Question for Dr. H on money management
maidenguy
Senior Member
I noticed at your site all of your bets are to win 1 unit so you are risking less money on dogs than on favorites. For ex., based on $100 bets, on a -120 favorite, one would bet $120 to win $100 but on the dog ones bet would be app. $83 to win $100.
That is different than most handicappers out there who bet to win 1 unit on a favorite and risk 1 unit on a dog. Just wondering if you would mind sharing your thoughts on why you bet that way.
That is different than most handicappers out there who bet to win 1 unit on a favorite and risk 1 unit on a dog. Just wondering if you would mind sharing your thoughts on why you bet that way.
Comments
http://www.sportsbookreview.com/forum/handicapper-think-tank/29841-maximizing-expected-growth-kelly-criterion-part-ii.html
That's probably over most people's heads, and that's fine. The key parts are here:
So we see that you should be betting 1.xx to win 1 on favorites, and .xx to win 1 on dogs, and that if you are risking 1 unit on dogs you are overbetting your dogs.
Basically, the bigger dog you play, the higher the risk of ruin (busto). So the counter that, we play dogs for less.
Intuitively it makes sense, too. Say we play a +180 dog and it closes +140. Great bet! But it's still probably going to lose. I don't want to be risking as much on that as I risk on a play that is probably going to win.
I have the Blue Jays at +2500 to win the World Series. They are currently +465 at CRIS. Good bet that's really likely to lose. Again, I don't want as much invested in that as I invested in my Washington -155 play last night, or even my SF +108 play tonight. I want to win as much, not to risk as much.
You released that game after lineups where out correct?
No, overnight. I didn't know Kemp was going to sit but I didn't know Span was out either.
I'M ON IT
I can get that, sure, not home at the moment though. I know that if wagering 1 to win 1.xx on dogs, my units are at 28.11 instead of 30.73, so not much difference, but slightly better the way I'm grading (which isn't my reason for doing it obviously).
How's your season going?
Pretty good, super hot start until May 15 or so and pretty much break even since.
Well if you lower your stake on favorites and keep your stake on dogs where it's supposed to be, you're doing 2 things - 1) lowering your overall unit size, which will serve to preserve bankroll longer but not promote as much growth as it should, and 2) you are now wagering more on dogs, relative to faves, than you should be.
I'd also add that looking only at your results to verify whether or not this is the correct way to stake ignores the calculus behind it. Your results may vary because it's not a large enough sample size, although probably not the case with you, OR it could show that your methods show a larger edge on dogs, which may well be the case with you. And if you have a bigger edge on dogs than on favorites, then you should be betting more on them relative to faves.
you cant really use a kelly system when you have no idea what the CL will actually be, especially on lines you actually bet.
and betting less on dogs or faves seems to me to be a waste of time, your return handles the variance of fave or dog.
and I have been betting risk max at BOL openers on my MLB games and so it means you always win 250 no matter what the line is. for me this is best because it gets as much of my money in the BOL soft openers as I can. but less on dogs or more on faves I dont see how mathematically that changes anything.
you bet 100 on a .168 fave 10 times you win 6.25 times and win 1000 bucks
you bet on the dog +155 10 times you win 3.92 times and win 1000 bucks
where is the happen to lose more. the only possible value at all in anyway is avoiding the expected longer droughts caused by betting lots of dogs. the probability of having an extended drought with lower implied probability is definitely higher.
example yankees -360 bankroll $50000, i need to bet 5500 bucks which is over 10% of my roll on 1 game.
I am not sure I am comfortable with that strategy!
Sides Sides Totals Totals Totals
Line Value 1.65% 2.12% 1.69% 0.07% 1.17%
guess>buy 2-2.5 3&+ 3-3.5 4-5.5 over 5.5
and you can see by this I am getting the most value at certain points , this gives me in a sort of weird sense a kelly type system. I want most of my money bet at where i get the most CLV . so I would bet big on sides that my guess and the bet line gap is over 2.5 and on totals I would make the most by betting more at 3 to 3.5 of gap. so what i do on mlb is I am tracking where i bet compared to my guess and see what range gives me the highest clv point.
and of course all these numbers could all be noise and it means nothing!
Conceive of two separate bets, one at +110 odds and one at +120 odds. Assume you have a 2% edge on each bet. Plug the numbers into kelly criterion and see what proportion of bankroll it recommends for each situation. You'll find that despite both bets having equal edge, the kelly criterion will recommend you bet less on the longer odds wager. This is an obvious contradiction to the convention of always betting 1 unit on dogs. You can extend this to even longer odds, holding edge constant, and watch as the recommended stake decreases further.
Obviously limits play a role here. If you have a large enough bankroll, or limits are prohibitively small, you'll just bet as much as you can.
also how do you know you have a 2% edge?
There's not really a need for more opinions,
when you understand what Granchow has explained here
Part 1: Expected Value vs Expected Growth
Part 2: Maximizing Expected Growth
then you can understand the faulty reasoning behind SportsInsights staking methods. Same applies to the "risk 1 on dogs and 1x to win 1 on favorites" method.
In the case of the 'bet the same amount regardless of the odds', you are increasing your 'risk of ruin' by underbetting on favorites and overbetting on dogs.
In the case of "risk 1 on dogs and 1x to win 1 on favorites", you are betting the correct amount on favorites, but you're overbetting your underdogs.
I bet 1 on dogs and 1x on faves because I am trying to get as much cash as possible on BOL soft openers.
I am saying when I place a bet other than the noisy past for bets similar there really is no way for me to know what the expected value of any sports wager will be since the odds will move after I bet (hopefully in my direction)
can you please put the math as if you were talking to a 2 year old for me so I can see why this is better. the absolute only advantage i see is protecting from bankrupty on your bankroll just a little bit longer. you are less likely to have a bad streak at a higher implied probability so if you have less cash in on the lower implied probability you are lowering that risk, that is the only advantage i see. Please I am super interested if you could explain this to me like I am 2, I would really appreciate it
Nope, won't.
My previous post is pretty clear.
yeah I understand because there is no math to it that makes sense! you learned it that way and you do it that way but I dont see any math to suppor that, its just a method.
if its impossible to have true expected value before you bet, kelly is useless!
No you don't.
The articles by Granchow not exhaustive enough ? Everything you need to know as to why and how to stake your bankroll is in there. Those articles are the end-point of the staking discussions. There is no better explanation out there a.f.a.i.c. (anybody who has a better article or book, feel free to post it here.)
The staking method I'm suggesting is the method I learned from Granchow's article, which is as clear an explanation of staking using the Kelly Criterion as there can be. If you don't understand the math and theories laid out in the article, then I can't help you.
All every sharp sports bettor is doing daily is comparing their probabilities of events happening (or not happening) vs the ones given out by the bookies and which are shaped up/down by the market. Probability x Price gives you the Expected Value. I hope you can understand that. That's all you need to know.
I said you don't understand it.
Granchow's article already did.
The bookies/market already gives you the value (or price) for every bet. I have no clue why you're stating that 'EV is unknown before betting'. Your statement doesn't make any sense.
Rockies today I have them -129 and bought them at -117 (they currently are sitting at -122)
I bet the BOL MAX on them which was 292.50
My question is if I was using Kelly how much should I have bet on the Rockies? What is my expected value on the Rockies game today?
I thought expected value was not the number bet at I thought it was the expected value the game closed at
how I determine expected value is I need to know the probability of the outcome, in sports until the line closes there is no way to know ( i can guess) the probabiltity of the outcome and that is still assuming that the market is effecient which we all know is an assumption and even if it is effecient in Game A it could just easily not be truly effecient in Game B. Please show me how to calculate the EV of an event we have no idea what the probability of the results occuring is! I say Kelly is flawed for sports betting because there is no way pre event to A know the expected value B even if we use the close as EV we are not truly sure the closed line is effecient!
For the sake of this example, let's assume there's a bettor out there who only bets at close (=most efficient line) and he only bets -110/-110 bets. NBA handicap lines. Every bet is a -110 bet. After many many many bets.....what is his Expected Value per bet? What value will every bet return him on average?
100/110 = 0.91
So as I said, he's not beating the market, he's betting at close and he only bets -110 lines. As a result his Expected Value per dollar he bets is 0.91 cents. Concluding: He's only losing the vigorish/juice/rake every time and over the long haul his bankroll will diminish. To break even he has to hit 52.38% of his bets, but as he's betting the closer, so he's only hitting 50%.
Now say another bettor has a model that is better and smarter than the other ones out there. It may also be that he has insider information on injuries that other bettors don't have. So let's assume this second bettor get's the best lines for his -110 NBA bets. Where the 1st bettor would bet the close at +6, he (the 2nd and smarter bettor) would have the opener at +8. So this smart bettor manages to consistently get better numbers. Not on every bet, but more-often than not, he'd be sitting on a better line. As a result he's not winning 50% of his bets like Bettor1 does, but he's winning 57% of his bets. In this case Bettor2 achieves 'Positive Expected Value'. For every dollar he bets, he gets more in return.
Now take these numbers and start at the 'Expected Value vs Expected Growth' article,.....then continue to the 'Maximizing Expected Growth' article as linked to in post #21.