How much should you bet per game?

[Beyond the Bets]

Way too many people worry about how much they should bet per game, which is a good thing and also a bad thing. It's a good thing because you care how much you bet per game, but a bad thing because you spend all your time worrying about much to bet rather than on who to bet and when to bet and what to bet.

So, for those of you who wrestle with trying to figure out the optimal bet size relative to your bankroll, let me try to dumb things down for you as best I can:

If your bet size is so much that a loss will ruin your day, it is too high.

If your bet size is so small that a win does not bring you any joy, it is too low.

Find a goddamn middle ground and go win some money!

[oddsok]

Haha, so succinct!

I generally flat bet 2.5% of my roll and work up or down from there, never going above 5%.

If I know I'm getting a really good number or am very confident in my bet I will up my bet size.

If I'm a little unsure (say betting a ML dog) or am getting a bad number I will drop my bet size.

[mount187]

Two good pieces of advice that I've received:

Never bet money that you can't afford to lose.

If you NEED a win, your mind is already focused on something other than making a good pick. How can you focus on making the right play, when you're running through a list of all the bad things that are going to happen if you lose?

Gambling should only affect your life when you win.

You should be able to lose games and not have it change your mood terribly. If you lose a game and you're mean to your friends and family afterward, you need to re-evaluate. Remember, even really good sports bettors are going to lose 9 out of 20 bets. It's going to feel like you're losing a lot, it's because .. you're losing a lot (you're also winning a lot too). If you're making good picks though, long term you'll come out ahead.

[Pujols_5]

(11-23-2011 09:27 PM)mount187 Wrote:  Two good pieces of advice that I've received:

Never bet money that you can't afford to lose.

If you NEED a win, your mind is already focused on something other than making a good pick. How can you focus on making the right play, when you're running through a list of all the bad things that are going to happen if you lose?

Gambling should only affect your life when you win.

You should be able to lose games and not have it change your mood terribly. If you lose a game and you're mean to your friends and family afterward, you need to re-evaluate. Remember, even really good sports bettors are going to lose 9 out of 20 bets. It's going to feel like you're losing a lot, it's because .. you're losing a lot (you're also winning a lot too). If you're making good picks though, long term you'll come out ahead.

I'm an ultra-competitive person so I hang on every game I bet like its the world. But that is just naturally. I find betting more of a hobby so I only bet a few $ at a time just to have a rooting interest and have some fun. However, if there is a matchup that I really feel confident about I will either bet pretty high or use the Kelly Criterion that way if I do lose it doesnt completely wipe me out and if I win it helps more than the usual couple of $ I would have bet in the first place.

This is really the first time I have started betting and I have watched more of sports than I ever have and I used to watched a lot of sports before. It's fun and I'm always looking for new theories or a competitive advantage.

[grantumchi]

I couldn't agree more with the above (Mount). Unless gambling is your sole means of income, it is a hobby. There are certainly different levels of hobby but at the end of the day even if I have a terrible day at the golf course, it doesn't ruin my day, my week or my mood, because it is a hobby.

The amount of money per bet/game is completely to the individual. If I'm willing to risk 10% of my bankroll on what I feel really good about then I should, because it's my decision. I know guys that won't bet over 5% of their roll, on ANYTHING! Its just a very personal decision.

Nice to be here, looks like a great site!

[mike laga]

I determined a long time ago that it is not always the money lost or money won that dictates my happiness...it is being right vs being wrong.

Being completely wrong coupled with a significant financial loss is a bad day. Because of this, I have scaled back the amounts that I wager.

I had a day last week where I went 7/7 in MLB. Taking time to cap each game, making my choices, and then having them come through is the true reward. And it keeps me in tune with what is going around the league, so I enjoy that.

[SkyBlues42]

Find the amount that would piss you off if you lost, and just never wager that amount. Other than that, the bet size will fluctuate depending on confidence and other factors.

[LOLstradamus]

Money management is so critical, especially for novice betters. I remember when I first started betting how many times I'd be cold early in the day and then chase big plays at night that had -EV, which over the long-run left me with nothing.

Set long-term goals for yourself (don't say I want to make $xxxx , stick with winning %'s) , you should be gradually building your bankroll over a long period of time. Start by creating a bankroll with an amount you can afford to lose. Someone new to sports betting I'd highly recommend nothing more than 2%. Your first few years the majority of learning will come from your mistakes.

A lot of people have varying opinions when it comes to money management. I advise that bet size should always be a single-flat % of your bankroll. I also think the unit game can be dangerous in the long-run. Every play should be with maximum confidence and I'd recommend thinking about why one play is 1-unit versus another that is 3-units.

Different strokes for different folks. The only universal rule is you should be able to afford losing your entire bankroll.

[SkyBlues42]

How do you calculate maximum confidence? Is it just a gut feeling, or do you have to use numbers to determine that? Seems like that would be difficult to figure out. If I'm very confident about Texas over LAA one day.. then very confident about NYY over Bal another, how would I determine whether or not my confidence level matches up?

[LOLstradamus]

(07-24-2012 02:02 PM)SkyBlues42 Wrote:  How do you calculate maximum confidence? Is it just a gut feeling, or do you have to use numbers to determine that? Seems like that would be difficult to figure out. If I'm very confident about Texas over LAA one day.. then very confident about NYY over Bal another, how would I determine whether or not my confidence level matches up?

In my opinion, it's purely subjective, varies on a case by case basis, and it's just something you get better at identifying through experience. From my observations when people bet with varying units, they're typically a volume bettor so they are getting action on a multitude of games and will differentiate confidence levels by unit-betting.

[VincentValour]

(07-24-2012 02:02 PM)SkyBlues42 Wrote:  How do you calculate maximum confidence? Is it just a gut feeling, or do you have to use numbers to determine that? Seems like that would be difficult to figure out. If I'm very confident about Texas over LAA one day.. then very confident about NYY over Bal another, how would I determine whether or not my confidence level matches up?

Sky Blues I'm constantly pulling my hair out over this. I have used a mix of linear and exponential bet sizes, loosely based on units, and thresholds of %'s of my bankroll. In general units are ok but sometimes a game has a much bigger edge and warrants a much bigger bet.

The easiest way to tell confidence level is to use differences in your lines vs. book's lines. In spreads for example, a 5 point difference will warrant a bet but, the question is how much to bet when the difference is 12 points, to go along with a situational angle. Surely this bet has to be bigger than 2 to 3 units.

Maybe a good way to research that is to back run your numbers, and find out how many times a given point edge shows up. If a 15 point edge only happens once a year, your bet should reflect that rarity.

Same goes for ML sports, also making your own lines and getting ahead of the market moves is a good way to determine advantage. Linde and Mount's Price machine does this very well if you look at the splits 10% 15% 20% 25%, that is the perceived edge, the biggest edge is 25% and in theory should warrant a bigger bet. (Correct me if I'm wrong guys)

If you are just using various stats and situational angles, you can do a simple count of how many angles/stats are going your way. "Rangers have 5 angles going there way so 5 units." Doing that is short sided however, because price is not factored in which IMO is the basis of sports betting, or any kind of market.

The Kelly is solid betting strategy, however the edge has to be known before betting, and in sports betting the edge is just perceived. I think Jeffery Ma said this is why blackjack is a perfect game, because the edge is always specific, giving a chance to bet accordingly and correctly.

In any case calculating an edge is very tough given the mixed results of any game. If you are confident in your numbers or edge, stick to your guns, eventually skill will weed out the variance and produce profit if your numbers or edge are solid.

I already ranted in another post about this, anyone else have theories or ways they calculate their edge? Would love to hear them.

[LOLstradamus]

The not so helpful answer: If you only bet single-units you completely bypass that frustrating process.

[louiemdj]

I like to bet enough that it hurts when I lose. Definitely not the best approach but it's money, you can always make more, right?

[mount187]

(07-24-2012 07:26 PM)VincentValour Wrote:  Same goes for ML sports, also making your own lines and getting ahead of the market moves is a good way to determine advantage. Linde and Mount's Price machine does this very well if you look at the splits 10% 15% 20% 25%, that is the perceived edge, the biggest edge is 25% and in theory should warrant a bigger bet. (Correct me if I'm wrong guys)

This was the theory behind tracking our plays with that information. If our line is 25% off of what the listed line is, then that's our perceived edge*. Unfortunately our 25% plays are 68-92, -15.19u right now, and our other ranges are doing significantly better. Go figure.

When I set up the tracking, I broke it down by home/away, dog/fave, and the combination of the two, so we could get a better view of where our edge actually is. It would be more useful if I had the time to break it down by m/l ranges, say +115 to +119, +120 to +124, etc. Then I would know our actual win percentages at those ranges according to how much of a perceived edge we have (10%, 15%, etc). It would also be useful to have our numbers broken down to have the line movement with each sub-section. That way we could look and see if, "Yeah, this perceived edge grouping isn't going well, but the lines have moved x% on average in our favor. The plays are probably still good, but the results aren't working in our favor." It's all part of just getting a better understanding of how your handicapping is translating into actual results.

Along with that, I don't think you can just plug in the percent edge into the Kelly formula and go, because that doesn't factor in your actual win rate on those games. I'm still working on how it all fits together, so if someone can provide some more info, that'd be great.

*It's important to note that that's our perceived edge of our value against the listed line, not our perceived edge in who's going to win the game. Having a 25% edge on a play doesn't mean we think that team has an amazing chance to win the game; it's an edge on what we think the line should be versus what it actually is.

[Turnstone Money]

Yeah I struggle with this idea...

[mount187]

More rambling from me ...

You have to have a lot of confidence in your bet tracking, and edge to use the Kelly model. Here's why:

If you look at our true line for CIN @ HOU for 7/23, we had CIN as a -124.84. The line came out CIN -150. That means we thought there was a lot of value on HOU that night. If you plug those numbers into the Kelly formula, you end up with something like this:

On a $10k bankroll, with odds of +160 offered, and an estimated probability for HOU of 44.64% (+124.84), you should wager 7.73% of your bankroll ($773).

If you were to use a conservative number, like we do (we pad our #'s both direction with 10% before we'll play them), you would still end up with offered odds of +160 and an expected win rate of 42.14% (+137.33). At that rate you'd still be betting 3.57% of your bankroll.

If I were to go back and look at our tracking, I'd see that on 20% variance plays (which that one was), I have a few numbers I can use to go from there. On all 20% plays, we're 49-39 (57%). On home 20% plays, we're 20-20 (50%), on dogs we're 55%, but on home dogs we're only 7-8 (47%). Those aren't huge sample sizes, but I at least get a sense that we're a little better than a coin flip in these situations. So betting a 42.14% dog in situations where we're generally a coin flip feels pretty comfortable for me, and that 3.57% bet seems justified. The 7.73% certainly is not. If the numbers were worse (say HOU is a home dog in the 10% range, where we're 9-19), I'm not going to feel great about that 3.57% number either.

Anyways, that's a little more in-depth look at how I think figuring your edge is supposed to work. I know some people hate the Kelly model, because getting your actual edge is really difficult. I've also read that people use half the Kelly listed bet %, because they feel it's too aggressive; your bankroll doesn't go to infinity, so even if theoretically you should come out ahead betting that 7.73%, if your bankroll runs out before then, that theory isn't worth much.

[Edit]

For the record, we have way too small of a sample size with our numbers to use Kelly against our plays. I don't, and was just using these numbers as an example.

[VincentValour]

(07-25-2012 10:10 AM)mount187 Wrote:  *It's important to note that that's our perceived edge of our value against the listed line, not our perceived edge in who's going to win the game. Having a 25% edge on a play doesn't mean we think that team has an amazing chance to win the game; it's an edge on what we think the line should be versus what it actually is.

In general that is how you get "your edge." A bet on the team just to win the game is rather arbitrary, because the real game is playing the number/price in whichever sport. Hence a losing record but + units due to dog winners.

Probably the best way to truly determine if your edge is real, and how much is to back run your model, or figure out your Z score. (How many standard deviations are your picks from the average expected # of correct picks) / (How likely it is to have a specific record.) But that would be really difficult with a ML sport. Think you would have to calculate you average chance of winning all your games, then run standard deviation and Z score.

But yeah I think your ramblings nailed it.