[Dark Horse]

Perhaps more interesting than flat versus Kelly is the question of 'flat' fractional Kelly or not. And another question is if any type of Kelly bet size curve, when capped off by a book's limit, should be adjusted accordingly.

[jolmscheid]

Perhaps more interesting than flat versus Kelly is the question of 'flat' fractional Kelly or not. And another question is if any type of Kelly bet size curve, when capped off by a book's limit, should be adjusted accordingly.

Good question Dark Horse...I like the notion of using flat fractional kelly or not...

[That Foreign Guy]

Perhaps more interesting than flat versus Kelly is the question of 'flat' fractional Kelly or not. And another question is if any type of Kelly bet size curve, when capped off by a book's limit, should be adjusted accordingly.

This is an interesting question and one that seems to be begging for a monte carlo sim this weekend.

[wantitall4moi]

a) You are arguing against yourself. The mist efficient market should be the hardest to beat for the player, right?

b) You are willing to say lots of things, nearly all of which are complete garbage.

That is why these arguments dont work. Bettors set the market so are they going to beat it or are they going to make it better? I say they do both. Thats why books keep winning. Where the books make their 'real' profits is in the vig they overcharge on losing bets.

I went back further than 5 years when I looked at openers versus closers. The source took an aggregate of all books. I was just making the point that on average lines arent going to be close to the actual result. Which you pretty much verified as well.

Kelly is based on an advantage to win. To me that is a constant. As in based on a players personal ability/opinion on a game. It doesnt go up or down based on a price they see or a spread they find. While that may in some way effect ther long term results it does not effect their probability of winning.

No one is good enough to look at a line and say a team -2.5 has a better shot to win than the team +3 does. In that particular situation anyway. You either like the dog or you like the favorite, the spread MIGHT play a part, and you always want to get the most favorable odds/points. But for the most part the only 'advantage' comes from getting the best price.

I dont ever remember kelly saying the price is going to set the bet amount, just the win probability. That is where kelly is flawed. I have said that forever. While guys can tweak it and fool with it and try and find a 'happy' medium. It still doesnt change the fact that Kelly isnt a valid approach to sports betting. Because guys that claim they use it arent. Unless any time you raise or lower your bet size by a certain percentage makes it a 'kelly based play'.

Now if you want to call it 'bargain shopping criterion' then by all means. As in you see a price super low on a side you love and you 'stock up' on it like you do on toilet paper when the price drops then fine. But the only thing that determines 'value' in those circumstances is your opinion of the side, and that for guys who 'originate' line moves is generally gone after they make their first bet. So then a 'truer' value can be determined because there is a more in depth comparison to be made. But it still doesnt mean the price is 'right' as shown by me and the other guy trying to debate me, the spreads arent close to the results and the changes in spreads dont matter too much either. So once again it comes down to the ability of the person to look past all the shit guys get caught up in, find a winning side and make a play at the best price they can find.

The irony is a guy with TRUE ability would avoid games that are going to be nip and tuck versus the spread. with so many games that arent going to be close if guys truly had the ability to not only figure out their advantage but use it on every play they would ALWAYS bet the games with the biggest 'advantage. Example CLippers versus Cavs last night. line opened 5 went up and down slightly finally closing 7.5 because the kid on Clev wasnt playing. Cavs never in jeopardy really of losing hthat game let alone not covering 7. So 'market' wrong, therefore most players I am sure were wrong, no matter what system they used to figure out their bet amount. Flip side you have Miin and mem, opened 8 moved around and finally closed at 7.5 with 'favorable' vig. Memphis was easily covering that spread throughout the game, they led by 15 in the 4th quarter. Then for some reason with 10 seconds left and a 10 point lead Mem fouls a guy on Minn to give them 2 free throws, then a desperation 3 gets them the cover on all numbers. So two games two mves. One win one loss, but one a clear cut example of 'wrong side' that a guy with TRUE ability would have had the 'right' side. The other a win, but a very lucky win. But in the tally sheet it doesnt matter.

Either way if youre truly good enough to factor your win probability you should very well be good enough to not bet games that arent even close.

[FourLengthsClear]

Another 8 paragraphs of complete drivel.

As mentioned people are perfectly entitled to the view that Kelly is not suitable for them in sports betting. If however, someone has an approach/model which has proved to be reliable and can derive a fair line (say +5.5) for a given game based on the information available at the time. Then calculating an edge and Kelly stake based on -104 odds at a +6.5 spread is the easy part.

Am I expecting to win every such game? Of course not. Am I expecting the favourite to win some of those games by 15 or 16 points? Yes I am. Does this invalidate the approach/model? No.

In the long run, all I am looking for is the winning percentage of such a bet to be in the 53.5 - 54.0% range that is consistent with the one point 'advantage' I have (in NBA) against the line bet and which is the source of the 5.0 - 5.9% full Kelly risk amount.

You can maintain until the cows come home that a half point or one point is too small a margin of error. My models will tend to show a fair line within a half point of NBA openers 90% of the time. It is the generally the games that fall outside this range that a) I looking to see if there is injury information or something else that I am missing and b) throw up wagering opportunities.

I don't expect you to get any of this, in one of your previous posts you described the chances of a half point differential affecting the outcome of a bet as "minuscule". Anyone who believes that has no understanding of simple probabilities and as such wouldn't know where to start in terms of regression/simulation based modelling.

[YOUSENKO]

The best way is you just start betting big on a start of winning streak, preferbly going all in for the first few games, then cool down and play smaller stakes. Withdraw when u hit the first lost. Variance isn't fun.

There are ways to find out whether winning streak starts, like betting casually first to find out your form.

WIn more lose less is the way to profit in gambling although its not easy. But if u realised most losers didnt quit at the right time but continue playing and get burnt.

You will lose anyway if you say it is a strategy to aim 55% long term, its not gonna get it done. They vig u out.

[hutennis]

Another 8 paragraphs of complete drivel.

As mentioned people are perfectly entitled to the view that Kelly is not suitable for them in sports betting. If however, someone has an approach/model which has proved to be reliable and can derive a fair line (say +5.5) for a given game based on the information available at the time. Then calculating an edge and Kelly stake based on -104 odds at a +6.5 spread is the easy part.

Am I expecting to win every such game? Of course not. Am I expecting the favourite to win some of those games by 15 or 16 points? Yes I am. Does this invalidate the approach/model? No.

In the long run, all I am looking for is the winning percentage of such a bet to be in the 53.5 - 54.0% range that is consistent with the one point 'advantage' I have (in NBA) against the line bet and which is the source of the 5.0 - 5.9% full Kelly risk amount.

You can maintain until the cows come home that a half point or one point is too small a margin of error. My models will tend to show a fair line within a half point of NBA openers 90% of the time. It is the generally the games that fall outside this range that a) I looking to see if there is injury information or something else that I am missing and b) throw up wagering opportunities.

I don't expect you to get any of this, in one of your previous posts you described the chances of a half point differential affecting the outcome of a bet as "minuscule". Anyone who believes that has no understanding of simple probabilities and as such wouldn't know where to start in terms of regression/simulation based modelling.

For the record. I think I understand and agree with everything here.
If I would have been able to establish an unquestionable edge, I would use Kelly every time.

[FourLengthsClear]

For the record. I think I understand and agree with everything here.
If I would have been able to establish an unquestionable edge, I would use Kelly every time.

The edge is not unquestionable it is merely a good estimate which is why I use 3/4 Kelly.
Sports markets and matches do not, and will never, have the certainties involved with flipping a coin.

[hutennis]

The edge is not unquestionable it is merely a good estimate which is why I use 3/4 Kelly.
Sports markets and matches do not, and will never, have the certainties involved with flipping a coin.

How good is your good?

B/c if it's not good enough (and you have no way of knowing the true extend one way or another),
you need to be very lucky not to get hurt.

[FourLengthsClear]

How good is your good?

B/c if it's not good enough (and you have no way of knowing the true extend one way or another),
you need to be very lucky not to get hurt.

A model breaking down is part of the game and being guided by the market to that possibly is the easiest way to avoid that hurt. If markets do not move consistently in the direction I expect them to then I am, of course, questioning whether that model and the variables within it are still usable.

[jolmscheid]

I would say half kelly so that the edges where one is "off" will not kill you...

Also, 4LC, do you use your entire bankroll when putting in your simultaneous kelly wagers, or do you break up your bankroll per sport?

Thanks for the insight..

[FourLengthsClear]

I would say half kelly so that the edges where one is "off" will not kill you...

Also, 4LC, do you use your entire bankroll when putting in your simultaneous kelly wagers, or do you break up your bankroll per sport?

Thanks for the insight..

I don't subdivide my bankroll in any way. It has been known for me to have almost my entire bankroll in play on Saturday afternoons.

[wantitall4moi]

Another 8 paragraphs of complete drivel.

As mentioned people are perfectly entitled to the view that Kelly is not suitable for them in sports betting. If however, someone has an approach/model which has proved to be reliable and can derive a fair line (say +5.5) for a given game based on the information available at the time. Then calculating an edge and Kelly stake based on -104 odds at a +6.5 spread is the easy part.

Am I expecting to win every such game? Of course not. Am I expecting the favourite to win some of those games by 15 or 16 points? Yes I am. Does this invalidate the approach/model? No.

In the long run, all I am looking for is the winning percentage of such a bet to be in the 53.5 - 54.0% range that is consistent with the one point 'advantage' I have (in NBA) against the line bet and which is the source of the 5.0 - 5.9% full Kelly risk amount.

You can maintain until the cows come home that a half point or one point is too small a margin of error. My models will tend to show a fair line within a half point of NBA openers 90% of the time. It is the generally the games that fall outside this range that a) I looking to see if there is injury information or something else that I am missing and b) throw up wagering opportunities.

I don't expect you to get any of this, in one of your previous posts you described the chances of a half point differential affecting the outcome of a bet as "minuscule". Anyone who believes that has no understanding of simple probabilities and as such wouldn't know where to start in terms of regression/simulation based modelling.

This is why it fails. You are still using these 'values'. More than likely values established by a book. Which are never constant. a half point at just about every spread other than 3 or 7 in the NFL are meaningless. In the case of 3 and 7 there is always a premium attached to them. Some books might offer 3 -110 the other book might offer 2.5 -130. Some people would say -2.5 -130 is a 'no brainer'. I could show it isnt in a lot of cases. But the 'value' of that half point regardless of how it is obtained its still set by how the books offer it. You could also buy it at the book with -3 -110, but more than likely would have to pay -135 for it.

Where people get in trouble is looking at half point differentials with varied vigs and trying to ascertain their 'value'. There is no real way to test it. The only way to try and obtain a value is looking at past games that have pushed or been close to the line, that is flawed as well. In each new game the value of the half point is unknown. But theoretically the 'better' the line the more 'valuable' the half point might be. The trick is (and it is impossible) to determine which lines are in fact 'better'. Which I addressed above. Why bet those games? Just bet the games with 'bad' lines.

I always laugh at people who claim they are good because theyre constantly on the 'right' side of these half or one point wins related to the spread. Why sweat it out and try to look 'smart' bet a game that isnt within 20 of the spread and have no worries. Like I say it is one of the greatest ironies there is.

As far as kelly is concerned, if a person feels they can hit 54% lifetime then they will more than likely be betting 5% of their bank roll per play too. So in essence theyre following kelly AND flat betting. Or if they only bet 3% per play then they could just as easily say theyre betting kelly there too just a bastardized form of it like everyone else does.

When I was betting full time I bet 1 to 1.5% of my bankroll per (one sided) play depending on how many games I was going to play one sided. I basically capped my over all exposure. So if I had 10 plays going at one time I might bet 1.5%, if I had 20 I would only bet 1%. Of course that was determined by the limits of the books I was at, in some cases 1.5% of my bankroll exceeded that. When I took leads I had no cap and no regard for percentages, I bet the limit, sometimes multiple times until the line changed enough to where I wasnt confident of having a profitable or at least break even buy back. I suppose in some way that could be construed as following kelly because with a 100% probability you bet 100% of your bank roll. But since I bet up to 8 leads I couldnt obviously bet 100% of my bankroll on one side of every one of them. So I had to pick and choose which ones I thought would move the most or look at the odds and see which ones offered the greatest potential with the smallest moves. In the end I was basically all in (had as close to 100% of my bankroll in action) on as many plays as I could get. Then it was a matter of collecting, tallying the returns and starting over the next day. But there were bankroll limitations after awhile. So that is something that not a lot of people address either.

So we can slice and dice it as many ways as you want, it is going to be mostly semantics. Kelly just tries to determine an OPTIMUM bet size rather than the willy nilly of 1,2,3,5, star/unit systems most guys use on their own. But for all intents and purposes theyre the same thing.

[FourLengthsClear]

This is why it fails. You are still using these 'values'. More than likely values established by a book. Which are never constant. a half point at just about every spread other than 3 or 7 in the NFL are meaningless. In the case of 3 and 7 there is always a premium attached to them. Some books might offer 3 -110 the other book might offer 2.5 -130. Some people would say -2.5 -130 is a 'no brainer'. I could show it isnt in a lot of cases. But the 'value' of that half point regardless of how it is obtained its still set by how the books offer it. You could also buy it at the book with -3 -110, but more than likely would have to pay -135 for it.
[/COLOR]

Again you display incredible ignorance for someone who claims to have been doing this for a long time.

I maintain my own push probability tables. Even in the NFL no half point is "meaningless".

A game where the fair base spread is -3.5/+3.5 and the projected total is 45 will have a different 'profile' to a game with -14/+14 and a expected total of 55.

But let's follow your line of thought. Suppose I have a game where the best line available in the market is +9 -110 and I have a calculated fair line of +7.5. Does the fact that 8 and 9 are unlikely margins of victory mean that I have no edge? No, it simply means that my edge is very marginal.

A projected push rate of 2.31% on the 8 and 1.32% on 9 would give me an expected win rate of just under 53% (50 + 2.31 + 0.66*).

*0.66% being half of the push probability on 9.

As far as kelly is concerned, if a person feels they can hit 54% lifetime then they will more than likely be betting 5% of their bank roll per play too. So in essence theyre following kelly AND flat betting. Or if they only bet 3% per play then they could just as easily say theyre betting kelly there too just a bastardized form of it like everyone else does.

Not all sports are spread based. And odds are not always -110.
Betting full Kelly really just means betting to win the +EV. It's not rocket science.

If the +EV was known (and in sports betting it cannot be) then Kelly is optimal. That is mathematical fact.

[wantitall4moi]

Again you display incredible ignorance for someone who claims to have been doing this for a long time.

I maintain my own push probability tables. Even in the NFL no half point is "meaningless".

A game where the fair base spread is -3.5/+3.5 and the projected total is 45 will have a different 'profile' to a game with -14/+14 and a expected total of 55.

But let's follow your line of thought. Suppose I have a game where the best line available in the market is +9 -110 and I have a calculated fair line of +7.5. Does the fact that 8 and 9 are unlikely margins of victory mean that I have no edge? No, it simply means that my edge is very marginal.

A projected push rate of 2.31% on the 8 and 1.32% on 9 would give me an expected win rate of just under 53% (50 + 2.31 + 0.66*).

*0.66% being half of the push probability on 9.

Your using a flawed database I am sure to get those probabilities, because more than likely your using past results to ascertain those percentages.

So with such small examples of pushes, and more than likely even sample size if your database has say 151 games with a line of -9, and 2 of those pushed on 9 (I extrapolated it) that is how you get your 1.32% push rate. Now if a guy has a much more extensive database and looks and see that the 2 times you saw pushes he saw both times that any other number other than 9 was available then he could say that there were zero pushes. The only valid claim you can make is that against CLOSERS in MY database the push rate since xxxx has been 1.32%. That in no way says that is what it will be going forward or even valid considering different line availability. But for shits and giggles I looked at my Db and found 143 games back to 1989 that closed -9. 2 of them did push. But is that a big enough sample size? Obviously it is 23 seasons so you can only use what you get, but is it enough games to determine probability? Now if you want to look at all games played and how many landed with a SU win margin of 9 points and do some fancy math there then fine. That would be 99 times, and a SU win margin obviously doesnt have any room for error like games closing -9 does. so 99 out of 5715 total games have ended in a SU win by 9 points. So in a vacuum one might try and say the probability of a 9 point differential in the NFL is 1.7%. I will let the probability guys try and figure out the other stuff.

Past results do not predict future expectations.

Back in 2001 the push rate on the 3 was over 10%, since then it has been less than 9. Lowering the push rate overall to just a little over 8%. So did the probability really drop that much in a decade or have rules changes taken place that have made games less likely to land on the 3 when lined -3?

When you factor in that from 1989 to 2000 450 of 2797 games finished with one team winning by a FG, and from 2001 to present 461 of 2916 games have had a team winning by a FG. That is 16% and 15.8% respectively. According to this database the push rates were from 1989 to 2000 47 of the 450 games pushed (10.4%) in the 461 games from 2001 to present 38 pushed versus the closer, thats only 8.2%.

So while the overall probability of a 3 point differential stayed fairly close, the push rates dropped off dramatically. So how can you even begin to try and figure out probability versus a spread using past results?

[FourLengthsClear]

Your using a flawed database I am sure to get those probabilities, because more than likely your using past results to ascertain those percentages.

So with such small examples of pushes, and more than likely even sample size if your database has say 151 games with a line of -9, and 2 of those pushed on 9 (I extrapolated it) that is how you get your 1.32% push rate. Now if a guy has a much more extensive database and looks and see that the 2 times you saw pushes he saw both times that any other number other than 9 was available then he could say that there were zero pushes. The only valid claim you can make is that against CLOSERS in MY database the push rate since xxxx has been 1.32%. That in no way says that is what it will be going forward or even valid considering different line availability. But for shits and giggles I looked at my Db and found 143 games back to 1989 that closed -9. 2 of them did push. But is that a big enough sample size? Obviously it is 23 seasons so you can only use what you get, but is it enough games to determine probability? Now if you want to look at all games played and how many landed with a SU win margin of 9 points and do some fancy math there then fine. That would be 99 times, and a SU win margin obviously doesnt have any room for error like games closing -9 does. so 99 out of 5715 total games have ended in a SU win by 9 points. So in a vacuum one might try and say the probability of a 9 point differential in the NFL is 1.7%. I will let the probability guys try and figure out the other stuff.

Past results do not predict future expectations.

Back in 2001 the push rate on the 3 was over 10%, since then it has been less than 9. Lowering the push rate overall to just a little over 8%. So did the probability really drop that much in a decade or have rules changes taken place that have made games less likely to land on the 3 when lined -3?

When you factor in that from 1989 to 2000 450 of 2797 games finished with one team winning by a FG, and from 2001 to present 461 of 2916 games have had a team winning by a FG. That is 16% and 15.8% respectively. According to this database the push rates were from 1989 to 2000 47 of the 450 games pushed (10.4%) in the 461 games from 2001 to present 38 pushed versus the closer, thats only 8.2%.

So while the overall probability of a 3 point differential stayed fairly close, the push rates dropped off dramatically. So how can you even begin to try and figure out probability versus a spread using past results?

I happened to pluck those numbers from thin air just to give an example. The methodology I use to determine push probabilities for wagering purposes is my business and I am not about to get into a debate about its validity with you (or anyone else for that matter).

If however, you take the view that "past results do not predict future expectations" then you cannot predict that 3, 7 and 10 will be key numbers next season in the NFL and for that matter you cannot say with any confidence that the Miami Heat will be a better basketball team than the New Jersey Nets for the remainder of this NBA season, right?

[Dark Horse]

The edge is not unquestionable it is merely a good estimate which is why I use 3/4 Kelly.
Sports markets and matches do not, and will never, have the certainties involved with flipping a coin.

3/4 Kelly? That's very high.

Unless the bet is purely mathematical, as in a prop, sports bettors deal with sample sizes that are generally considered adequate in the 300-400 range (per model). A statistician will frown upon that number, and will set his required sample size closer to 10,000. But that is not practical in sports betting, and to protect themselves against fluctuations that the uncovered 9,600-9,700 sample may yet have in store, sports bettors reduce their risk by turning to fractional Kelly.

Even if a sample size is considered clean, it may not be. Let's say you perceive a winning play over 50 games. It just jumps off the page. Money in the bank. After those 50 games you start betting it. You didn't include the 50 games, so statistically your sample is clean. And yet in many cases it may not be. Because you caught a fluctuation. If it continues for the rest of the season, that fluctuation will make your 300-400 sample look very good. But the 10,000 sample will have additional fluctuations, and they may not all be in your favor.

It is certainly possible to develop models that are good each season. But that is a huge step beyond where most players are. In the big sports this may be easiest in baseball, which in my experience is the most mathematically clean; not to mention that the teams play twice the number of games as NBA and NHL teams, and ten times that of NFL teams. Hit a rough stretch in your NFL season and the ATS effect will be beyond repair for that season. Hit a rough stretch in a MLB season, and it will hardly matter in the end.

In any case, a Kelly discussion to me is about taking it out of the realm of theory and into the reality of sports betting. That reality includes the fact that the sports betting market has become much tighter after the anti-gambling legislation from a few years ago. Who do you think did not walk away? Do you think it affected the lines? If so, what does that tell you about long term models?

[hutennis]

If however, you take the view that "past results do not predict future expectations" then you cannot predict that 3, 7 and 10 will be key numbers next season in the NFL and for that matter you cannot say with any confidence that the Miami Heat will be a better basketball team than the New Jersey Nets for the remainder of this NBA season, right?

I don't want to insult any one's intelligence, but this is beyond childish.

Don't you understand that IT IS ALREADY PRICED IN?

Its like those shady commercials a few years back encouraging dumb segments of a population to buy gasoline futures to make a killing because summer driving season is upon us.
TYVM for letting us on a huge secret that summer comes after spring.

Do you know one spread related set of odds that would not discount 3,7 and 10 numbers in itself?

And if you would to make a prop bet on relative performance Heat vs Jet, would you offer Heat at even money?
At even money I'd say past performance does have a great chance to guarantee future result of making money on your bet.
But once you set the proper price, coming ahead financially on future bets does not look like such a sure thing anymore.

[FourLengthsClear]

I don't want to insult any one's intelligence, but this is beyond childish.

Don't you understand that IT IS ALREADY PRICED IN?

Its like those shady commercials a few years back encouraging dumb segments of a population to buy gasoline futures to make a killing because summer driving season is upon us.
TYVM for letting us on a huge secret that summer comes after spring.

Do you know one spread related set of odds that would not discount 3,7 and 10 numbers in itself?

And if you would to make a prop bet on relative performance Heat vs Jet, would you offer Heat at even money?
At even money I'd say past performance does have a great chance to guarantee future result of making money on your bet.
But once you set the proper price, coming ahead financially on future bets does not look like such a sure thing anymore.

Of course it is already priced in.
The job in hand is finding deficiencies in the method of pricing. It is the same with any type of bet.

[hutennis]

So how does past performance guarantees anything in your quest to find those new, ever changing deficiencies in new, ever changing equations with new, ever changing variables?

Its like draw cards from the deck trying to come up with its exact composition without knowing that the deck you draw from is just a 52 cards random pick from 20 deck pile of cards
and it is a new pick before your every draw.

[FourLengthsClear]

3/4 Kelly? That's very high.

Unless the bet is purely mathematical, as in a prop, sports bettors deal with sample sizes that are generally considered adequate in the 300-400 range (per model). A statistician will frown upon that number, and will set his required sample size closer to 10,000. But that is not practical in sports betting, and to protect themselves against fluctuations that the uncovered 9,600-9,700 sample may yet have in store, sports bettors reduce their risk by turning to fractional Kelly.

Even if a sample size is considered clean, it may not be. Let's say you perceive a winning play over 50 games. It just jumps off the page. Money in the bank. After those 50 games you start betting it. You didn't include the 50 games, so statistically your sample is clean. And yet in many cases it may not be. Because you caught a fluctuation. If it continues for the rest of the season, that fluctuation will make your 300-400 sample look very good. But the 10,000 sample will have additional fluctuations, and they may not all be in your favor.

It is certainly possible to develop models that are good each season. But that is a huge step beyond where most players are. In the big sports this may be easiest in baseball, which in my experience is the most mathematically clean; not to mention that the teams play twice the number of games as NBA and NHL teams, and ten times that of NFL teams. Hit a rough stretch in your NFL season and the ATS effect will be beyond repair for that season. Hit a rough stretch in a MLB season, and it will hardly matter in the end.

In any case, a Kelly discussion to me is about taking it out of the realm of theory and into the reality of sports betting. That reality includes the fact that the sports betting market has become much tighter after the anti-gambling legislation from a few years ago. Who do you think did not walk away? Do you think it affected the lines? If so, what does that tell you about long term models?

I suspect we deal with slightly different methods in terms of modelling. Mine tend to be holistic with data as granular as possible. That might be why my attempts at MLB have always sucked, maybe I am missing the wood because of all the trees!

I am not American and most of my efforts are still on 'European' sports where the legislation you refer to are not really a factor.

My experience of 'modelling', dates back to 1984 when as a 13-year old and armed with a pencil, paper and a ZX Spectrum (replete with 48k RAM) I attempted to reverse engineer the odds offered on Saturday soccer matches by the big UK High Street books. These were published on a coupon every Monday and remained valid until Kick-off. Starting out I knew nothing of Poisson processes, (not that they would be of any use nowadays) or binomial distributions but after a year or so I got there. I then turned my attention to finding deficiencies in that pricing model even though I knew they couldn't be readily exploited (odds were heavily juiced and the player was taxed at source in the UK back then).

Pricing models and markets have moved on a lot since then but my approach is still very much aimed at a) projecting what the books' opener will be and b) having an edge against that number. Needless to say it is a lot easier in Euroleague basketball than it is in NBA but by the same token across all the sports/leagues that I have created models for, the data I have is both extensive and clean enough to trust my edge calculations.

[wantitall4moi]

Like someone said its priced in and uses people's insecurities against them. Why do you think books get away with charging so much? Superbowl was a great example, guys laying 125. 130, and 135 and thinking they were 'smart' for doing it. This is why books always win because they have people fooled.

3 and 7 are 'key' numbers, but the only reason they push more is because the overall chance of them hitting is higher. But the fact that the overall numbers highly 'outperform' actual push rates shows that books still arent 'efficient' enough to predict a valid spread.

But as I have said before people worrying about even the 3 are pissing in the wind. Because more often than not spread other than 3 is available even for less vig than one would think it was 'worth'.

As I have said spreads arent predictors, theyre numbers made to get action as close to balanced as possible.

Again to cite this past Superbowl. it was -3 at most books. It did go to 2.5 really late, and even though most people thought Giants was the 'right' side a lot of people took NE -2.5 because it was the 'better' number. So if that doesnt show how lines effect how people play nothing will. All week we heard how Books were getting killed or how they lost their asses. Once the tally was dont the truth came out and once again books did better than fine.

I will tell you the very best approach anyone who wants to gamble can do. Take a list of the games each week with no spreads, no odds, no nothing gambling related and simply circle teams you think can win the game. Then do some handicapping look at injuries, schedule, anything you feel is relevant then re-evaluate your first impressions. Then bet accordingly. Dont worry about spreads or lines or vig. Obviously try to get the best price you can on your side but dont get swayed off your opinion. That goes for every single sport. Obviously baseball is a little different, but for spread based sports that is something everyone should do.

Once you figure out if you can pick a winner or enough winners ot have a chance then start getting a little more advanced.

But if anything this whole thread shows how everyone over thinks everything to the point where they dont even know what theyre talking about. keep it simple stupid has always ben a phrase everyone should live by.

[FourLengthsClear]

So how does past performance guarantees anything in your quest to find those new, ever changing deficiencies in new, ever changing equations with new, ever changing variables?

Its like draw cards from the deck trying to come up with its exact composition without knowing that the deck you draw from is just a 52 random pick from 20 deck pile of cards
and it is a new pick before your every draw.

With respect, it really isn't.

Question for you though. Do you actually risk money in sports betting? If so what is the rationale behind the wagers you make? By the tone of your posts it sounds like you believe these markets are unbeatable.

[hutennis]

With respect, it really isn't.

Question for you though. Do you actually risk money in sports betting? If so what is the rationale behind the wagers you make? By the tone of your posts it sounds like you believe these markets are unbeatable.

Oh, I see. Thats easy. I dont bet. I mean I dont handicap. I trade.
My firm stand is that the only way (outside of being lucky or manipulating the market) to get ahead in a losing game (50/50 minus commission) is to be small when I lose and big when I win.
I dont want to sound like talking bumper stickers, but that's still the best way to put it. Instead of trying to predict random future, I try my hardest to skillfully react to what have already took place. Instead of trying to buy low or sell high I love cutting losses and let profit run.
Nowadays, I found it to be much easier done in sports than lets say in financial markets. More transparency, muuuuuch less whipsawing, just about enough liquidity. Yep, I like it here.

[FourLengthsClear]

Oh, I see. Thats easy. I dont bet. I mean I dont handicap. I trade.
My firm stand is that the only way (outside of being lucky or manipulating the market) to get ahead in a losing game (50/50 minus commission) is to be small when I lose and big when I win.
I dont want to sound like talking bumper stickers, but that's still the best way to put it. Instead of trying to predict random future, I try my hardest to skillfully react to what have already took place. Instead of trying to buy low or sell high I love cutting losses and let profit run.
Nowadays, I found it to be much easier done in sports than lets say in financial markets. More transparency, muuuuuch less whipsawing, just about enough liquidity. Yep, I like it here.

Fair enough, I am guessing you are not American.
I do a bit of trading myself but as you might guess, I am incredibly anal about it.
http://www.sportsbookreview.com/forum/players-ta...fair-bots.html

[hutennis]

Yeh, good read.
Not me though. I have nothing, and i mean nothing, to do with scalping, bots, line skewing etc.
I do thank bots (or whoever) for being there for me when I need to sell, though

[Dark Horse]

This is an interesting question and one that seems to be begging for a monte carlo sim this weekend.

Look forward to the results.

[135steward]

Why do you think books get away with charging so much? .... why books always win because they have people fooled....

numbers highly 'outperform' actual push rates shows that books still arent 'efficient' enough to predict a valid spread.

****

As I have said spreads [are] made to get action as close to balanced as possible.

I will tell you the very best approach anyone who wants to gamble can do. Take a list of the games each week with no spreads, no odds, no nothing gambling related and simply circle teams you think can win the game. Then do some handicapping look at injuries, schedule, anything you feel is relevant then re-evaluate your first impressions. Then bet accordingly. Dont worry about spreads or lines or vig. Obviously try to get the best price you can on your side but dont get swayed off your opinion. That goes for every single sport. Obviously baseball is a little different, but for spread based sports that is something everyone should do.

Once you figure out if you can pick a winner or enough winners ot have a chance then start getting a little more advanced.

But if anything this whole thread shows how everyone over thinks everything to the point where they dont even know what theyre talking about. keep it simple stupid has always ben a phrase everyone should live by.

I'm concerned that the upper portion of your analysis (italicized) contains contradiction. But the last few paragraphs say the same thing 4LC says. Your only disagreement is to what degree you carry out your handicapping. Both ways, btw, have been shown to work. Wasn't that guy in the Casino movie doing what you're talking about, wantit? And use of computer modeling is simply another way to gather and sort data, right?

[jolmscheid]

If you don't know your exact edge (nobody does).. just play 1.5-2% per play and re-calc bank daily...if you indeed have an edge, then volume will be your friend...double your bankroll every calendar year and see how quick that adds up...

[135steward]

...double your bankroll every calendar year ...

I'm in!

[p19101]

If you are new to sportsbetting, stick to flat betting.

When you become proficient at handicapping, that is a good time to change to various unit-sized betting. While handicapping helps you find value in places, finding lines that are off does also. If you liked Pittsburgh -7 this week, and found Pitt -6.5, wouldn't you bet more?

While I understand how you're thinking, some of the best punters I've come across in my life still lives and dies by the flat betting. I have myself tried to practice flat betting but I feel it's damn hard to do.

[sneakerhead]

Go big or go home