Thanks was just wondering. Need all the help I can get this year.GL everyone!!

[Dark Horse]

So the next logical question may be, for those of us betting blindly on all dogs before New Year's Day, - do we bet 'em all now or wait and bet them before kickoff of each game? Maybe wait and allow "JQP" more time to bet the chalk, but watch the lines and jump on any dog that starts losing points earlier?

Typically, bet favorites early and dogs late.

[hhsilver]

If you bet the before 1/1 dogs, I hope you have the guts to stick with it. Last year looked like a year this would not hold up. After the first 8 games, dogs were 2-6 with only Troy and San Jose covering.

[Ganchrow]

This could be just a case of a freak 9 year win streak for dogs.

You're 100% correct. It certainly *could* be. We need to look at the standard error of the measured statistics to determine how likely that might be.

For bowl games commencing prior to New Year's Eve, the dogs have a records of 97 wins and 56 losses, which is 63.40%. This means that the standard error is then 3.92%. 63.40% is then 3.42 standard errors above 50%. Performing a 2-tailed z-test (2-tailed in that we'd have been equally interested had faves outperformed to the same extent as dogs) we get a p-value of 0.0631%.

This means that (if this were the only "trend" we were examining) there is about a 0.0631% probability (the exact figure is from the binomial distribution, which is 0.1150%) that this data is representative of but "a freak 9 year win streak".

Now what does this suggest about the probability of this continuing into the future? Absolutely nothing whatsoever.

I think I will play the dog theory for a small amount each game.

Just to be clear on terminology, what I've presented isn't a theory -- rather it's experimental data in search of a theory. That over the past 9 years there is only a 0.1150% probability of dog performance having been other than the result of a fundamental misvaluation of dog prices is "fact" and not "theory" (assuming this were the only "trend" we were examining).

Why that might have been true is up to theory (not presented here) to describe.

Similarly, what we can expect to observe in the future based on said probabilistic fact would also be up to theory (again not presented here either) to predict.

When betting on dogs I like to bet 2X on the spread and X on the Moneyline.

In general, this is probably not a good idea. That said, if you have any evidence whatsoever to backup your assertion that this is a sensible strategy, I'd certainly love to see it.

[Ganchrow]

How much longer you think this is going to last?

I have no view on that.

Do I expect this inefficiency is going to continue forever? No. Markets do tend to eventually correct themselves.

to have something be 3 SDs away from the mean is extraordinarily high for a 50/50 proposition outcome.

From the standpoint of market efficiency you're obviously correct.

I will note that the data suggests that over the in-sample period dog victory was likely not a 50/50 event. If the structural issues that caused this dynamic are for whatever exogenous reason not corrected and continue unabated into the future then we'd expect outperformance (in units of standard dev) to increase unboundedly.

The way I would interpret this info is that the bowl lines are shaded heavily toward the dog more than anything the dog is doing.

I'd also point out that the Pope is Catholic.

[Ganchrow]

What's the breakdown by lines? Curious as to how Bowl Dogs +7 and higher do.

I posted the complete data set here, but to save you a little leg-work:

<hr>
<style> .GANCH { font-size: 10px; font-family: Verdana, Arial, Helvetica, sans-serif; }</style> <table border=1 cellpadding=2 cellspacing=2> <tr> <td align=left class=GANCH> Description </td> <td align=left class=GANCH> Dog Wins </td> <td align=left class=GANCH> Dog Losses </td> <td align=left class=GANCH> Freq </td> <td align=left class=GANCH> Std. Err. </td> <td align=left class=GANCH> Z(50%) </td> <td align=left class=GANCH> Z(52.381%) </td> <td align=left class=GANCH> p-value(52.381%)
</td> </tr> <tr> <td align=left class=GANCH> Before New Year's </td> <td align=center class=GANCH> 97 </td> <td align=center class=GANCH> 56 </td> <td align=right class=GANCH> 63.40% </td> <td align=right class=GANCH> 3.92% </td> <td align=right class=GANCH> 3.42 </td> <td align=right class=GANCH> 2.81 </td> <td align=right class=GANCH> 0.49%
</td> </tr> <tr> <td align=left class=GANCH> After New Year's </td> <td align=center class=GANCH> 38 </td> <td align=center class=GANCH> 34 </td> <td align=right class=GANCH> 52.78% </td> <td align=right class=GANCH> 5.97% </td> <td align=right class=GANCH> 0.47 </td> <td align=right class=GANCH> 0.07 </td> <td align=right class=GANCH> 94.70%
</td> </tr> <tr> <td align=left class=GANCH colspan=8> </td> </tr> <tr> <td align=left class=GANCH> spread ≥ 7 </td> <td align=center class=GANCH> 57 </td> <td align=center class=GANCH> 27 </td> <td align=right class=GANCH> 67.86% </td> <td align=right class=GANCH> 5.16% </td> <td align=right class=GANCH> 3.46 </td> <td align=right class=GANCH> 3.00 </td> <td align=right class=GANCH> 0.27%
</td> </tr> <tr> <td align=left class=GANCH> spread < 7 </td> <td align=center class=GANCH> 78 </td> <td align=center class=GANCH> 63 </td> <td align=right class=GANCH> 55.32% </td> <td align=right class=GANCH> 4.22% </td> <td align=right class=GANCH> 1.26 </td> <td align=right class=GANCH> 0.70 </td> <td align=right class=GANCH> 48.59%
</td> </tr> <tr> <td align=left class=GANCH colspan=8> </td> </tr> <tr> <td align=left class=GANCH> spr ≥ 7,Before New Year's </td> <td align=center class=GANCH> 38 </td> <td align=center class=GANCH> 13 </td> <td align=right class=GANCH> 74.51% </td> <td align=right class=GANCH> 6.22% </td> <td align=right class=GANCH> 3.94 </td> <td align=right class=GANCH> 3.56 </td> <td align=right class=GANCH> 0.04%
</td> </tr> <tr> <td align=left class=GANCH> spr < 7,Before New Year's </td> <td align=center class=GANCH> 59 </td> <td align=center class=GANCH> 43 </td> <td align=right class=GANCH> 57.84% </td> <td align=right class=GANCH> 4.94% </td> <td align=right class=GANCH> 1.59 </td> <td align=right class=GANCH> 1.11 </td> <td align=right class=GANCH> 26.86%
</td> </tr> <tr> <td align=left class=GANCH> spr ≥ 7,After New Year's </td> <td align=center class=GANCH> 19 </td> <td align=center class=GANCH> 14 </td> <td align=right class=GANCH> 57.58% </td> <td align=right class=GANCH> 8.87% </td> <td align=right class=GANCH> 0.85 </td> <td align=right class=GANCH> 0.59 </td> <td align=right class=GANCH> 55.82%
</td> </tr> <tr> <td align=left class=GANCH> spr < 7,After New Year's </td> <td align=center class=GANCH> 19 </td> <td align=center class=GANCH> 20 </td> <td align=right class=GANCH> 48.72% </td> <td align=right class=GANCH> 8.21% </td> <td align=right class=GANCH> -0.16 </td> <td align=right class=GANCH> -0.45 </td> <td align=right class=GANCH> 65.57%
</td> </tr> <tr> <td align=left class=GANCH colspan=8> </td> </tr> <tr> <td align=left class=GANCH> ALL </td> <td align=center class=GANCH> 135 </td> <td align=center class=GANCH> 90 </td> <td align=right class=GANCH> 60.00% </td> <td align=right class=GANCH> 3.28% </td> <td align=right class=GANCH> 3.05 </td> <td align=right class=GANCH> 2.32 </td> <td align=right class=GANCH> 2.02% </td> </tr> </table>

[HedgeHog]

Wow, look at Dogs +7 and higher before 1/1: 38-13 (74.51%) nearly four standard deviations to the good. I just got the biggest stiffy ever. After I measure, I'll post how many standard deviations above the norm I am.

[TLD]

“That over the past 9 years there is only a 0.1150% probability of dog performance having been other than the result of a fundamental misvaluation of dog prices is ‘fact’ and not ‘theory’.”

Doesn’t this need to be qualified with some sort of “assuming no data mining” clause? (Or is that covered by your “if this were the only ‘trend’ we were examining” in an earlier paragraph?) I would think there are plenty of trends hitting better than 63.4% that have nothing to do with a fundamental misvaluation that the market needs to correct for.

[Ganchrow]

“That over the past 9 years there is only a 0.1150% probability of dog performance having been other than the result of a fundamental misvaluation of dog prices is ‘fact’ and not ‘theory’.”

Doesn’t this need to be qualified with some sort of “assuming no data mining” clause? (Or is that covered by your “if this were the only ‘trend’ we were examining” in an earlier paragraph?) I would think there are plenty of trends hitting better than 63.4% that have nothing to do with a fundamental misvaluation that the market needs to correct for.

It's not the 63.40% figure that's by itself relevant, rather it's the Z-score of 2.81.

And yes, the assumption that this were the only "trend" we were examining (which for the sake of clarity I just edited into the above post a second time) was intended to allow us to eliminate the possibility of data mining from our analysis.

[eric dy]

I did this a few years back and broke about even......nothing special for me.

[HedgeHog]

Best Bets Given this info:

Navy +8.5
S Miss +11
E Car +10.5
C Mich +8.5

All are getting 7+ points and playing prior to Jan1st.

spr ≥ 7,Before New Year's 38 -13 74.51% (Dogs)

[BandwagonFan]

Since we are all talking about dogs doing well ATS, does anyone have records for dogs straight up. It seems like if one were to bet blind on every dog ATS at -110 or -105, you might as well bet them blind SU at +money.

[hhsilver]

So the next logical question may be, for those of us betting blindly on all dogs before New Year's Day, - do we bet 'em all now or wait and bet them before kickoff of each game? Maybe wait and allow "JQP" more time to bet the chalk, but watch the lines and jump on any dog that starts losing points earlier?

someone answered the above with this >>>>"Typically, bet favorites early and dogs late."

but how about the fact that more and more people are realizing dogs have done well in recent years? A quick glance at the lines so far shows that more numbers have dropped than risen to this point. Any data on line movements for bowl games with, say, opening lines > 3?

[seaborneq]

I just entered the Bodog contest with this concept. Let's see how it goes.

[HedgeHog]

someone answered the above with this >>>>"Typically, bet favorites early and dogs late."

but how about the fact that more and more people are realizing dogs have done well in recent years? A quick glance at the lines so far shows that more numbers have dropped than risen to this point. Any data on line movements for bowl games with, say, opening lines > 3?

General, and useless, info (locking in Favs early) unless you're hoping for a small middle. Why bet the side you don't like first (the Fav)? Remember the Kiss formula...Keep It Simple Stupid. Bet the Big Dogs (+7 and up) prior to 1/1. Solid Angle.

[seaborneq]

Time to test the theory.

[Doug]

someone answered the above with this >>>>"Typically, bet favorites early and dogs late."

but how about the fact that more and more people are realizing dogs have done well in recent years? A quick glance at the lines so far shows that more numbers have dropped than risen to this point. Any data on line movements for bowl games with, say, opening lines > 3?

This might be like betting #12 seed ML to beat #5 in March.

over-rated, and no longer value.

[pharmorjac]

Glad to see this thread is still alive, i'm definetly interested to see if it pans out.

Either way I put no money on it though....

[wtt0315]

its really simple math. Just take the dogs that win and take the favs that win then you will be 100 % just a little joke

[20Four7]

its really simple math. Just take the dogs that win and take the favs that win then you will be 100 % just a little joke

I didn't realize it was that easy..... I"m going to try that and see how I do...

[seaborneq]

1 down 31 to go.

[seaborneq]

60% through the first 5 games(3-2), up 1 unit with dogs.

[HedgeHog]

Best Bets Given this info:

Navy +8.5
S Miss +11
E Car +10.5
C Mich +8.5

All are getting 7+ points and playing prior to Jan1st.

spr ≥ 7,Before New Year's 38 -13 74.51% (Dogs)

This trend is now 41-13 as Navy, S Miss and E Car all won with the points (EC outright). I'm bumping this thread as a reminder to consider C Mich tonight.

[20Four7]

I cashed on all 3 HH. I"m considering passing tonight as a 3 and 0 record is good enough for me.

[HedgeHog]

20-4-7:

I'm playing CM for just as much as the other three. If I'm wrong, then I'll happily live with 3-1. And thanks to Ganch for exposing this angle!

[20Four7]

20-4-7:

I'm playing CM for just as much as the other three. If I'm wrong, then I'll happily live with 3-1. And thanks to Ganch for exposing this angle!

I ended up doing the same. Although towards the end of the first half I was wishing otherwise. Once it was tied up again I felt better. I certainly am putting this angle down in the book.

I need to go back through this thread and find out where New Years day falls into the category.

[20Four7]

Ganch, is there an easy way to data mine the info at covers? I've gone there and there doesn't seem to be an easy way to get at old info. Am I missing something. BTW thanks for the insite as I did hit up the pre NYD dogs so far.

[Ganchrow]

Ganch, is there an easy way to data mine the info at covers? I've gone there and there doesn't seem to be an easy way to get at old info. Am I missing something. BTW thanks for the insite as I did hit up the pre NYD dogs so far.

I wrote an HTML scraper in Perl. I certainly wouldn't want to do it by hand.

I've updated the data in this post to be current up to 12/28/2007.

[20Four7]

That's what I figured you did. I haven't done any programming since my thesis in University. I guess it's time to learn this stuff to help utilize covers to it's fullest.

Thanks again.

[seaborneq]

An initial look, it seems as though you would hit well over 60% of the games.

Well maybe about 60%. I have counted 15 dogs that have covered with 2 games left.

[pokernut9999]

Well maybe about 60%. I have counted 15 dogs that have covered with 2 games left.

Favorites are 15-14 and then the Kentucky game that depends when you got down on that game.

[Actionbrett]

I'm good with numbers/and math but what's the definition of these symbols and what do the %'s relate too?

Z(50%) Z(52.381%) p-value(52.38%)

What year was the starting date for the records compiled above?