[austintx05]

I have always flat bet, in my short 2 years of wagering on sports. As many of you know, I am fairly new at this compared to the veterans, but the reason I have always flat bet is because the day you go 1-1, you still have the potential to be down money if the 1 loss you have was a 2+unit play vs the 1 unit winner. I also realize that there are certain pays that might warrant a multiple unit wager and have read some thoughts at other sites that are felt to be leaving money on the table.

So as a rookie to the sports betting world, I would like to hear your thoughts on this issue.

Thanks

[Justin7]

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?

[TLD]

The reason I have always flat bet is because the day you go 1-1, you still have the potential to be down money.


First thing: When you realize something you’re doing in sportsbetting is motivated by a rule like this, examine the rule itself to see if it’s justified.

So you want to use a betting system where if you bet multiple games and win at least 50% of them, it isn’t possible to lose (other than a little juice). Why? Can you show logically, mathematically, that you will do better over the long run adhering to such a system than you would do with a system that violates this rule?

Or is it purely a psychological thing—the “feeling” of going 1-1 and losing bothers you more than the prospect of going 1-1 and winning pleases you? Something just doesn’t seem right about it?

If the latter, then this isn’t a justified sportsbetting rule but simply a personal mental block you picked up along the way, and one that will pointlessly restrict your strategizing.

[TLD]

And in answer to the original question:

Optimum bet size is a function of three factors: The size of your bankroll, the likelihood of the wager winning, and the odds at which you’ll be betting it.

If for every bet you ever make all three of these factors are precisely identical, then flat betting is correct.

Otherwise it is not.

[Mudcat]

I have been betting on sports for about 25 years and my stand on flat betting is: that's all I do. I determine a unit size and bet it like clockwork when I have an angle.

I periodically reevaluate and change my unit size but that still qualifies as flat betting.

I used to vary my bet sizes. In fact I have tried MANY systems of variable bet sizes over the decades. Flat Betting is where I have arrived.

There is good theory behind the use of variable bet sizes. I don't doubt there are people who do it properly (mathematically) and with discipline. However my observation has been that most people use variable bet sizes to accomodate their own psychology. They will tell themselves they have good capping reasons for placing a larger bet - and probably really believe themselves - but really they are just rationalizing facts to fit the conclusion they want to reach, and they are really chasing/impatient/feeling invincible/whatever.

I have done my share of that over the years.

As I say though, I don't doubt there are people who use variable bet size properly.

If I ever found an angle that was just SO MUCH BETTER than my other angles, I could see myself using variable bet sizes. For example if I found an NFL angle that produced a unit profit for every 3 bets and it was based on a really big sample - like hundreds and hundreds of games - that would get my attention and I would bet more than I would on my normal angles which are all in the same range (producing roughly a unit every 8-12 bets.)

But should I bet more on an pointspread angle that has hit at 57% for a long time than one that hits at 55.5% ? I guess maybe. Should I bet 3% more; should I bet three times as much? I don't know. I just place my bets and I'm happy with the results.

I simply don't have any spectacular angles that are way out of that range and I just grind it out with flat bets.

[jjgold]

There is no such thing as one game better than the other, you have an equal shot at each therefore you should bet the same exact amount on each game.

[austintx05]

Thanks for the replies.

[koko]

If you're a winning player, you should have an edge on every wager you make.

Sometimes you'll have a bigger edge then other times. The only reason to not bet more when you have a bigger edge (and less when you don't) is if you are looking to trade away some EV to reduce variance.

If your roll is big enough to handle the variance, there is no good reason to willingly give up EV to reduce variance. If you can't mentally handle the swings your betting to big - regardless of the size of your roll.

Bet more when you have a bigger edge. Bet less when your edge is smaller. Pick bet sizes that your roll can handle.

If you do these three things, you'll maximize your EV in the long run.

[austintx05]

Good stuff koko

There have been some games during this NCAAF season where I felt I had the games nailed and sure enough they went that way, but I never strayed away from playing every game at 1 unit. I was just trying to stay disciplined and focused.

[Ganchrow]

But should I bet more on an pointspread angle that has hit at 57% for a long time than one that hits at 55.5% ? I guess maybe. Should I bet 3% more; should I bet three times as much? I don't know.

Well, if you're betting a -110 line set at -110, then your edge on the 55.5% proposition is 5.95% and your edge on the 57% proposition is 8.82%.

So if your preferences are of a logarithmic type (ŕ la Kelly) and you're betting 1% of your bankroll on the 55.5% prop at -110, then you should be betting a little less than 1.5% of your bankroll (8.82% / 5.95% ≈ 1.48%) on the 57% prop at -110 (assuming the individual bets were not contemporaneous).

[Ganchrow]

If you're a winning player, you should have an edge on every wager you make.

Sometimes you'll have a bigger edge then other times. The only reason to not bet more when you have a bigger edge (and less when you don't) is if you are looking to trade away some EV to reduce variance.

If your roll is big enough to handle the variance, there is no good reason to willingly give up EV to reduce variance. If you can't mentally handle the swings your betting to big - regardless of the size of your roll.

I've noticed that it's demonstrably more difficult for neophyte handicappers to determine unbiased and consistent estimators of the likelihood of any particular bet's winning. As such, an advantage bettor may actually be giving up EV by utilizing a betting scheme such as Kelly.

If lowering the variance of your bankroll's growth is of great concern to you (and it certainly should be) and you frequently find yourself betting across a wide variety of money lines, then you might want to consider moving away from fixed unit staking towards a "fixed-profit" staking plan. Fixed-profits staking refers to betting to win a constant amount on all bets. So in other words, if a fixed-profits staker were to bet 1 unit at a line of +100, he would be betting 1.1 units on a money line of -110, and ˝ of a unit on a a money line of +200.

Joseph Buchdahl in Fixed Odds Sports Betting: Statistical Forecasting and Risk Management demonstrates how a bettor engaging in fixed-profits staking can reduce both his standard deviation and his risk-of-ruin versus a flat bettor with the same average bet size.

(Fixed-profits staking, btw, is actually implicit in Kelly betting.)

[Mudcat]

Well, if you're betting a -110 line set at -110, then your edge on the 55.5% proposition is 5.95% and your edge on the 57% proposition is 8.82%.

So if your prefernces are of a logarithmic type (ŕ la Kelly) and you're betting 1% of your bankroll on the 55.5% prop at -110, then you should be betting a little less than 1.5% of your bankroll (8.82% / 5.95% ≈ 1.48%) on the 57% prop at -110 (assuming the individual bets were not contemporaneous).

Sure, that makes sense. If I was confident that, because an angle has hit at a certain percent for the last 20 years, that means it will continue at that exact percentage for the next 20, then I might be more conscious of fine-tuning points like that.

The problem for me is that sports-betting is an ever-changing system. The sharpness of the public, linesmaking and rule changes can all make subtle - or not so subtle - differences in the expectation of a betting angle. And often I don't become aware that a change has happened until after I've placed a whole bunch of bets and am looking back at the results.

Because I have had angles that have not only wavered in reliability but some that were reliable for many years and then literally dropped of the face of the earth and became nothing, I don't see things as being quite as precise as that. I see a 57% prospect beside a 55.5% prospect and my reaction is, 'Same thing.'

Now if I started looking at some 66% prospects, that would be different.

I do appreciate your comment though and will give it more thought.

[TLD]

I think it’s a mistake to think of flat betting as the default strategy, or the appropriate strategy for situations of uncertainty.

When you flat bet, you aren’t being agnostic as to the degree of your edge; you’re taking a stand that the edge is equal for all bets.

[Dark Horse]

Flatbetting is best for beginners. Least amount of risk. No real need to move away from it, unless you're interested in maximizing profit. When that becomes important enough, you'll find a way to vary betsizes in the best way for you. Whatever you do, do not use Kelly, or trust anybody that tells you to bet 15% of your bankroll on one bet.

This link has some good points.
PRO INFO SPORTS Money Management

[Patrick McIrish]

When you flat bet, you aren’t being agnostic as to the degree of your edge; you’re taking a stand that the edge is equal for all bets.

Agree. The only time flat betting should really be encouraged is when you are new and have no idea how to assess what kind of edge you have. At that point the most important thing for you should be protecting the bankroll. Stick to betting 1% of your roll and try to keep 8% or less in action each day and you will stick around a while.

That is why a lot of people don't really get down fairly common principles like Kelly and so on, they don't stick around long enough to do so. Stick to flat betting when you are starting out but be learning everything you can, believe me you will know when it's time to start varying your bet size.

[onlooker]

Flat betting is fine. I usually have a set amount, unless I get some good information on a game, then I usually double or triple up on it.

But I do increase my units from time to time as my bankroll increases.

[Ganchrow]

Flatbetting is best for beginners. Least amount of risk.

As I've mentioned, fixed profits staking is associated with lower standard deviation in finishing bankroll after a given number of bets (and lower risk-of-ruin) than traditional flat betting.

Whatever you do, do not use Kelly, or trust anybody that tells you to bet 15% of your bankroll on one bet.

This link has some good points.
PRO INFO SPORTS Money Management

There's no question that if a bettor is able to accurately judge his expected win probability for a given bet then wagering the Kelly stake will result in the highest possible expected bankroll growth over time. The link you've provided mentions "Peak Profit Percentage" which, from the two examples given, appears to be nearly identical to Kelly staking.

[Dark Horse]

Nearly identical has the same ring as almost perfect. You know that Kelly advocates much higher percentages than the ones suggested in the article.

[Scorpion]

Debunking the Kelly Criterion

[Ganchrow]

Nearly identical has the same ring as almost perfect. You know that Kelly advocates much higher percentages than the ones suggested in the article.

The formula that appears to be used in the article is:

PPP = p * (o-1) - (1 - p)

where p = probability of the bet winning and o = the decimal odds offered. (So in other words, the PPP is just the % edge.)

While the Kelly stake is given by:

k = PPP / (o-1)

which means that in the case of a line of -110, Kelly will be exactly 10% higher than PPP. So in the example given of a win % of 55%, the PPP is 5% and the Kelly stake is 5.5%. For money lines > +100, the PPP would actually be higher than the Kelly stake.

The difference between PPP and Kelly is precisely analogous to the difference between fixed-stake and fixed-profit unit wagering.

[Dark Horse]

I seem to recall that the author of that article suggests betting between 2-5% of bankroll. Whether or not the formula is identical or not, I know that Kelly suggested much higher risks.

[Ganchrow]

Debunking the Kelly Criterion

I thought this horrendous piece of misinformation had been put to rest some time ago. If you were to search, I'd bet you could probably find a dozen critiques of this particular piece of JR Miller nonsense.

The flaw in Miller's "reasoning" lies in this statement:

The only way to fairly compare the Kelly system (or any other progressive betting scheme) to flat betting is to use a flat bet the same size as the average size of all the Kelly bets. That way, you’re risking the same total amount against the same overall won-lost results. No fair risking more money overall with one system than the other. That obviously skewers the results.

So what JR Miller is arguing is that in order to compare flat and Kelly betting you'd need to use the same average bet size. However, your actual average bet size under Kelly isn't known until all bets have been placed. In other words, Miller's thesis would require knowledge only available ex-post, ex-ante.

If a particular realization of the experiment outlined by Miller happened to result in large profits (implying a larger average bet size) then Miller is implicitly stating that the player should have bet the larger size from the start. Conversely, after a poor realization of the experiment, Miller is saying the player should have bet less from the start.

So Miller tells us that when a player a wins, he should have bet more; and when he loses, he should have bet less. Yeah, that's real sage advice. The advantage of Kelly is that provides the optimal bet size to maximize future expected bankroll growth, based on information available at the time the bet is placed.

I'll agree Miller is correct that if we all had time machines we could construct a betting system significantly more effective than Kelly at maximizing expected bankroll growth. But lacking that time machine ... Kelly provides the single best method for maximizing expected bankroll growth.

[Ganchrow]

I seem to recall that the author of that article suggests betting between 2-5% of bankroll. Whether or not the formula is identical or not, I know that Kelly suggested much higher risks.

The author starts the article by suggesting betting the PPP (the edge, really). In the case of a bet made at -110, this would be 90% of the Kelly stake; and in the case of a bet made at +200 would be 200% of the Kelly stake.

Later in the article, the author does suggest capping max bet size at 5%. His rationale for chosing the particular figure of 5% is not made readily apparent.

[Dark Horse]

Do you know any gamblers who bet full Kelly (and haven't gone broke)?

[Ganchrow]

To be clear, there are definitely legitimate criticisms of the Kelly Criterion:

1. Kelly requires a player to fairly accurately estimate his win probability over a wide swath of bets. This can be especially difficult for bettors that base their handicapping decisions off of largely qualitative considerations.
2. The standard Kelly formula does not apply to multiple bets placed simultaneously. While it's a fairly straightforward computational task calculating proper Kelly stakes for a small number of contemporaneous bets, as the number of bets increases it would fairly quickly becomes necessary to utilize a hardcore optimization engine.
3. Kelly serves to maximize expected bankroll growth -- but who's to say that this accurately reflects any given bettor's preferences? In fact, the only bettors for whom the maximization of expected bankroll growth serves as a proper objective function are those with logarithmic utility. But evidence from financial economics suggests that logarithmic utility represents an insufficiently risk averse utility function for most investors. Insofar as this also holds for a given bettor, Kelly will suggest bet sizes that are theoretically too large based upon the bettor's actual preferences. (Note that this does not preclude the use of a "Kelly-like criterion" that might employ a potentially more appropriate (although less computationally friendly) utility function.)

So these are the real digs at Kelly, There may well be some other minor issues that have escaped me, but for the most part, these three issues represent the only real theoretical problems with Kelly staking. The significance of these issues is by no means minor and may very legitimately stand in the way of an advantage sports bettor using Kelly.

But that's really about it. JR Miller simply has it wrong. If you aren't using Kelly ... just make sure you aren't using it for the right reasons.

[Ganchrow]

Do you know any gamblers who bet full Kelly (and haven't gone broke)?

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.

[nosuzieno]

Do you know any gamblers who bet full Kelly (and haven't gone broke)?

While the optimum bet for the greatest growth of bankroll is making the full bet suggested by the Kelly criterion, this can produce a volatile result. There is a 1/3 chance of halving the bankroll before it is doubled. A popular alternative is to bet only half the amount suggested which gives three-quarters of the investment return with much less volatility. Where money would accumulate at 10% compound interest with full bets, it still accumulates at 7.5% for half-bets.

Certainly over-betting beyond that suggested by Kelly is counter-productive as the long run return will fall, dropping to zero (with the loss of all the bankroll) when the Kelly bet is doubled. Using half-Kelly bets also safeguards against being ruined by unconsciously overbetting, as it can be easy to over-estimate the true odds by a factor of two...

To the original question: If you bet multiple games each day and pick more winners than losers over the long run, I would suggest you stay with flat betting at least for the time being. In the future, perhaps track various bettor strategies at the same time and see if you find a niche. Martingale my favorite Good luck!~

[RickySteve]

Do you know any gamblers who bet full Kelly (and haven't gone broke)?

This reckless log-of-expected-bankroll-maximizing lunatic does OK.

This wacko too.

[Lucas]

This debate is nice, but in many cases academic - a bet is often as big as bookie allows.

[RickySteve]

This debate is nice, but in many cases academic - a bet is often as big as bookie allows.

You should consider opening an account with more than one bookmaker.

[pags11]

I agree a lot with what Mudcat has written...I use that strategy mainly for college hoops...all of my plays are for a unit...occasionally (like maybe 5 times a year, I'll make a 2 unit or 1/2 unit play)...that's worked best for me in basketball...in football, as I've stated before, I play all of my games for the same ammount each week, but the size of the unit starts out small, gets bigger, and then gets smaller towards the end of the season...

[imgv94]

I find it hard to respect people who flat bet. Don't get me wrong I respect pags a lot but believe he is doing himself an injustice sometimes when only wagering a set amount when he has a really solid wager going.

Mudcat I sure would like to see your 57% angles you have? I believe to this day you haven't posted any angles and don't give me your B.S that you don't want to give out your secrets cause no one believes that B.S..

I think you are bullshiet..

[Dark Horse]

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.

Some derivative. As in less than full Kelly.

I don't know any serious sports bettor who uses Kelly without toning it way down. Meanwhile, less experienced sports investors could easily walk into the trap of such a 'respected' theory, especially if backed by respected posters who only later qualify that it was about blackjack.

[Dark Horse]

In blackjack a 60% chance of winning is based on the present deck(s). A deck is a very limited 'universe'. The positive expectation here is mathematical.

In sports betting, where the bet is made before the start of the game, a 60% winning expectation is based not on the present, but on the past (stats, etc). It is only hoped that the game will have a direct resemblence to the past. And naturally, a sports game is not a limited universe like one or a few decks of cards. In sports the possibilities for each single play are vast, and much less predictable.

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 the price during a live betting event would not change, even with one team way ahead, that might compare to betting into a positive deck; and open the door to full Kelly...).

[Ganchrow]

Some derivative. As in less than full Kelly.

I don't know any serious sports bettor who uses Kelly without toning it way down. Meanwhile, less experienced sports investors could easily walk into the trap of such a 'respected' theory, especially if backed by respected posters who only later qualify that it was about blackjack.

Actually, by Kelly derivative I was referring to Kelly staking adjusting for simultaneous bets rather than some largely artifical "partial Kelly" construct.