[justbet]

i have spent quite a bit of time looking at the articles at the professionalgambler.com and it seems that J.R. Miller is convinced that the star system is poor money management. i noticed that several of the people who have pics in the wagering section use a star system, and i was wondering why. i have to agree with j.r. that it is smarter bankroll management to simply bet a set unit on every game. i am wondering what everyone's opinions are, and whether you think it is smart to use a "star" system.

[Razz]

The closest thing that I can compare the star system to is counting cards in blackjack. Is it poor money management for the card counters to bet 10 unit on a hand when the true count is highly in their favor, and only bet 1 unit on a hand when the true count is against them? I don't think so.
To me, some plays are invariably stronger than others. While a 1* play is still worth playing, I do not expect it cover as often as a 2 or 3 unit play. The cure is either to rank plays, or be extremely selective, and only play the creme de la creme of plays (the ones that I would consider a 2* or 3* play). If you are going to play all plays for the same amount, the latter method is necessary, but to say the star system is poor money management borders on stupidity.

[Mudcat]

I think you have to file that one under, to each his own.

I just use one bet size. I've been operating that way for quite awhile and I don't foresee that changing - although, theoretically speaking, if I ever discovered an angle that had a significantly higher winning expectation than I'm used to, anything is possible.

I have definitely seen posting forum cappers playing around with multi-star plays who don't have a clue what they're doing. They would be wise to listen to J.R. (The truth is, they're probably not even really betting - they're just paying posting forum games).

But for a real bettor, Razz's card counting analogy makes perfect sense.

[JayEgdarWho]

I think you have to file that one under, to each his own.

A good player MAY be able to use the star system to his advantage, but if we are honest I think we will recognize that a mediocre player is probably likely to:

(1) use the lower-star designation as an excuse for playing some no-advantage games just so that he can be in action; and

(2) use the higher-star designation much more often when he is on tilt, which is not what God intended.

And while Razz makes a decent point, I think it is also true that, even in most BJ systems, bet variance is a two- or three-tier scheme at the most -- with the dividing lines falling at ADVANTAGE/NO ADVANTAGE/DISADVANTAGE. If you have an advantage, no matter what it is, many systems are going to recommend the same bet size on every "ADVANTAGE" bet.

BJ is different because you have to play even when you have no advantage or a disadvantage. In a sportsbook you do not (unless you are desperate for action).

A smartass might say that J.R. Miller IS recommending varying your bet size:
ADVANTAGE? Bet 1 unit
NO ADVANTAGE or DISADVANTAGE? Bet 0 units

[MrX]

I read the article there on the Kelly Critereon. Horrible. Really horrible article. Nonsense math, nonsense reasoning. Just ridiculous. I didn't read any other articles after that. Maybe there's some good stuff there, I don't know, but what I do know is that the arguments made against the Kelly Critereon are not sound.

If you have a sound reason to believe that one pick is worth twice as much as another, and that both picks are positive expectation, and you are attempting to maximize bankroll growth, you are correct to bet both picks and to bet more on the better pick, period. It's a proven mathematical fact.

That said, star systems and the like are often abused in an attempt to recover a bad day/week/month. If you start betting more on some games for emotional reasons like this, it certainly is poor money management.

If anyone wants to read a very good article on the Kelly Critereon as it relates to sports betting, blackjack, and the stock market, they should find a book called "Finding the Edge" (2000). Amazing stuff in there, but not for the faint of heart.

[JoshW]

Great comments by everyone in the thread. In major sports, I tend to think close to flat betting is called for. But if you are a winning player with access to lots of different odds. When you find odds that are far enough off on plays you already like, double size bets and larger are justified. The same could be said for secondary sports where you might have a huge edge and are justified in betting even more. I have found situations that would justify putting 20% of my bankroll on them. I can't because of limits but in these cases I just try to get as much money as I can down. When you are looking at 50% ROI, is tough not to bet as much as you can. But it all comes down to finding winners and edges.

I am recommending the book Fortunes Formula that actually made some non-fiction best seller lists last year. Book is largely about Kelly Criterion and provides some very accessable analysis of betting size.

[MrX]

I should add, making only single unit bets is not inherently bad by any means. First of all, I think some, maybe many, handicappers have done solid research showing them that when certain criteria are present, a bet will be positive E.V. There may not be enough data to give more conifidence to any one wager over another. In a case like this flat betting is clearly proper strategy, although your bets should increase as your bankroll grows.

In a more common scenario, even if you know certain bets have more value than others, you may still want to flat bet to reduce fluctuations. Betting for longterm maximum bankroll growth comes at a cost. Most people find that they are not comfortable with the swings involved with adhering rigidly to the Kelly Critereon. Most will tone it down either by betting 60-80% of the actual value determined by Kelly, or by limiting the spread of bet values (closer to flat betting, thus reducing the impact of certain high-value games), or both.

[Mudcat]

(2) use the higher-star designation much more often when he is on tilt, which is not what God intended.

That's what I like. A post with some authority.

[JayEgdarWho]

(2) use the higher-star designation much more often when he is on tilt, which is not what God intended.

That's what I like. A post with some authority.

The "he" in my post being the player. In case there is any confusion.



To my knowledge, "He" has not been on tilt since the Old Testament.

[justbet]

or since hurricane katrina

[Ganchrow]

I read the article there on the Kelly Critereon. Horrible. Really horrible article. Nonsense math, nonsense reasoning. Just ridiculous. I didn't read any other articles after that. Maybe there's some good stuff there, I don't know, but what I do know is that the arguments made against the Kelly Critereon are not sound.

If you have a taste for nonsense you might also want to read Miller's Crash Course in Vigorish. Right conclusions but laughable reasoning. Wow.

That said, star systems and the like are often abused in an attempt to recover a bad day/week/month. If you start betting more on some games for emotional reasons like this, it certainly is poor money management.

In effect, a "star-based" system is just a dumbed-down version of the Kelly, utilizing the concept of discrete expectation tranches as opposed to mathematically more pleasing continuous expectation values. As such, given Miller's virulently irrational opposition to Kelly, there is some internal consistency to his logic. At least that's something.

Still, the decision not to use Kelly or "starring" isn't necessarily irrational. "Garbage in, garbage out" as the saying goes. Indeed, if one's estimates of EV are are sufficiently flawed by virtue of their bias or inconsistency, then the damage dealt to one's bankroll can be considerably greater than if a flat betting scheme had been utilized. (And of course even in the case of an unbiased and consistent estimator, a high variance estimate can still raise the variance of one's recorded results to unacceptably high levels.)

This being said, I do tend to think that laziness is often (although not always) the reason why many handicappers shy away from Kelly and its progeny. While it may be relatively easy to determine if a given bet has any edge, it takes considerably more work to determine what exactly that edge might be. That's why I think that whenever possible it's good practice not only to predict winning wagers, but also to assign one's own money lines and (if applicable) spreads or totals, and then continually refine the estimation process over the course of time (real or properly back tested). This way, when examining multiple candidate wagers, one can readily determine relative wager levels. Of course, estimating fair lines can be as much art as science, but at least doing so provides a framework from which bet size analysis can easily be performed.

[maritime]

I agree with Razz. This thread made me think of the Bowling Green vs Miami Ohio football game this past season. BG was a double digit dog and their QB Omar Jacobs wasn't supposed to play. When I found out that he was going to play, I felt I had an advantage and increased my play on BG moneyline. They won handily as Jacobs had huge game. In my opinion, the more advantage you believe you have, the more your bet should be. The keys for me are staying disciplined and staying honest with myself (not letting a bad day influence future plays).

[Dark Horse]

What J.R. Miller means, and I know this from having asked him, is that you have a maximum bet size (percentage of your bankroll). No matter how strong a bet, you never bet more than that amount. It is ok to bet under that maximum amount. Miller says that doing so makes only the big bets 'count', but I disagree. I would be crazy to bet the same amount on basketball games as on football.

Miller does have the math to back up his claims. And he increases the bet size when he reaches the next plateau. He suggests 2% of the bankroll per bet, but his brother preferred 1%. Then again, Miller also says that he considers 2% too little, and prefers to bet 1 level higher (3% if I remember correctly). So this is flexible. Some recommend 5% of the bankroll. Seems a lot. What Miller is very much against are progressive gambling schemes, where you try to make up for losses by increasing the bet size.

Emotionally, losing a huge bet is far more damaging than winning a large bet is helpful. Gamblers call this having your nose open, a mental state where a gambler makes wild, undisciplined bets. If you know you have an edge, just keep working it and be patient. If you act like you're not that interested, that beautiful lady (LUCK) will come to you. If you chase her as if your life depends on it, hey, she's seen it a million times.

My own rule of thumb is to bet whatever you're comfortable with. That can be 10% for a $1000 bankroll or 1% for a 100K bankroll. Not everyone can disconnect gambling money from its value in the real world.

[imgv94]

Anyone who plays the same amount on every game is a fool to me..

Your gonna try to tell me you like every game the same? Foolish..

I do not mean to bash,but thats crazy to me that someone would bet
the same amount on every game. It shows me there are not that sharp
and do not take advantage of an edge. Look at me and Razz our +2 unit
games are gold. our 1 unit games are avg. All the Pros I know bet more
or less depending on how much they like it.

If you love a game pound it. but not too much. I had 9 units on the
Spurs today,key word today. I do not make +5 unit wagers everyday
but when I do there gold.

[dave11486]

You guys know what you are talking about. People should be confident about every bet they make, but when a great opportunity comes along you have to seize it. I think it all comes back to being patient and waiting until those great opportunities come along. When you see something that is an exception, lay more cheddar.

[imgv94]

You guys know what you are talking about. People should be confident about every bet they make, but when a great opportunity comes along you have to seize it. I think it all comes back to being patient and waiting until those great opportunities come along. When you see something that is an exception, lay more cheddar.

Exactly! Well written.

Dave I am very impressed with you young fella.
Keep it up!!

[Dark Horse]

I'm going to research this field in more detail. This is just as important as good capping. The information that is out there is oversimplified, not focused enough. It's either flat betting, or Kelly, or progressive schemes, all in relation to winning percentage. Usually the information doesn't include the ideal number of games played for greatest profit, which lowers the winning percentage. There has to be a formula that works best for everyone, and that can be personalized to fit each style.

A chain is only as strong as its weakest link. To be a great capper, but an average money manager, translates into being an average sports bettor.

[imgv94]

I fvckin Loved the Spurs tonight! Was I only suppose to bet 1 unit on
it? LOL!! Thats lame betting the same amount on each bet..Sorry if you
do but I think it's lame.

[Dark Horse]

No, I don't think that at all. What I'm saying is that the info that is out there isn't that great. We can do better. If you want to do it intuitively, go for it. Go with what works. I prefer to have some hard numbers if I'm going to bet 5% of my bankroll.

Capping is one half. Management the other half. But most of us, including myself, spend about 99% on capping. That is where the books benefit...

I'm glad you won with the Spurs, but it was just one bet. I don't want to have attachment to a bet, so I play lower limits. I want to get to where I couldn't care less if I won or lost, because I know that my winning percentage, combined with a maximum profit money management system, takes care of business. Makes sense?

[dave11486]

I hear ya DH, capping is a marathon not a sprint.

[imgv94]

DH I wasn't meaning you bro. I was talking to anyone who bets the same
amount on every game. I agree though capping is 60% and money mgmt
is 40%. I had a problem one time and it costed me plenty, I now know
the benefits to money mgmt and will forever.

[135steward]

I agree though capping is 60% and money mgmt
is 40%. I had a problem one time and it costed me plenty, I now know
the benefits to money mgmt and will forever.

Amen, brother! I lost $10k in the fx/commodities futures markets earlier this year by not covering margins. Live and learn, brothers. Live and learn.

[Sawyer]

I'm a strong believer in flat betting. No one game should be more important then another. You should not be wild about a specific game. You gotta keep is simple stupid. But only few people can do it. You should bet like a robot, no emotions. Many people fell in love with a play and boost units, bets x5 of the regular stake..and you can guess the result, lol. There isnt anything such as Game of Week, Game of Year or Game of your Career.. If you use different units, then you increase the variance.

[Sawyer]

The closest thing that I can compare the star system to is counting cards in blackjack. Is it poor money management for the card counters to bet 10 unit on a hand when the true count is highly in their favor, and only bet 1 unit on a hand when the true count is against them? I don't think so.

Sports betting is not black jack. There's no card counting in sports betting.

[wal66]

I'm no pro but if I am putting my money at risk I am doing so with some sort of confidence. To that end I treat all plays as equal. Obviously a recreational wager on a higher risk play like a parlay is for less but individual plays are all treated as having the same potential for winning.

I am also a new fan of risking a unit instead of betting to win a unit.

[Inspirited]

I am probably moving away from flat betting to fixed-profit betting.

[podonne]

I'd go with flat betting 100% of the time. It is very hard (I would argue impossible) to be able to quantify your edge in any given sports-betting situation. If your line and the book's line disagree, either they are right or you are right, and since you dont know which is definetly the case, or by how much, you have to just decide whether to bet or not. "Calculating an edge" is pointless.

Too many people assume that when thier line is different from the book's line, then it has somehow resulted in a set of precise statistics that you can perform complicated math on. It hasn't, you just have a bunch of estimations by a bunch of different people that have come up with a number that looks very precise (0.4525), but isn't.

BlackJack and other table\card games are completely different. Everything is known about a blackjack game, there are precise probabilities associated with all outcomes and counting cards could very well give you a precise, mathmatically defined edge that you could use to do proportional (Kelly) betting. That will never be the case in sports betting.

I wish more people understood the difference between sports betting and table\card games. Much of what you know about probability and betting doesn't apply to sports.

[ScreaminPain]

Wow! …this thread is still around since ’06 and with some interesting comments in here.

Let me make a case for varying bet size, as it applies to my July baseball work.
I’ve determined that I can win 63% of my one star (*) bets and I can expect to win 77% of my three star bets (***). I have further proven that the average “*” bet is made laying odds of -131, while my average “***” bet is made at much higher odds of -163. I had 59 “*” bets in July finishing 37-22 and 39 “***” bets finishing 30-9. So, for this dissertation I will use $100 as a basis. Of course to determine any of this, one must keep some sort of record, spreadsheet or journal.

Now considering my “*” bets. Assume I’m playing “to win”, I won 37 and lost 22 at an average lay of -131. The breakdown is such. I won 37 bets at $100 for $3700 and lost 22 at -131 for -2882. Therefore, I realized a net profit of $818 on one star bets. Now for my “***” bets, I won 30 at $200 ea for $6000 and lost 9 at -$326 ea for -$2934 and a profit of $3066. This netted me $3884 total. ($818+$3066).
Now had I flat bet, my “***” bets would have been
30 wins at $100 $3000
9 loses at $163 $1533
Net profit of $1467 on the “***” and the same $818 on the “*” bets for a net profit of $2285 for the month.

So you see, I would have “cheated” myself out of $1599 profit by flat betting all my plays—one star and three star. I used to flat bet everything, only to be awakened when I was finally able to develop a model and spreadsheet to accurately track my plays. To paraphrase, as it was once stated in another thread, “if you don’t have a positive expectation of victory, you’re optimum bet is zero”. With that in mind, my best advice is to create some sort of method for tracking, thereby allowing for one to take advantage when an extremely advantages position is presented. It will maximize your profits.

[chunk]

I'd go with flat betting 100% of the time. It is very hard (I would argue impossible) to be able to quantify your edge in any given sports-betting situation. If your line and the book's line disagree, either they are right or you are right, and since you dont know which is definetly the case, or by how much, you have to just decide whether to bet or not. "Calculating an edge" is pointless.

Too many people assume that when thier line is different from the book's line, then it has somehow resulted in a set of precise statistics that you can perform complicated math on. It hasn't, you just have a bunch of estimations by a bunch of different people that have come up with a number that looks very precise (0.4525), but isn't.

BlackJack and other table\card games are completely different. Everything is known about a blackjack game, there are precise probabilities associated with all outcomes and counting cards could very well give you a precise, mathmatically defined edge that you could use to do proportional (Kelly) betting. That will never be the case in sports betting.

I wish more people understood the difference between sports betting and table\card games. Much of what you know about probability and betting doesn't apply to sports.

I agree with everything that you have stated with the possible exception of flat betting 100% of the time. Right now I tend to lean to this star system with small incremental differences. Always on the lookout for improved staking methods, but difficult for the reasons that you laid out in your post.

[dogman]

Everyone has their own way of betting. Personally I have mostly used the RETURN method spoken in another spread for most of my plays the last few years but if you can spot bigger edges you should bet more. I am talking about betting teasers and some correlated stuff than can have bigger +EV's(not as much as in years past with most sportsbooks making it harder by cutting you off, limiting you or changing the payoffs).

Your betting BANKROLL should also make a difference in the size of your bets.

[podonne]

After much consideration, I will revise my statement slightly. I would use flat betting on 100% of bets created by a single individual model. Model in this sense means system, person, process, comuter program, any thing that picks a series of bets and is evaluated by saying that following that model\person\system and making every bet results in +11.5 units.

When you say that a series of 95 bets results from a model results in +11.5 units, that is all you are saying. Every bet in the series and every future bet made by the model has the value of + (11.5/95)= +0.121 units. The fact that you get twice as good odds on bet #34 in that series as bet #56 does not mean that you should bet more on #56 than #34, because they both have a positive expected value of +0.121 units.

The reason is that when you qualify a model further, say by stating that you will bet more if the team is at +200 as another team at +100, you are adding a condition to the model, and thereby creating a completely different model that has to be backtested and subjected to minimum sample size requirements and statistical tests.

An illustration (the numbers are fantasy, only the reasoning is important):

Model A: Bet all home dogs during 2007 results in +36.7 units on 567 bets.

I constantly hear people saying "that's good, so that implies that a fair odds line on any home dog in 2008 is +125, so you could get better results if you only bet home dogs at +125 or higher". The I might hear "oh, and apply kelly betting so that if its higher than +125 you bet more because you have a more positive EV."

And I say wrong, wrong, WRONG! What you've done is to create a different model, and it has no relation to Model A. Lets call it model B:

Model B: Bet all home dogs at +125 or higher.

Model B is not Model A. The results of model A are in no way related to model B. Model B is not guaranteed to have a better result because you took a broad trend in Model A and applied it only in higher EV situations to create Model B. In my experience, Model B will probably have a worse result, but it may have a better or worse result, the point is that you don't know unless you backtest Model B and get stastically valid results.

So back to my original statement, if you have two models\systems\persons and one is +10 units, and a second is +20 units, you can absolutley bet more on the second than the first, and kelly betting is perfectly acceptable here. But notice that we are applying it to the choice between bets created by two different models\systems\persons, not to the indivudal odds/bets within that model\system\person. Within a single model\system\person, only flat betting is appropriate, as per the above.

Whew.

[vyomguy]

kelly criterion is the best.

[Don_C]

I'm familiar with JR Miller, and I would have to agree. The truth is, there are no "locks" and no guarantees in the world of sports betting. You either find an overlay in a game, or you don't. A true sports betting investor spends countless hours analyzing almost every line, every day. He generally only places only "one unit", 1-2 % of his bankroll, on every matchup he has found an overlay in. If someone thinks they're finding 10* value on one line, 5* value on another, and only 1* on others and they are actually placing money on those 1* games; than that person is a gambler - not a true sports betting investor.

[wiffle]

I'm familiar with JR Miller, and I would have to agree. The truth is, there are no "locks" and no guarantees in the world of sports betting. You either find an overlay in a game, or you don't. A true sports betting investor spends countless hours analyzing almost every line, every day. He generally only places only "one unit", 1-2 % of his bankroll, on every matchup he has found an overlay in. If someone thinks they're finding 10* value on one line, 5* value on another, and only 1* on others and they are actually placing money on those 1* games; than that person is a gambler - not a true sports betting investor.

lol