[mikeanite]

i would like to ask ppl with experience. how would you go about maximize your profit if on average i go 5-3 for the day? i'm super greedy and always fall flat on my face.

[Stinger]

f l a t b e t

[mikeanite]

f l a t b e t

sometimes the juice is too much. i don't have a bankroll of 25,000 or anything.

[kisado]

You basically answered your own question. Stop being super greedy and you won't fall flat on your face.

[tenbas]

You will never maximize your profits if you don't work on that greed thing of yours. I know I learnt, cost me plenty.

[Pokerjoe]

If you're going to go 5-3 on the day, it is mathematically undeniable that you should bet all you have and can borrow.

If you're merely going to average 5-3 on the day, bet a little less.

[Pokerjoe]

Stop being super greedy and you won't fall flat on your face.

That's not true, in that he can stop being super greedy and will still fall flat on his face. This is because anyone who thinks they're going to hit 62.5% has already failed. "Greed" isn't the right word for what ails him.

[JustinBieber]

If he wants to maximise profit then the correct answer is Kelly.

[wrongturn]

simple, bet 1/8 of all you have on each bet, and after you see 3 losses, go all-in on each bet rest of day since they can't lose.

[Grind-It-Out]

If he wants to maximise profit then the correct answer is Kelly.

60% win average. The line really matters, but if it's on a -110 line, you should be betting 16% of your roll on each bet. Of course, that's if the bets are one at a time. You have to adjust the bet size to be a bit lower if you bet multiple games at the same time.

http://www.sportsbookreview.com/betting-tool...alculator.aspx

[Arilou]

This situation calls for meta-uncertainty. In those scenarios where you actually WILL hit 60%+ over the long term betting eight games a day you're going to be the biggest force in sports betting quickly, so there's no reason to bet as if that was your edge and you're going for maximum bankroll growth. Also keep in mind that a sufficiently large bankroll doesn't scale like you think it does because getting down gets harder - in a major American sport betting $1000 a game gives you tons of places to go that are giving you a great price, $10,000 a handful that will have sharper lines and $100,000 requires going to multiple places or wagering multiple times even if you hit the top guys. Thus, the calculator's assumption that it's as good to double as it is bad to lose half is incorrect even if you're all about the main lines. If you're doing props or other low-limit fare that goes double.

If all you care about is maximizing profits and you know beyond a doubt you're going to average 5-3 then yes you bet about 16% of your bankroll per game, assuming all bets are of equal strength. Of course, you're not actually going to go 5-3 if these are -105 style bets rather than favorite MLs (even if you're going -110 you can shop around for numbers, no one trying is truly paying -110) because no one is that good.

[Wrecktangle]

What's your long term win%? Find that number and if you want to be hyper aggressive, bet it at 1/4 Kelly.

But if you want to survive to the end of the season, flatbet it using 1/100 of your bankroll.

On second thought, if you are asking these kinds of questions: only flatbet.

[Grind-It-Out]

What's your long term win%? Find that number and if you want to be hyper aggressive, bet it at 1/4 Kelly.

But if you want to survive to the end of the season, flatbet it using 1/100 of your bankroll.

On second thought, if you are asking these kinds of questions: only flatbet.

Can I ask why you think 1/4 Kelly is the most aggressive one should go?

Is it because of the points in the post above you? to account for errors in estimated win %? or something I'm missing?

[JustinBieber]

Kelly is more optimal for growth than 1/4 kelly.

[bachngocduong]

i would like to ask ppl with experience. How would you go about maximize your profit if on average i go 5-3 for the day? I'm super greedy and always fall flat on my face.

nothing

[jgilmartin]

The answer is full Kelly, but I'd want to be much more certain about my edge than "on average i go 5-3 for the day"

[carparts1000]

arilou, that is a great , quality post.

[bandit bettor]

flat betting is key i know from expierience i had many 4-1 nights that i came away in the negative 4-0 on early games so u dump the profit all on a late game for that super big win that misses and end up losing money... try flat betting for a week or two and you will notice a nice change in your bankroll even if u love certain games so much more try it just for a few weeks..

[Grind-It-Out]

flat betting is key i know from expierience i had many 4-1 nights that i came away in the negative 4-0 on early games so u dump the profit all on a late game for that super big win that misses and end up losing money... try flat betting for a week or two and you will notice a nice change in your bankroll even if u love certain games so much more try it just for a few weeks..

Even if you can't quantify your edge precisely, if you have sufficient evidence that the games you "love" have a higher EG than the games you "like" then you would be silly to flat bet.

[Yi]

If your games are nicely spaced out, I would try a conservative progressive betting system like (Labby or Oscar). From my experience, if you can get "+ev (very important)" bets then using progressive betting system will maximize your profit.

[Wrecktangle]

Can I ask why you think 1/4 Kelly is the most aggressive one should go?

Is it because of the points in the post above you? to account for errors in estimated win %? or something I'm missing?

A few of my friends have seen me go full-Kelly in my wagers at times. I've always done it with very small bankrolls. If you can hit high win%s (I'm looking to get above 57%) the exponential growth really tends to take off above 60%. Unfortunately, nothing is constant in sports, even a league that you have a bead on, can change in a few days. When I'm doing this, I'm running Bayesian forecasters of various types in order to shut down these outrageous BR fractions, even to the point of stopping betting entirely, in defense my BR against league changes.

The point is, unless you really do the math, in an environment of change (this is NOT Blackjack with calculable odds) you haven't a good constant win% to rely on, thus I see 1/4 Kelly as the largest you should consider UNLESS you are a decent mathematician and have spend a lot of time working Kelly betting both in simulations and with real data not in your forecast db. Even at 1/4 Kelly you should be damn good in math, if not, leave it alone as you will only get burned.

The average punter should always be very defensive and considering betting 1/100 of your BR per game. This system, called the century system, keeps you in the enjoyment of betting your favorite sport and allows you to get through the entire season without burning thru your BR.

[Wrecktangle]

If your games are nicely spaced out, I would try a conservative progressive betting system like (Labby or Oscar). From my experience, if you can get "+ev (very important)" bets then using progressive betting system will maximize your profit.

conservative and "progressive betting systems" are an oxymoron

Progressive systems are a fun way to crash and burn in an extremely short time. Much quicker, BTW, than Full-Kelly, a system many folks akin to hari-kari.

If SBR had that self-emolation smiley in the side panel you see at some of the other forums, I'd have put that here.

[That Foreign Guy]

I am now changing my tune on this:

Had I been flat betting I would be +20u on moneylines over the last 200 bets. Because I was Kelly betting (full Kelly for the fraction of my bankroll devoted to this) I am slightly below breakeven - my top twenty bets performed significantly worse than random selections despite beating closing lines.

If you don't have the mental fortitude and confidence in your model / system to deal with these swings and the ability to sustain them financially (I don't need to make draw downs) then flat betting is the way forward. Of course it means you miss out on some of the upside when things go well. But winnings forsaken are a lot less bitter than losses taken.

[Wrecktangle]

I am now changing my tune on this:

Had I been flat betting I would be +20u on moneylines over the last 200 bets. Because I was Kelly betting (full Kelly for the fraction of my bankroll devoted to this) I am slightly below breakeven - my top twenty bets performed significantly worse than random selections despite beating closing lines.

If you don't have the mental fortitude and confidence in your model / system to deal with these swings and the ability to sustain them financially (I don't need to make draw downs) then flat betting is the way forward. Of course it means you miss out on some of the upside when things go well. But winnings forsaken are a lot less bitter than losses taken.

TFG: in all likelihood you have a misaligned model.

Many times I've found that I simply didn't have all the factors in a model/simulation that should have been there. It doesn't mean you should toss your model, even though it probably does need to be improved. You can counter the problem that by having lower and UPPER bounds on your selection criteria "window." i.e. due to misalignment, your "better" picks are not better, instead you have a "sweet spot" and can't really grade your selections as to quality.

But back to the point, you can Kelly bet the sweet spot area, but they should all go off at the same Kelly fraction.

[smitch124]

What about 1/2 to full Kelly if you have a more quantifiable edge like an off line in an efficient market?

[Wrecktangle]

What about 1/2 to full Kelly if you have a more quantifiable edge like an off line in an efficient market?

It's your money, bet it like you want. If you have a long term "backcast" look at your perceived edge, it may be real. My point earlier was: each sport changes within the season: pre-season, first several games of season, reg season to trade deadline, trade "transition", period between trade transition and post-season ramp down games, last few games before post-season (ramp down), playoffs, and inter-league (in MLB) can all be quite different than a calced long term "norm". Depending on the sport all of these periods can be stronger or weaker, i.e. seen clearly or not against the league norm.

Then for each season, you have new coaching trends as well as new league rules (a particular hazard in the NFL) or league chaos (WNBA & Arena Football). Nothing is really the same all the time and when you bet a large fraction of your bankroll against a changing win%, you can quickly burn up a BR with almost any fraction of Kelly, especially full Kelly. And then if you are betting simultaneous games with Kelly (depending upon how you divide up your BR against selected games) it can melt away even faster.

[That Foreign Guy]

Wreck - thanks for making me re-evaluate things.

These are moneyline bets between 1.20 (-500?) and 18 (+1700?). I had sliced them a few ways (by bet size, by odds, vs closing line, vs actual result) without noticing anything until finally realising I had the "edge" in my spreadsheet even if it wasn't explicitly there (kelly=edge/odds, so edge=odds*kelly - thank you high school maths).

I graphed my believed edge against cumulative bankroll effect (so starting at 0 with the lowest edge bet and adding wins and subtracting losses). It's amazing what this reveals (I'm going to round numbers because that's how people are wired):

Edges under 2% seem to be fairly neutral EV (the cumulative profit hovers around zero).

Edges above 2% seem to have +EV (there is a consistent trend upwards in cumulative profit)

Edges above 10% seem to be -EV traps. (my cumulative profit basically falls off a cliff). Total loss for these bets is about 50% of initial bankroll. I had 1 win of the 3.4 I'd expect if I was getting 0EV bets.

Now, given a choice between "I am really unlucky" and "I am not as smart as I think I am". I know which one I should choose, and I know which one my ego wants me to choose (has been choosing). Further supporting evidence to the theory that my model is misaligned is that that half of these losing bets totalling 2/3rds of the losses and 1.4 of the "missing" wins, are in the same group of leagues.

I think from now on I will focus on the sweet spot between 2% and 10% and re-evaluate in another 200 bets.

[Wrecktangle]

TGF: I think that you have also touched on another area that I think is important. You probably have already done this, but if not consider constructing a db of games that can be binned into every way that you think can be important. Test your model against every bin that has enough data to statistically say something (at least 30 games, but your really want these numbers to be in the hundreds) and then see if your win% holds up. In the test, if your win% falls significantly, then a mod to the model needs to be considered embracing that bin.

And one other in terms of ego; the market is much cleaver than one might think. In NCAA FB I was wondering if Homecoming games really had an effect. I didn't have them in the db so I had to drill them out. A week or so later I found not only the long term value for HC, but that the market had almost exactly compensated for it in point value.

[hoku808pc]

simple, bet 1/8 of all you have on each bet, and after you see 3 losses, go all-in on each bet rest of day since they can't lose.

^^^ this actually made me Laugh Out Loud


PS...Sorry for interrupting your guys' input, but I'm trying to figure out what you mean by "kelly" and how does it work???

[Data]

I think from now on I will focus on the sweet spot between 2% and 10% and re-evaluate in another 200 bets.

I would suggest rethinking this because it sounds very similar to Bill the Cop's approach to teasers selection.

[sideloaded]

I would suggest rethinking this because it sounds very similar to Bill the Cop's approach to teasers selection.

Is all you do is nitpick? Give Bill the cop a break. Data remember that episode of star trek when they insert that chip in you to have human emotions?

[Data]

Is all you do is nitpick?

It is given that we may have a different opinions on the importance of points I bring up.

Give Bill the cop a break.

I have never given him a hard time.

You seem to be actively responding to my posts lately while I never responded to yours. I am puzzled of what could be the cause.

[sideloaded]

Come on Data show some emotions.

[Toples]

with 62,5%, and only 1 bet a day (365 per year with -110 line) if you want to be 99,73% sure you'll make profit, you gotta bet 15% of your bankroll to maximize your profit.

if you had 1000$ bankroll to start with, you can retire cuz you will win 230 million $ on average (median is over 400 000$)
IF you hit 62,5%, which is possible (well, even 100% is possible), but HIGHLY unlikely...

[That Foreign Guy]

I would suggest rethinking this because it sounds very similar to Bill the Cop's approach to teasers selection.

Possibly, although it makes sense - lower edge plays the book and I are both about right, medium edge plays I have an edge, large edge plays there is something I missed.

Bookies don't tend to be out by 10% very often at least not on the markets I play. Now it's possible that I occasionally find a gem, but the closing lines weren't backing that up (line didn't close better for the big edge plays). Setting the exact lower and upper values in stone is probably a mistake, I'll continue to paper trade the plays outside that range and re-evaluate every couple of hundred bets.