My official betting thread

Remove this Banner Ad

Status
Not open for further replies.
One play for tomorrow:

NYK -1.5 @ 55%

I'm long overdue on housekeeping. Will do that tomorrow, i'm busy tonight.

There's a lot of good points coming out in the posts above which I'll give my 2 cents on tomorrow.

I think Tigez is referring to the Efficient Market Hypothesis (EMH). I believe in the weak-from variant of this theorem, as it is clearly evident there are entities which make long term profits in markets (sports betting or otherwise). Hence the strong form version of this theorem can be likely rejected.


http://en.wikipedia.org/wiki/Efficient-market_hypothesis

I'd like to see some stats on the % of the time that the bookies odds are correct (with maybe a 3-4% margin of error because most punters won't take on an option if it is only 3-4% better odds than they believe it should be).
Click to expand...
I can tell you for one, that the mean absolute error of the closing line in the AFL is roughly 25 points. It seems like a lot, but try predicting a margin, and see how much error there is compared to the actual score line. Get under 25 points in average error, and you're beating the AFL closing line (which means you would be crushing the sport).

Getting to within about 28 points in error is enough for 55% ATS winners if you cherry pick your best overlays. You don't need to beat the closing line to make money.

Bettors have 'first mover advantage' over the books, hence books need to react to action and correct soft lines if they had been posted.
 
brett128 said:
I can tell you for one, that the mean absolute error of the closing line in the AFL is roughly 25 points. It seems like a lot, but try predicting a margin, and see how much error there is compared to the actual score line. Get under 25 points in average error, and you're beating the AFL closing line (which means you would be crushing the sport).

Getting to within about 28 points in error is enough for 55% ATS winners if you cherry pick your best overlays. You don't need to beat the closing line to make money.

Bettors have 'first mover advantage' over the books, hence books need to react to action and correct soft lines if they had been posted.
Click to expand...

Yeah ATS is a good way to look at it. So basically, while the books average 25 points error in AFL lines, that covers every game. As punters we can pick and choose which games we think they erred in, and act accordingly. Gives us a slight edge in that regard if we can act with discipline and consistency.
 
NFL for the weekend:

WAS +3.5 @ 55%
OAK +9.5 @ 57%
BUF +3 @ 57%
KC +10.5 @ 57%
CHI -2.5 @ 57%
STL +2.5 @ 55%
CAR -2.5 @ 60%

Was hoping for some big plays considering a lot of public action on Thanksgiving long weekend over there, but the model doesn't seem to think so.
 

Log in to remove this ad.

CIN +3.5 @ 57% WIN
CLE + 3.5 @ 62% LOSS
BUF +10.5 @ 60% LOSS
WAS -3 @ 60% LOSS
JAC +5.5 @ 60% LOSS
TB + 1.5 @ 60% WIN
NYG -3 @ 60% LOSS
DAL +4 @ 55% LOSS
PHI +3 @ 57% LOSS
PHO +13.5 @ 55% LOSS
UTA +5 @ 55% LOSS
SAS -1.5 @ 55% LOSS
SEA -6 @ 60% WIN
DEN -3.5 @ 55% WIN
SF -11 @ 55% LOSS
DAL -1 @ 57% WIN
HOU +1 @ 55% WIN
GSW +7 @ 57% LOSS
PHI +5.5 @ 57% WIN
DAL -6.5 @ 55% LOSS
ATL +1 @ 57% LOSS
UTA +5.5 @ 57% WIN
PHO +3 @ 55% LOSS
Penn State +12 @ 57% LOSS
Charleston -4 @ 55% LOSS
NYK +6 @ 57% LOSS
ATL -2 @ 57% WIN
Eastern Illinois +10.5 @ 57% WIN
Ball State -6 @ 57% LOSS
San Diego -2 @ 57% LOSS
Princeton -3 @ 57% LOSS
CLE +9 @ 62% WIN
GB -3 @ 62% WIN
NE -9.5 @ 57% WIN
JAC +15 @ 55% WIN
CIN -3.5 @ 55% WIN
SD +8.5 @ 57% WIN
DePaul -16 @ 57% WIN
Pittsburgh -20.5 @ 55% LOSS
DAL +2.5 @ 57% WIN
HOU +8 @ 55% LOSS
Virginia -17 @ 57% WIN
Arizona -17 @ 55% WIN
Cal Poly -2.5 @ 55% LOSS
Ball State -2 @ 55% LOSS
Canisius -2.5 @ 55% WIN
NYK -1.5 @ 55% LOSS



21-26 for the period


60%+ plays: 99-88-4 (0.529) Z Score actual vs expected: -2.01
57%+ plays: 87-80-2 (0.521) Z Score actual vs expected: -1.29
55%+ plays: 64-72-1 (0.470) Z Score actual vs expected: -1.86

250-240-7 (0.510)

Current Index Number: 18
 
NCAAB:

Rice -11 @ 55%

EDIT:

My mistake. I listed the play wrong. It is actually Georgia Tech -11 @ 55%

(Game not started yet).
 
NBA:

GSW -2.5 @ 55%
UTA -1.5 @ 55%

NCAAB:

Dayton -10 @ 55%
Illinois Chicago -5 @ 55%
Samford +3 @ 55%
 
NBA:

ORL +5.5 @ 55%

NCAAB:

Auburn -9 @ 55%
Drexel -13 @ 55%
West Virginia -3 @ 55%
 
brett128 - thoughts on what was discussed a couple of days ago, re the viability of the Kelly method, please?

I've got a number of things I'd like to discuss which I think are accurate re Kelly, and which - in my opinion - make it not viable as a method, however I would like to hear what your thoughts on Kelly are now, after having used it full on for a good period.
 
Duritz said:
brett128 - thoughts on what was discussed a couple of days ago, re the viability of the Kelly method, please?

I've got a number of things I'd like to discuss which I think are accurate re Kelly, and which - in my opinion - make it not viable as a method, however I would like to hear what your thoughts on Kelly are now, after having used it full on for a good period.
Click to expand...



I've been meaning to go through some older posts and put down some thoughts. I'll do that some time today.

NFL TNF:

NO +3.5 @ 55%
 
Taking some dome games now. The rest I wait for weather effects closer in to game time.

DET -4.5 @ 57%
PHI +10.5 @ 60%
 

(Log in to remove this ad.)

d3ft said:
Brett, valid concern here - if your true success rate is 55% (not even the 57% to 62% you've been putting up lately) you have a 0.06% chance of having a run this bad (2 successes in 16 trials). Are you still putting it down to variance, or are you claiming your results don't follow a binomial distribution?
Click to expand...


Don't dissect (or cherry pick) results. You'll commit what's called type 1 error. Conduct the test on the whole dataset.

Take the NFL over the last 3 weeks. I'm 15-3 ATS. Conducting a binomial test on this record shows I would be in the 99.7% percentile in a binomial distribution against a 55% record. Hence I could reject the null hypothesis that I'm a 55% handicapper, and instead assume a new hypothesis that I'm somewhere beyond a 55% handicapper.

Does that mean I'm now an NFL handicapping god? No. Because I'm cherry picking my results, I'm committing type 1 error.


Another scenario. Take a coin, wait for it to go 19 of one side and 1 of another in a 20 coin flip sample. Conduct a binomial test. It will tell you the coin just has to be biased. But we know the coin is not, as we understand the coin flip sample comes from an infinite population that has a long term probability of 0.5 for either side.

These kinds of tests are just about worthless for these sample sizes.
 
Duritz said:
How do you determine if odds are accurate? IE, after the event, how do you determine what the true probability of each team was?

Say the odds on an event are -

Team A $1.20
Team B $6.00

(normalised to 100% for ease of maths here)

Their implied probabilities of winning therefore are -

Team A 0.83
Team B 0.17

(rounded to 2 places, to make it easy)

So, they play, Team B, the outsider, gets up.

Was Team B's probability greater than 0.17 in reality? Or was this win merely representative of the 1 time in 6 that this team would get up?

To know if the odds of the books are generally right long term, you need to know what the true probability was for each event. How do you measure that?
Click to expand...

I don't bet money lines, so I don't have this problem, I can always take the error (MAE or MSE) of the spread.

I recently read a horse book; "Precision: Statistical and Mathematical Methods in Horse Racing", by CX Wong. Wong described how we measured the efficiency of vig adjusted Hong Kong tote prices using a type of R-Squared test. He found that a model needed an r2 of 0.1237 in 1988 to breakeven against the tote, whereas by 2008, the r2 requirement was 0.1741.

The accuracy required by a model to breakeven against the HK tote had risen markedly over 10 years. The same thing is happening in all sports.

Sports handicapping is an arms race with people constantly innovating to find better methods, even Australian sports will have NFL-like efficiency within 5 years. You'll find overseas handicapping groups will make their way to Australia to crush some of the last soft markets left.

Actually this is already happening over here too, the MAE/MSE of the NRL line has decreased drastically over the last 3 years. Someone (or some syndicate) with a lot of money is hammering the line to efficiency. When I went to Sydney, I met an NRL handicapper who said some English quants had set up shop in Sydney a few years ago. This is not coincidentally when I see the NRL line beginning its MAE/MSE drop.

The MAE/MSE trend of the AFL line is nowhere near as bad.
 
brett128 said:
I don't bet money lines, so I don't have this problem, I can always take the error (MAE or MSE) of the spread.

I recently read a horse book; "Precision: Statistical and Mathematical Methods in Horse Racing", by CX Wong. Wong described how we measured the efficiency of vig adjusted Hong Kong tote prices using a type of R-Squared test. He found that a model needed an r2 of 0.1237 in 1988 to breakeven against the tote, whereas by 2008, the r2 requirement was 0.1741.

The accuracy required by a model to breakeven against the HK tote had risen markedly over 10 years. The same thing is happening in all sports.

Sports handicapping is an arms race with people constantly innovating to find better methods, even Australian sports will have NFL-like efficiency within 5 years. You'll find overseas handicapping groups will make their way to Australia to crush some of the last soft markets left.

Actually this is already happening over here too, the MAE/MSE of the NRL line has decreased drastically over the last 3 years. Someone (or some syndicate) with a lot of money is hammering the line to efficiency. When I went to Sydney, I met an NRL handicapper who said some English quants had set up shop in Sydney a few years ago. This is not coincidentally when I see the NRL line beginning its MAE/MSE drop.

The MAE/MSE trend of the AFL line is nowhere near as bad.
Click to expand...

Yeah I've read that, but his thinking is flawed. A LOT of his thinking in that book is flawed, TBH, but that's another matter.

I'm more interested in your thoughts re Kelly as a betting medium. I have some thoughts on this I'd like to discuss with you, but I'd first like to hear your thoughts, you're in the unique position of having used Kelly quite heavily in real time, with real $$$.
 
brett128 said:
I don't bet money lines, so I don't have this problem, I can always take the error (MAE or MSE) of the spread.

I recently read a horse book; "Precision: Statistical and Mathematical Methods in Horse Racing", by CX Wong. Wong described how we measured the efficiency of vig adjusted Hong Kong tote prices using a type of R-Squared test. He found that a model needed an r2 of 0.1237 in 1988 to breakeven against the tote, whereas by 2008, the r2 requirement was 0.1741.

The accuracy required by a model to breakeven against the HK tote had risen markedly over 10 years. The same thing is happening in all sports.

Sports handicapping is an arms race with people constantly innovating to find better methods, even Australian sports will have NFL-like efficiency within 5 years. You'll find overseas handicapping groups will make their way to Australia to crush some of the last soft markets left.

Actually this is already happening over here too, the MAE/MSE of the NRL line has decreased drastically over the last 3 years. Someone (or some syndicate) with a lot of money is hammering the line to efficiency. When I went to Sydney, I met an NRL handicapper who said some English quants had set up shop in Sydney a few years ago. This is not coincidentally when I see the NRL line beginning its MAE/MSE drop.

The MAE/MSE trend of the AFL line is nowhere near as bad.
Click to expand...

In that book, you'll have noted he referred to a bloke I've referred to several times, Bill Benter... he used him as an a example of a greatly successful professional in HK. This Benter was... however, their R2 method of market analysis is flawed... it assumes that you have to be more accurate than the market, or more predictive, to beat it. You don't. You can be far lower on an R2 test and still beat a market, if you're finding the bigger value flaws.

Not restricted to racing, of course. Same with sports. Find the holes and exploit them. You don't have to "beat the market at it's own game".
 
I'll answer this sometime time later on today.

NBA:

DET +7.5 @ 55%

NCAAB:

Arkansas State -3 @ 55%
St Josephs +6 @ 55%
Louisville -11 @ 55%
Davidson -14.5 @ 55%
Princeton +3.5 @ 55%
 
I've taken this now, so I'll put this play out here.

CAR -3.5 @ 62%

This was originally going to be a 62% play at -3 for me. I was looking at line movement history on some games at Covers, and I happened to see an article about the CAR @ KC game.

I found out a KC linebacker had just committed a murder suicide. The books had moved the spread from 3 to 3.5 with the news. Head coach Crennel and the KC leadership group have opted to play on Sunday.

I've taken note of unfortunate incidents over the years, and they offer some really good +EV spots. Deaths, accidents, funerals, etc do not make the affected team play harder (which average Joe often likes to mention). It's human nature, people have their minds on the incident and want to mourn, not play pummeling professional football (in any code).

Yes I understand this is some extremely sad news. But the market waits for no one. An emotional, very flat KC in combination with a KC team that is by far the worst team in the NFL should see an easy CAR cover.

As the unscrupulous Baron Rothschild once said, "buy when there's blood on the streets". Sports betting is no different, sometimes you need to take advantage of very unfortunate situations.
 
Just got down at the TAB in cash. The TAB is the only Australian book that has the game up.

The TAB's CAR -2.5 @ $1.60 is the equivalent of CAR -3.5 @ 1.91 (and better than Pinnacle's $1.885).

CAR at -2.5 is an 81% play at $1.60 (moving from -3.5 to -2.5 is a 19% difference in win %).
 
Week 13 NFL:

Already released:

NO +3.5 @ 55%
PHI +10.5 @ 60%
DET -4.5 @ 57%
CAR -3.5 @ 62%

The rest:

ARZ +6 @ 62%
MIA +7.5 @ 60%
TB +8 @ 60%
PIT +9.5 @ 55%
WAS +3 @ 55%

I've averaged just over 1 x 60% play per week since I re-graded my plays. This week I have 5. I just respond to what the market throws up.


 
Status
Not open for further replies.

Remove this Banner Ad

Similar threads

H
HTP Purton Bets
Replies
11
Views
1K
HTPunter
H

Remove this Banner Ad

Back
Top