Anyone like AFC -6? for Superbowl? (thread from '07)

[imgv94]

Looks really good 2 me

[LT Profits]

Pass. Saints +6 might not be bad, but I will wait for actual match-up.

[Korchnoi]

This is my technical/quant/scalper non-fundamental/handycaper analysis:

I think Colts are roughly 39% and Pats are roughly 26.5%, so that's 65.5% for AFC to win. All books have roughly these odds.

According to Wong -3 is equivalent to roughly 61.5% (or a -160 ML). Moving -3 to -6 is worth roughly 11.5%. So betting AFC -6 is roughly equivalent to betting them at a ML implying 73% which is much higher than their actual probability of winning which is 65.5%.

Ipso facto, the bet is crap.

But I know absolutely nothing about specific matchups, etc. so this analysis might be overly simplistic.

Thoughts?

[gummo]

-5 at 5dimes. I like it.

[Korchnoi]

-5 at 5dimes. I like it.

is like 70.5%

[Razz]

If you play this, you better root your ass off for the Bears this week.

[Ganchrow]

This is my technical/quant/scalper non-fundamental/handycaper analysis:

I think Colts are roughly 39% and Pats are roughly 26.5%, so that's 65.5% for AFC to win. All books have roughly these odds.

According to Wong -3 is equivalent to roughly 61.5% (or a -160 ML). Moving -3 to -6 is worth roughly 11.5%. So betting AFC -6 is roughly equivalent to betting them at a ML implying 73% which is much higher than their actual probability of winning which is 65.5%.

Ipso facto, the bet is crap.

Excellent analysis. I'd certainly have to agree with your conclusion.

Nevertheless, I do have one problem with the specifics of your methodology. Namely, you're implictly assuming that the transformation of win probability to point spread would be distributive over composite probabilities. This does not have to hold in the general case.

The way I might go about it would be as follows:

From Pinnacle:

Prob of Winning Super Bowl:
CHI	18.06%
IND	38.86%
NE	26.97%
NO	16.11%

Prob of Winning Championship Game:
CHI	55.23%
IND	61.50%
NE	38.50%
NO	44.77%

And after doing a bit of linear algebra one comes up with the following probabilities of winning a given Super Bowl matchup, along with the associated probability of that matchup occuring:

NO	30.70%
NE	69.30%
17.23% matchup probability

NO	39.29%
IND	60.71%
27.53% matchup probability

CHI	29.33%
NE	70.67%
21.26% matchup probability
	
CHI	34.82%
IND	65.18%
33.97% matchup probability

As a rough estimate we could pretty much stop here as by insepction it's readily apparent that the following associated approximate point spreads (and matchup probabilities) imply an expectation worth fewer than 6 points for the NFC.

NO  +6  vs. NE  (17.23%)
NO  +3  vs. IND (27.53%)
CHI +6½ vs. NE  (21.26%)
CHI +4½ vs. IND (33.97%)

However, if we really wanted to be strict about this we would first impute from each of the above four matchup probabilities the probability of covering a 6 point spread. Once this was accomplished determining the fair value for AFC -6 would become a simple excersise in expectations.

So for example, as a first-order approximation (through linear interpolation of a score frequency chart) we could say that given the above probabilities of winning the respective matchups, the probability of the AFC winning by more than 6 points (conditioned on not pushing) would in each case be:

NE  -6 / NO:  49.57%
IND -6 / NO:  40.70%
NE  -6 / CHI: 50.98%
IND -6 / CHI: 45.31%

Hence, the probability of winning AFC -6 would be 49.57%*17.23% + 40.70%*27.53% + 50.98%*21.26% + 45.31%*33.97% = 45.98%.

So by this analysis this means that AFC -6 would be a good bet at about +118 or better, and
at -110 the expectation of AFC -6 would be ~ -12.2%.

[Korchnoi]

Excellent analysis. I'd certainly have to agree with your conclusion.

Nevertheless, I do have one problem with the specifics of your methodology. Namely, you're implictly assuming that the transformation of win probability to point spread would be distributive over composite probabilities. This does not have to hold in the general case.

The way I might go about it would be as follows:

From Pinnacle:

Prob of Winning Super Bowl:
CHI	18.06%
IND	38.86%
NE	26.97%
NO	16.11%

Prob of Winning Championship Game:
CHI	55.23%
IND	61.50%
NE	38.50%
NO	44.77%

And after doing a bit of linear algebra one comes up with the following probabilities of winning a given Super Bowl matchup, along with the associated probability of that matchup occuring:

NO	30.70%
NE	69.30%
17.23% matchup probability

NO	39.29%
IND	60.71%
27.53% matchup probability

CHI	29.33%
NE	70.67%
21.26% matchup probability
	
CHI	34.82%
IND	65.18%
33.97% matchup probability

As a rough estimate we could pretty much stop here as by insepction it's readily apparent that the following associated approximate point spreads (and matchup probabilities) imply an expectation worth fewer than 6 points for the NFC.

NO  +6  vs. NE  (17.23%)
NO  +3  vs. IND (27.53%)
CHI +6½ vs. NE  (21.26%)
CHI +4½ vs. IND (33.97%)

However, if we really wanted to be strict about this we would first impute from each of the above four matchup probabilities the probability of covering a 6 point spread. Once this was accomplished determining the fair value for AFC -6 would become a simple excersise in expectations.

So for example, as a first-order approximation (through linear interpolation of a score frequency chart) we could say that given the above probabilities of winning the respective matchups, the probability of the AFC winning by more than 6 points (conditioned on not pushing) would in each case be:

NE  -6 / NO:  49.57%
IND -6 / NO:  40.70%
NE  -6 / CHI: 50.98%
IND -6 / CHI: 45.31%

Hence, the probability of winning AFC -6 would be 49.57%*17.23% + 40.70%*27.53% + 50.98%*21.26% + 45.31%*33.97% = 45.98%.

So by this analysis this means that AFC -6 would be a good bet at about +118 or better, and
at -110 the expectation of AFC -6 would be ~ -12.2%.

That's exactly the analysis I was too lazy to do, thanks for that Ganchrow.

My simplistic back-of-envolope analysis actually gets the bet to be ~11.5% which is pretty close. The bet looked so bad that I feel I didn't even need to bother going into the detailed analysis. Oh, and I have no way to place a bet on that line anyway

[Ganchrow]

My simplistic back-of-envolope analysis actually gets the bet to be ~11.5% which is pretty close. The bet looked so bad that I feel I didn't even need to bother going into the detailed analysis.

Definitely close enough. The difference between 11.5% and 12.2% would be well within the margin of error of the frequency chart anyway.

BTW, Bet Jamaica's currently offering NFC +5½ +100.

[Korchnoi]

Definitely close enough. The difference between 11.5% and 12.2% would be well within the margin of error of the frequency chart anyway.

BTW, Bet Jamaica's currently offering NFC +5½ +100.

Final nail in the coffin. I certainly think img94 has his answer.

[imgv94]

Thanks a lot fellas.

You guys are too good.

[onlooker]

If the Bears make it, then yes. If the Saints make it, then no. I will just wait to see what the match up is, so I can cap the two teams.

[maxedout42]

Can someone explain to me the math behind the individual match-ups percentages? I know it requires some linear algebra which I haven't done in quite some time.

[TankHankerous]

[rm18]

Good call son