[Ganchrow]

The huge difference between the real, -based on the present-, mathematical expectation in blackjack, and the hoped for, -rooted in the past-, expectation for a sports event is one reason why full Kelly can never work in sports. It's not about percentages, which may look deceptively similar, but about past and present.

If I'm understanding you correctly and your argument is that Kelly is ineffective in sportsbetting because handicappers are unable to adequately judge their true winning percentages for any given bet, then insofar as your premise might be true for any given individual, I certainly have no argument with your conclusion. (You'll notice that I listed this as Kelly problem #1 above).

I hope you realize, however, that this is not an issue with Kelly qua Kelly, but rather one that stems directly from from imperfect forecasting methodology.

"Garbage in, garbage out," as they say. (In other words, if your forecasts are faulty you shouldn't be using Kelly.)

[Dark Horse]

If every sports game were as neatly packaged as a new deck of cards I would be all over Kelly.

Imperfect forecasting methodology is inherent in sports betting. Doesn't that fluidity present a far greater challenge than the dry card counting in blackjack?

[Dark Horse]

Come to think of it. Ganch, you don't strike me as the kind of guy who would be into sports betting..., unless he had put together a very advanced computer program.

[Ganchrow]

JR Miller's Debunking the Kelly Criterion

So just for fun, I wrote a quick Perl simulator to run JR Miller's "experiment" (60 deals from a 2-deck shoe. Player wins on 7 through King, loses on 2 through 6, and pushes on Ace) for both full Kelly staking and flat-betting using a bet size equivalent to the Kelly staking average of 25.056% (if any bet represented more than the current bankroll, the bet size was set to equal that bankroll.)

The results were far from surprising:

Statistic	"Full Kelly"	Flat Betting
N	1,377,000	1,307,000
Avg Finishing Bankroll	322.480%	284.885%
Std Dev of Finishing Bankroll	173.682%	170.328%
Probability of Profitability	85.642%	80.229%
Probability of losing 90% or more of initial bankroll	0.965%	18.730%

(Bet size and finishing bankroll statistics are given as a percentage of initial bankroll. Source code available upon request.)

It pays to mention to that the only reason why the average flat betting returns are so relatively large compared to the Kelly returns (an average flat betting bankroll growth rate of about 83% of that of Kelly), is because the flat bet size chosen is very close to the correct Kelly weighting. This is obviously because we've implicitly used the Kelly criterion to determine our flat bet size. Good luck coming up with that figure otherwise.

So I wonder when I'll be getting my thank you note from JR Miller?

[Dark Horse]

The deck just went significantly positive on the advanced program.

[Ganchrow]

If every sports game were as neatly packaged as a new deck of cards I would be all over Kelly.

Imperfect forecasting methodology is inherent in sports betting. Doesn't that fluidity present a far greater challenge than the dry card counting in blackjack?

Once again it all comes down to the confidence a bettor has in his forecasts. An argument could easily be made the noise inherent in estimating the likelihood of hitting a middle or perhaps a correlated parlay or a teaser is on par with the noise inherent in common card counting strategies (especially when executed under pressure).

[Dark Horse]

In the long run I might agree. The only problem is that there may not be a long run if one uses full Kelly in sports betting.

One problem I've encountered is that even reliable methods can vary -quite severely- from season to season. Statistically, and based on sample size, these systems are very good; yet they may not produce as expected the next year. Quite possibly they need to be qualified further, but that could always be the case. If I had bet the farm, which I should have according to Kelly, on my best system from last season, I wouldn't be very upbeat right now. (My 'solution' is to play the system at lower betsizes until it kicks in, as I know it eventually will; a form of money management comparable to the positive deck approach, but not the same). Mind you, the system is still winning; just not at the expected rate!

In other words, there are no hidden elements in blackjack. To make the same assumption towards sports betting could be very dangerous.

[TLD]

I still have two problems with an argument for flat betting based on the difficulty of ascertaining the likelihood of the bet’s winning.

1. If you don’t know the likelihood of a bet winning, then you don’t know if it’s above the “break even” level relative to the juice you’re paying to make it bettable in the first place. If you can be sufficiently confident for sportsbetting purposes that a game is above or below 52.4%, how is it all of a sudden impossible to ascertain if it is above or below 55%, above or below 60%, etc.?

2. To restate a point I made earlier in the thread, when you flat bet you aren’t avoiding the question “How likely are these plays to win?” Your implicit answer to the question is “Every game I bet has an equal chance of winning.” That’s not neutral.

So even if the premise “You can’t reliably estimate the likelihood of winning a bet in sportsbetting” is conceded for the sake of argument, the conclusion “Therefore you should flat bet” simply doesn’t follow.

[Ganchrow]

I still have two problems with an argument for flat betting based on the difficulty of ascertaining the likelihood of the bet’s winning.

1. If you don’t know the likelihood of a bet winning, then you don’t know if it’s above the “break even” level relative to the juice you’re paying to make it bettable in the first place. If you can be sufficiently confident for sportsbetting purposes that a game is above or below 52.4%, how is it all of a sudden impossible to ascertain if it is above or below 55%, above or below 60%, etc.?

2. To restate a point I made earlier in the thread, when you flat bet you aren’t avoiding the question “How likely are these plays to win?” Your implicit answer to the question is “Every game I bet has an equal chance of winning.” That’s not neutral.

So even if the premise “You can’t reliably estimate the likelihood of winning a bet in sportsbetting” is conceded for the sake of argument, the conclusion “Therefore you should flat bet” simply doesn’t follow.

You're certainly on point here and I'm in agreement with you in theory even if not in practice.

1. You're right. It does seem a bit arbitrary doesn't it? But anecdotally at least, it does certainly appear that there exist profitable handicappers who, for whatever reason, are unable to handicap there own handicapping.
   
   Now because I do lack any real data to back up this assertion, I obviously can't definitively state whether this is a real phenomenon or merely a case of urban sportsbetting legend (which is to say that these handicappers either a) aren't actually long-term profitable, or b) are in fact better at coming up with unbiased forecasts than they are aware). Nevertheless, I can think of a reasonable hypothesis as to why this might be the case. Specifcally, the skill set necessary for determining actual forecasts (or even ordinal forecast tranche ranking, as in the case of a "starring" system) doesn't match up exactly with the skill set involved in the more qualitative approach to handicapping undertaken by most bettors. If determining accurate forecasts is a learned skill related to, but which does not follow completely and directly from, the initial act of handicapping as commonly practiced, then it's easy to see how a few realizations of forecast noise (especially when that bettor is first starting his career) might discourage him from honing his forecasting skills even as his skills at binary forecasting might continue to improve.
   
   I myself have always encouraged neophyte sports bettors to determine their own lines before making handicapping decisions. This is really simplest way to come up with forecasts. A bettor compares the probability implied by his own money lines with the lines offered by the market (adjusting as necessary for differences in point spreads and totals by utilizing either a push probability chart or some more advanced methodology) and then comes up with ready-made forecasts. It's my opinion that even if a bettor intends to flat bet, he should still engage in this exercise, as it's only through practice that he will ever improve on the accuracy of his forecasts.
   
   Nevertheless, this is not a catch-all. One profitable source of betting value for some handicappers may result from systematic market mispricings. Given such a scenario, a bettor may very well correctly believe that a particular line holds value even without having any particular opinion on the fair pricing of the line himself. Now certainly these systematic mispricings can be difficult to identify and are by their very nature evanescent, but they could on occasion present a bettor with profitable angles that are strikingly difficult to quantify.
   
   
2. I'm going to have disagree a bit on point #2. If we do choose to accept the premise that "a bettor can’t reliably estimate the likelihood of winning a bet in sportsbetting", and we further accept that he does have reason to believe that the bet in question is profitable (and I hope I've successfully argued above why this could potentially be the case) then his bet sizing decision reduces to a function of his current bankroll, payout odds, risk tolerance, and expected overall average profitability. If the associated win probability of any particular event is unknown to the bettor (even though he may well know his own expected win probability across all his bets) then after adjusting for payout odds, there is no reason to treat any two bets differently. Now once again, I'll admit that this represents a decidedly suboptimal scenario, but as long as you accept the premise, it does follow logically.

So I think fundamentally the two of us are are in agreement on how a bettor should behave under optimal circumstances, and it's just some minor details of behavior under suboptimal circumstances on which the two of us disagree. But what I hope I've successfully demonstrated is why it could be the case that a long-term profitable bettor might be unable to successfully differentiate between levels of profitability. Furthermore, if you do accept that premise, then I think I've shown that this hypothetical bettor is behaving rationally by implicitly treating each bet (at a stated payout level) as having an equal likelihood of winning.

[Ganchrow]

In the long run I might agree. The only problem is that there may not be a long run if one uses full Kelly in sports betting.

One problem I've encountered is that even reliable methods can vary -quite severely- from season to season. Statistically, and based on sample size, these systems are very good; yet they may not produce as expected the next year. Quite possibly they need to be qualified further, but that could always be the case. If I had bet the farm, which I should have according to Kelly, on my best system from last season, I wouldn't be very upbeat right now. (My 'solution' is to play the system at lower betsizes until it kicks in, as I know it eventually will; a form of money management comparable to the positive deck approach, but not the same). Mind you, the system is still winning; just not at the expected rate!

In other words, there are no hidden elements in blackjack. To make the same assumption towards sports betting could be very dangerous.

There are always "hidden elements" in any endeavor governed by the laws of probability. If the team on which you've bet +14 gives up 31 points in the fourth quarter after six key third quarter injuries, that's simply an eventuality that you in general expect to help you as frequently as you expect to hurt you. That's white noise and is much a part of sports betting as it is of stock trading or indeed of life in general.

While forecast volatility is certainly a very real phenomenon, if the disturbances are white noise, then they should have no impact on the success of your estimation of the underlying event. In other words, if your forecasts are truly unbiased and consistent estimators of the underlying probabilities then the standard deviation of your bankroll under Kelly won't be impacted by these "hidden elements" -- as much as it might feel otherwise. To give a specific (albeit contrived) example: if you identify a particular bet as having a 55% likelihood of victory, while in reality some "hidden element" impacts the outcome of the event such that with 50% likelihood your bet really only has a 40% chance of victory and with 50% likelihood it really has a 70% chance of victory, well guess what ... neither your bankroll nor Kelly care about this on average, and the event behaves just as if it always had that 55% likelihood. A binomial event is a binomial event is a binomial event -- no matter what the underlying "hidden factors".

I think too often when people talk about "forecast volatility", they're unwittingly using it as euphemism for the failure of forecasts that are more the result of data mining than they might care to admit, and consequently have little predictive value. Remember: a trend that only occurs only in prime numbered years is probably not a trend worth investing in.

Another possibility is that these seemingly noisy forecasts represent underlying market factors that are either cyclical in nature (which should be fairly easy to identify) or are autocorrelated. This latter possibility could present difficulties for a bettor, but this is by no means a problem unique to sports forecasting and it simply behooves the prudent handicapper to perform the appropriate statistical tests before investing in any particular angle.

[Dark Horse]

To me personally, money management is based first on defense (not going broke), and only then on offense (maximizing profit). In theory you may not be able to go broke if you continue to bet any but 100 percent of bankroll, but that theory doesn't necessarily pay the bills.

The chance of a 60% system losing five times in a row is 0.4x0.4x0.4x0.4x0.4= 0.01. One in a hundred. In other words, just a matter of time.

For a -110 bet with a 60% winning percentage Kelly suggest to bet 16% of your bankroll. Let's say your bankroll is 100K. After five straight losses you would have 41.8K. Even if you now won five in a row you would only be back to roughly 82.5K. You're now 5-5 but have handed in 17.5% of your bankroll.

Streaks are real, and the best gamblers know how to ride them. This is an intuitive element not covered by math theory.

Just my two cents, but a full 100 percent.

[Ganchrow]

To me personally, money management is based first on defense (not going broke), and only then on offense (maximizing profit). In theory you may not be able to go broke if you continue to bet any but 100 percent of bankroll, but that theory doesn't necessarily pay the bills.

The chance of a 60% system losing five times in a row is 0.4x0.4x0.4x0.4x0.4= 0.01. One in a hundred. In other words, just a matter of time.

For a -110 bet with a 60% winning percentage Kelly suggest to bet 16% of your bankroll. Let's say your bankroll is 100K. After five straight losses you would have 41.8K. Even if you now won five in a row you would only be back to roughly 82.5K. You're now 5-5 but have handed in 17.5% of your bankroll.

You've said nothing at all untrue. Kelly maximizes the expected growth of a bankroll over time, and you're indicating that this is not your objective. That's completely fair and you're certainly not alone in your particular risk aversion. As I indicated in criticism #3 above logarithmic utility represents an insufficiently risk averse preference set for most bettors. That said, and for what it's worth, over the long term you simply can't expect your bankroll to appreciate in value any more quickly than it would with Kelly.

It's a characteristic of all percentage betting structures that while you expect to grow more quickly when using them, it's more difficult to recover from losses over the short term. Still, if you look at the results of the simulation I ran based upon JR Miller's Kelly experiment it's pretty clear that your results of a disastrous occurrence are considerably lower under Kelly than under a flat betting scheme with the same average size bet (a bet which to be fair is considerably higher than that with which you yourself would likely be comfortable). I should probably run the simulation with a lower flat bet size to compare the results. Any particular interest in selecting a faly bet size?

Streaks are real, and the best gamblers know how to ride them. This is an intuitive element not covered by math theory.

Now this, I'm sure you realize, I consider to be complete and utter hokum. However, if you wanted to come up a testable hypothesis I'd be more than happy to design an experiment to test it.

[JoshW]

Good discussion. I generally flat bet but often the limits I have of late make it so whatever the max is (not at the big books) is what I bet. Another consideration is that there isn't often time to come up with exactly how good a bet is. If you make 50 plus bets a day, that is certainly a consideration.

[Dark Horse]

Ganch, I lowered my bet size today based on intuition. Went 1-3, with one game left. Yesterday, with higher bet sizes, I went 7-2. Intuition is real and can be developed. But it is beyond intellect, and therefore cannot be analyzed in a way that would satisfy today's academics. However, it can be demonstrated. If you want to see it in action watch the show Mind Freak.

[Ganchrow]

Ganch, I lowered my bet size today based on intuition. Went 1-3, with one game left. Yesterday, with higher bet sizes, I went 7-2.

So basically what you're saying is that because you had more confidence that your bets were going to win last week you increased your bet size, and similarly because you had less confidence this week you decreased your bet sizes. That sounds suspiciously similar to adjusting your wagers based on the strength of your forecasts.

Come to think of it ... it sounds identical.

Intuition is real and can be developed. But it is beyond intellect, and therefore cannot be analyzed in a way that would satisfy today's academics.

Beyond intellect? What lies beyond intellect? Stupidity? It's not about "satisfying today's academics". It's about making a claim that can eventually either be proven true or false. But if such a claim isn't even being proffered ... well I would refer you to my signature, but I'm still waiting for an assertion.

However, it [intuition] can be demonstrated. If you want to see it in action watch the show Mind Freak.

Whether Criss Angel, the star of Mindfreak, happens to be a particularly intuitive fellow or not I really couldn't say ... but just to be clear Criss Angel is simply an illusionist who, like all magicians, uses trickery to entertain his audiences. One particulat trait he shares in with some of his predecessors, specifically magicians such as Penn & Teller, Lance Burton, David Copperfield, Doug Henning, James Randi, and Harry Houdini (beyond, of course, their extraordinary skill in the arts of both stage and close-up magic) is a complete and utter distaste for those who would attempt to dupe the public into believing absurd claims of paranormal ability.

Criss Angel himself has never claimed to be anything other than an illusionist and has spoken quite favorably of the skeptical James Randi Educational Foundation in general and specifically of its One Million Dollar Paranormal Challenge.

[TLD]

Thanks for the posts in this thread, Ganchrow.

I think our positions are pretty close. I would agree that the skill set to ascertain if a bet is above or below the line for it to have positive EV, and the skill set to ascertain if it is above or below other levels of value are non-identical, as you point out. Still, I would maintain the skill sets overlap so much that any differences are probably pretty trivial. Perhaps not, and perhaps there are examples of gamblers who are good at the first but not at the second, but I’m dubious. (Obviously there are a lot of flat betting gamblers who are good at the first and choose not to do the second, but that’s a different point.)

For one thing, the hypothetical person would have to be incapable of making even the most simplistic value comparisons. He could know that Eagles +3 has enough value to be bettable, and that Eagles +4 has enough value to be bettable, but would have to be clueless that one has more value than the other.

[sixzero]

I hope bumping this thread back up isn't frowned upon. I'm fairly new and threads like these have opened my eyes. Thanks again for everyone that contributed in this thread and those that continue to educate players like me. Thread saved as favorite.

[dj_destroyer]

Great thread... I'm running my formulas over at this thread and I'm using Kelly!

[Marginalis]

Even discounting the blackjack card counting world (where Kelly is used extensively), I am certainly aware of advantage bettors who succesfully use some derivative of Kelly.

lol

[Dark Horse]

Five years ago already. I hope I learned a few things since then. But maybe not. lol

The mistake made by many investors, not just sports bettors, is that, after a strong start during which their talent was awake to the present, they begin to rely too much on what once worked, and are not awake enough to the changing environment. That is the true hurdle. And very few have the flexibility to overcome it.

[wantitall4moi]

If math worked 5 years ago it should work now. That is what math is supposed to do, be a constant. But it is also why I say math has no place in sports betting beyond balancing your account.

Its like the BCS, that is all math based bullshit too. Supposed to take the emotions out of it. Well they have tweaked and changed and reformatted that thing every single year since they made it up and it hasnt been right yet., and the way math is used there it ia actually semi valid. Certainly more than how most people correlate it to sports betting.

If you have to tweak and change your math formula to adjust to changes then that basically shows math use is invalid. I know 2+2 has never changed. It is always 4. Has been for the 40 or so years I have used it since I learned it. I ahve yet to see a single math equation in regards to sports or sports betting that hasnt had some sort of change put into it to try and make it fit the answer the person is trying to get. Which is basically what math guys do. They know the answer, or at least the answer they want, then they try and work backwards to get an equation that gives them that answer.

But sine this thread was mostly based on the Kelly Criterion and not specifically math it gets lost a Little. But since Kelly doesnt even work in a vacuum, then I doubt it works for people who couldnt come close to determining their win expectation and thus formulate how much money they should be betting. Kelly is great for people to look at when theyre wining and it shows how much they 'could' have or 'should' have won if they were betting 'optimally'. But ultimatley it takes a simply 1unit, 2 unit, 5 unit appraoch and makes it even more convoluted, but cites math as the reason.

Bottom line is no one knows before hand their win probability. The best anyone can hope for is comparison shopping and looking to see what sort of 'advantage' they have with a certain price. If they look and see Yankees -140 at one book or one time of the day and at a different book or different time theyre -160 the probability of the Yankees winning didnt change. Just the price is different. So if Pinnacle has -145 and Greek has -160 Yankees have a higher probability of winning if you bet them at the Greek?

Or as some suggest would you bet more at a lower price? So if Pinny is -140 and Greek is -160 would you bet more at Pinny because of that? Or would you cite some Pinny 'lean' and be sacred to bet them because they 'wanted' you to.

The only advantage to any of this is you win more if you win or lose less if you lose, thats it. Nothing makes your chances of winning better or worse, other than your opinion. Be t your own personal opinion or your choice(opinion) to follow someone elses opinion.

When less than 1% of the people are able to make their own lines, or better yet prediction of a final score, they are therefore relying on the books lines to determine how they should bet. In a lot of cases they rely on those lines as well as how other people that are deemed 'experts' bet, and subsequently move the lines, to determine how they themselves would bet as well. Most people rely on so many outside influences that thinking they can determine something on their own is a joke.

[Teamprofit101]

wantiall4moi,

Awesome post!!!

[durito]

You are dumber than bigdaddyqh

[jgilmartin]

If math worked 5 years ago it should work now. That is what math is supposed to do, be a constant. But it is also why I say math has no place in sports betting beyond balancing your account.

Its like the BCS, that is all math based bullshit too. Supposed to take the emotions out of it. Well they have tweaked and changed and reformatted that thing every single year since they made it up and it hasnt been right yet., and the way math is used there it ia actually semi valid. Certainly more than how most people correlate it to sports betting.

If you have to tweak and change your math formula to adjust to changes then that basically shows math use is invalid. I know 2+2 has never changed. It is always 4. Has been for the 40 or so years I have used it since I learned it. I ahve yet to see a single math equation in regards to sports or sports betting that hasnt had some sort of change put into it to try and make it fit the answer the person is trying to get. Which is basically what math guys do. They know the answer, or at least the answer they want, then they try and work backwards to get an equation that gives them that answer.

But sine this thread was mostly based on the Kelly Criterion and not specifically math it gets lost a Little. But since Kelly doesnt even work in a vacuum, then I doubt it works for people who couldnt come close to determining their win expectation and thus formulate how much money they should be betting. Kelly is great for people to look at when theyre wining and it shows how much they 'could' have or 'should' have won if they were betting 'optimally'. But ultimatley it takes a simply 1unit, 2 unit, 5 unit appraoch and makes it even more convoluted, but cites math as the reason.

Bottom line is no one knows before hand their win probability. The best anyone can hope for is comparison shopping and looking to see what sort of 'advantage' they have with a certain price. If they look and see Yankees -140 at one book or one time of the day and at a different book or different time theyre -160 the probability of the Yankees winning didnt change. Just the price is different. So if Pinnacle has -145 and Greek has -160 Yankees have a higher probability of winning if you bet them at the Greek?

Or as some suggest would you bet more at a lower price? So if Pinny is -140 and Greek is -160 would you bet more at Pinny because of that? Or would you cite some Pinny 'lean' and be sacred to bet them because they 'wanted' you to.

The only advantage to any of this is you win more if you win or lose less if you lose, thats it. Nothing makes your chances of winning better or worse, other than your opinion. Be t your own personal opinion or your choice(opinion) to follow someone elses opinion.

When less than 1% of the people are able to make their own lines, or better yet prediction of a final score, they are therefore relying on the books lines to determine how they should bet. In a lot of cases they rely on those lines as well as how other people that are deemed 'experts' bet, and subsequently move the lines, to determine how they themselves would bet as well. Most people rely on so many outside influences that thinking they can determine something on their own is a joke.

wantiall4moi,

Awesome post!!!

Yeah!!! No one could possibly ever be better than a sportsbook at estimating probabilities!!! If you have a bigger edge, why would you ever want to bet more money??? That's just crazy talk!!!

[FourLengthsClear]

You are dumber than bigdaddyqh

-110 he is BigDaddyQH.

The schtick might be different, no Vegas syndicate here, but the writing style is the same.
The jumbled up 'thinking' is familiar too.

[Teamprofit101]

If math worked 5 years ago it should work now. That is what math is supposed to do, be a constant. But it is also why I say math has no place in sports betting beyond balancing your account. Its like the BCS, that is all math based bullshit too. Supposed to take the emotions out of it. Well they have tweaked and changed and reformatted that thing every single year since they made it up and it hasnt been right yet., and the way math is used there it ia actually semi valid. Certainly more than how most people correlate it to sports betting. If you have to tweak and change your math formula to adjust to changes then that basically shows math use is invalid. I know 2+2 has never changed. It is always 4. Has been for the 40 or so years I have used it since I learned it. I ahve yet to see a single math equation in regards to sports or sports betting that hasnt had some sort of change put into it to try and make it fit the answer the person is trying to get. Which is basically what math guys do. They know the answer, or at least the answer they want, then they try and work backwards to get an equation that gives them that answer. But sine this thread was mostly based on the Kelly Criterion and not specifically math it gets lost a Little. But since Kelly doesnt even work in a vacuum, then I doubt it works for people who couldnt come close to determining their win expectation and thus formulate how much money they should be betting. Kelly is great for people to look at when theyre wining and it shows how much they 'could' have or 'should' have won if they were betting 'optimally'. But ultimatley it takes a simply 1unit, 2 unit, 5 unit appraoch and makes it even more convoluted, but cites math as the reason. Bottom line is no one knows before hand their win probability. The best anyone can hope for is comparison shopping and looking to see what sort of 'advantage' they have with a certain price. If they look and see Yankees -140 at one book or one time of the day and at a different book or different time theyre -160 the probability of the Yankees winning didnt change. Just the price is different. So if Pinnacle has -145 and Greek has -160 Yankees have a higher probability of winning if you bet them at the Greek? Or as some suggest would you bet more at a lower price? So if Pinny is -140 and Greek is -160 would you bet more at Pinny because of that? Or would you cite some Pinny 'lean' and be sacred to bet them because they 'wanted' you to. The only advantage to any of this is you win more if you win or lose less if you lose, thats it. Nothing makes your chances of winning better or worse, other than your opinion. Be t your own personal opinion or your choice(opinion) to follow someone elses opinion. When less than 1% of the people are able to make their own lines, or better yet prediction of a final score, they are therefore relying on the books lines to determine how they should bet. In a lot of cases they rely on those lines as well as how other people that are deemed 'experts' bet, and subsequently move the lines, to determine how they themselves would bet as well. Most people rely on so many outside influences that thinking they can determine something on their own is a joke.

Yeah!!! No one could possibly ever be better than a sportsbook at estimating probabilities!!! If you have a bigger edge, why would you ever want to bet more money??? That's just crazy talk!!!

I'm sorry for not being specific on which part of his post I agreed with. I was agreeing with this part....

If you have to tweak and change your math formula to adjust to changes then that basically shows math use is invalid. I know 2+2 has never changed. It is always 4. Has been for the 40 or so years I have used it since I learned it. I ahve yet to see a single math equation in regards to sports or sports betting that hasnt had some sort of change put into it to try and make it fit the answer the person is trying to get. Which is basically what math guys do. They know the answer, or at least the answer they want, then they try and work backwards to get an equation that gives them that answer.

[hutennis]

I'm sorry for not being specific on which part of his post I agreed with. I was agreeing with this part....

If you have to tweak and change your math formula to adjust to changes then that basically shows math use is invalid. I know 2+2 has never changed. It is always 4. Has been for the 40 or so years I have used it since I learned it. I ahve yet to see a single math equation in regards to sports or sports betting that hasnt had some sort of change put into it to try and make it fit the answer the person is trying to get. Which is basically what math guys do. They know the answer, or at least the answer they want, then they try and work backwards to get an equation that gives them that answer.

I totally agree with this part too.

As far as somebody somewhere being able to beat mature, liquid market with huge trading volume? Sure.
In fact, huge number of participants makes it an absolute certainty that somebody somewhere will do that just by chance alone.

I'm not saying that people who win "long term" (would love to see definition of "long term" btw) are not handicapping geniuses. I'm just saying that there is an alternative explanation for their success.

And, since alternative explanation requires much less assumptions, it should be a default one.
Anything above that calls for a proper examination and prove to be considered seriously.
Just like with Big Foot and UFO.

[wantitall4moi]

Yeah!!! No one could possibly ever be better than a sportsbook at estimating probabilities!!! If you have a bigger edge, why would you ever want to bet more money??? That's just crazy talk!!!

Books arent in the business to predict scores if they did you would see a lot of double digit spreads. They simply post numbers that make people bet both ways or as close to both ways as possible. In some cases that includes putting up a 'too good to be true' number where some peopel will either lay off the game or worse bet the other side thinking the book is trying to trick them. How many 'trap game' posts have we heard, or how many after the fact 'trap games' have people cited?

Sports are unpredictable sometimes games come out in a way you cant imagine, those are the ones books love because it makes them look really smart. But hen you see a line or odds or a spread or a total that doesnt make sense or looks too good to be true you bet it. That doesnt mean it was too good, it just means you have a strong opinion.

Now should you bet more? Depends on your ability. if youre a coin flipper type person then no. Since the vast majority of bettors are coin flippers they technically, even according to kelly shouldnt be betting anything. The only miniscule advantage they have is the price they get. More often than not their so called advantage is strictly short term as well.

But the short answer is no one no where has a model or an equation or a system that can predict sports outcomes well enough to say what there TRUE advantage is. If you have a model or a system or an equation that says Lakers should be +12 tonight and the line is +4 what do you do? Do you bet two times as much as a game where it might say the line should be +6? If your system, model, equation says 12 point differential and it is a 7 point differential is that a 'win' for you? Or how do you weight it? It had the 'correct' side, but was still way off on the prediction. So down the road you have a game that is lined -4 but has the other side favored by 3. (7 points different) The fav wins by 3, another 'cover' but again way off the prediction. Is that still a valid result? Models are going to be right and wrong, the aim is to get the right result from the team the model picked. But is the degree of difference valid? I would say not even close. As in the degree of difference between a personal opinion/model result cant be verified before hand. It would take a lot of games and plays to build a big enough sample size to even see if the system/model/equation was even valid to use let alone use with the advanced goal of trying to determine strength of probability.

Like I have said a thousand times over it all comes down to volume. If you can pick more games and see more profits than losses you show the ability to make money. The more games you play and STILL show a profit the more you show you have the ability. ANYONE can have some short term successes, even small volume success. But after tens of thousands of plays is where true ability starts to show. Most people dont last that long s they go broke a long time before then. But professional and 'expert' gamblers can have thousands of plays a year and still show a profit. But most of those are due to their ability to get winning odds on both sides of the game. Which turns them into the bookmaker and takes gambling completely out of the equation.

That is also addressed with Kelly but the fact is when you cant lose then you should bet 100% of your money on the play. So you dont need an equation to tell you that. Those are the only true +ev plays people can make. Arbitrages and scalps basically. If you have 20K bankroll and see two line of -120 and +133 then you can put all 20K into those two lines and have ZERO risk of loss. Now you could gamble and try and justify why one line was 'sharper' or smarter' than the other and try and make a case to bet the best number on the smartest side, but that is no guarantee. You obviously have a much bigger profit if you happen to be right, but if youre wrong and bet too much of your bankroll you take a hit. So you can either take a guaranteed 500 or so bux for your 20K invested and move on or you could bet whatever percentage of your bankroll you feel or kelly calls for and go that route. But basically if it is calling for less than a 2.6% bet sze youre better off scalping it.

Obviously scalps like that are non existent these days. But you can get odds like that and better if you take leads and look to buy back. Those do have some risk but with people who know what theyre doing it can be less than a 1% risk.

[jolmscheid]

As ganch and 4LC said, if you know your edge, playing half kelly or even quarter kelly is best for growth...

[wantitall4moi]

As ganch and 4LC said, if you know your edge, playing half kelly or even quarter kelly is best for growth...

NO ONE KNOWS THEIR EDGE. anyone that claims they do is lying or fooling themselves. At least in sports betting. In cards or dice or dominoes or something that has a finite set of possibilities, sure you can get close (dead cards even make cards an iffy calculation).

Sports is sports and the price or spread in no way shape or form changes your edge. All it does is give you a better price. Even in spread based sports, the point spread is so small an actual factor in determining wins and losses it is almost negligible. Adjusting vig on those spreads mkes it even more convoluted.

The Superbowl was a 'perfect' example. Guys laying -150 for +3.5, guys laying -130 for +3, some guys taking +115 on the ML. Or even +2.5 -105. So guys had to first 'value' the spread and how much to them each increment was 'worth' and then play accordingly. Obviously in the end the spread was meaningless, as it so often is. But that sense of security is what people are paying for. And most, even the so called smart ones just dont get it.

[Dark Horse]

NO ONE KNOWS THEIR EDGE. anyone that claims they do is lying or fooling themselves. At least in sports betting. In cards or dice or dominoes or something that has a finite set of possibilities, sure you can get close (dead cards even make cards an iffy calculation).

Players develop their own styles, and if they all agreed the bookmakers would be out of business. People can closely estimate their winning expectation. Fractional, instead of full, Kelly covers the margin for error. And sports betting can add certain advantages that cards can't. No model will stay hot for an entire season. Comparing each season to a horse race, is the model highly accurate early in the race, as the horses are in the first turn? This has some value, but is likely to fizzle out. Much preferable is when model has a strong grip late in the season, as the horses are in the final turn. In that scenario congratulations are in order, because that type of grip is very likely to last for the remainder of the season. And it would be a mistake to assume that this type of success has any bearing on the start of the next season. People who understand this have a significant advantage over those who pile all seasons on one big heap. If you have 100 proven models running in the background, do you think they're all going to be in sync with the season in progress? If you wish to burn a hole in a piece of paper with a magnifying glass would you flat bet or increase your bet size when the sun comes out of the clouds?

[jgilmartin]

This is about as productive as trying to debate a creationist on the age of the planet. If one believes it impossible for a probability estimation to have any value whatsoever, the other points are totally irrelevant.

[FourLengthsClear]

This is about as productive as trying to debate a creationist on the age of the planet. If one believes it impossible for a probability estimation to have any value whatsoever, the other points are totally irrelevant.

Indeed. I have wasted far too much time with wanty already. Life is too short.

[sharpcat]

NO ONE KNOWS THEIR EDGE. anyone that claims they do is lying or fooling themselves. At least in sports betting. In cards or dice or dominoes or something that has a finite set of possibilities, sure you can get close (dead cards even make cards an iffy calculation).

Sports is sports and the price or spread in no way shape or form changes your edge. All it does is give you a better price. Even in spread based sports, the point spread is so small an actual factor in determining wins and losses it is almost negligible. Adjusting vig on those spreads mkes it even more convoluted.

The Superbowl was a 'perfect' example. Guys laying -150 for +3.5, guys laying -130 for +3, some guys taking +115 on the ML. Or even +2.5 -105. So guys had to first 'value' the spread and how much to them each increment was 'worth' and then play accordingly. Obviously in the end the spread was meaningless, as it so often is. But that sense of security is what people are paying for. And most, even the so called smart ones just dont get it.

Why do you continue to post?

You have already admitted many times before that you no longer wager on sports because once the glory days of arbitrage and bonus whoring dried up you have been unable to adapt and can no longer profit from sports wagering. So just because you are unable to adapt you feel the need to come to the boards everyday and tell everyone that because you are no longer able to profit that it means nobody can????? I hate to break it to you but the way you were making money in the glory days did not make you a genius, any idiot should have been able to arb and whore their way into profits. Get over yourself!!!!!

Sports is sports and the price or spread in no way shape or form changes your edge. All it does is give you a better price.

By far the most idiotic rant I have ever heard from you

If I like the yankees and book A is offering them at -220 and book B is offering them at -190 you are actually going to sit here and tell me that I do not have an increased advantage long term with lower odds?

WTF would you flat bet and not exploit your increased advantage????

[PackerBacker]

I have always viewed flat betting as somewhat of a bumpy ride. Alot of uphills with flat betting.