FLAT BETTING : When to recalculate your bet size as your bankroll changes ?

[iparout]

I have to admit that I did not pay too much attention to the odds. As an American I don't have any interest in soccer and completely unfamiliar with the betting so I pretty much passed that over. If he is betting odds that will allow you to win with a sub .500 hit rate, he may very well have a good model. If he has lost over 2,600 bets though, I think he is sunk and will sink further over time. That ain't gonna work even with some positive CLV. I've got positive CLV.

I have not lost after 2600 bets. I am winning. My ROI was ~4%, went back to 3% because of my recent 3-month HUGE downswing, and I am now trying to build it back. I indeed lost 50% of my bankroll during my downswing, but the bankroll was already big enough due to my previous winning months and, thankfully, thanks to being strict with my money management policy, I didn't go busto...

Problem was that, due to readjusting my stakes every 1000 bets, last Saturday (which was the lowest part of my 3-month downswing and still far away from readjusting my stakes) I was actually betting ~4% of my roll on each bet and given the fact that I had 35 simultaneous games I wanted to put money on, my bankroll couldn't handle that... I lowered my stakes by 50% in order to put money on all 35 selection and, guess what ??? The day was a huge winner but I only managed to earn half of what I would have earned had I not reduced my stakes by 50% !!

As I said, Murphy's law...

[BTL]

OK then. Apparently he did not lose most of his bankroll as I believed I read. Carry on. Sorry for the trouble.

[antonyp22]

I get nervous when I read that you are betting specifically on dogs with odds greater than 2.00, you shouldn't be limiting yourself as to what odds you bet on, you should be betting wherever your own probability projection is greater than the implied probability of the odds.

Any model that is not able to spit out a probability projection is flawed.

[MoMoneyMoVaughn]

Average SBR poster has $100 per post as bankroll. Roughly. My Br is decent. '

JJ Gold BR is over 10 mill

[LT Profits]

Average SBR poster has $100 per post as bankroll. Roughly. My Br is decent. '

JJ Gold BR is over 10 mill

I wish that was true.

[DiggityDaggityDo]

Average SBR poster has $100 per post as bankroll. Roughly. My Br is decent. '
JJ Gold BR is over 10 mill

[bihon]

Someone mentioned 1% of the bankroll and with an edge of 3% this seems very right, even 0.5%-0.75% per bet could be more suitable.
Don't even think of the biggest hype in the betting community - Kelly criterion and its variations, a sure way to disaster. It is way too aggresive for any reasonable betting and while being advertised as the tool for maximizing profits, its Risk of Ruin and Drawdown risks are absolutely insane. Most times a player will end betting pennies/cents.

[blizzztik]

Hi.

I have been betting on a daily basis for 15 full months and, unfortunately, after 10 winning months (out of 12) I have bumped into a consecutive 3-month losing streak that made me reconsider my betting approach and bankroll management.

I am using level stake betting, risking 2% of my bankroll on each bet. I make around 220 bets per month, apart from June-July when most leagues are on vacation so my bets are around 30-40. Weekends are the most busy days (as expected) with about 25-30 simultaneous bets on Saturdays and around 20 simultaneous bets on Sundays.

Up to now, I recalculated my 2% bet size after 1000 bets (i.e. after 4 months) but this 3-month losing break almost lost me my entire bankroll since I ended up betting a very big percentage of my roll.

Thus, I was wondering, when is the best time to recalculate this 2% per bet as the bankroll changes ? It is impossible to do it after each bet since I have 20-30 simultaneous bets every weekend. I was thinking of adjusting my bet size after each day, but I don't know if this is the best way to go : Losing streaks are followed by winning streaks and vice versa, so when a losing streak occurs (usually after a winning streak), I will start losing much money and also, when a winning streak comes (usually after a losing streak) I will be betting very few on each bet, thus resulting in a very slow comeback.

I was thinking of adjusting my bet size each month, but I am still not sure if this is the best way to go.

What's your opinion, based on my betting habits ?

Thanks !

I think you should bet size based on your confidence in the pick. If you think something is 2% roi and something else is 8% roi you should be maybe .5 units on the 2% and more like 2.5 units on the 8%. Obviously depends specifically but you can lower your risk of ruin by manually altering the bets based on others confidence or your own.

[TheLock]

. AFter a certain period of time, you don't need it because ERA tells you what you need to know..

Incorrect

[Toples]

if you flatbet 2% with your numbers (3,8 roi 2646 bets, 2.18 odds) you will go broke 55% of time becuz of not so big roi compared to odds
though AN AVERAGE profit will be 2 times your bankroll if your ROI is correct

if i understood correctly you adjusted your bet size after 2000 bets.
that still leaves you with 13% chance of going bankrupt after 646 bets using 2% stakes

that being said, im gonna explain how i do it
im adjusting stake size after i reach certain treshold

if you start with 5K (2% is $100 per bet), then change stake size when you hit either 5500 or 4500. so next bet size will be either 90 or 110. and so on...

[Playon]

I lowered my stakes by 50% in order to put money on all 35 selection and, guess what ??? The day was a huge winner

That's the way goes, everytime. And it should, stakes must be reduced when in downswing. You guys are talking about recalculating the stakes weekly or monthly, but I do it every day. Even twice a day if for example I have half of the bankroll at risk and I'm still betting with other half available, then again I reduce staking.

Right now it seems Marathon saved big money for me by setting limits just when my bankroll growth was at its peak and downswing about to start. I need to thank Marathon they gave me so much money and then saved the rest of it.

[Toples]

That's the way goes, everytime. And it should, stakes must be reduced when in downswing. You guys are talking about recalculating the stakes weekly or monthly, but I do it every day. Even twice a day if for example I have half of the bankroll at risk and I'm still betting with other half available, then again I reduce staking.

Right now it seems Marathon saved big money for me by setting limits just when my bankroll growth was at its peak and downswing about to start. I need to thank Marathon they gave me so much money and then saved the rest of it.

Marathon limited me to silly amounts (like $3 max on superbowl so you can expect the same soon

[achilleas 21]

hi.
you calculate your variance/drawdown and your max winless streak?
that is very important for you to know .Check also if you follow exactly the correct strategy as you started with.

if you want to increase the stakes ,in my opinion you have to see first your capital to be +50% .But you dont increase at once .You still wait up to the time that you will not go down again of that 50% line.So then you go to 150 if each stake was 100.That is a safe way since your hit rate is close to 47%

Maybe this negative you are now is absolutely normal but you can not know for sure before you check all
But notice that if you have an almost 3-4% roi ,after thousands of bets ,you will see that maybe at the last 200-300 or..bets ,your roi will go at -8% and later maybe go to plus 8 or something.That negative time if your stake is bigger than have to be it will be a big minus for your capital.(Better to check also ROI every 100-250-500-1000 bets)
So a much more safety solution for such a big number of bets,is 1% of your bank roll and increase 50% the bet after you go more than 60-70 % of your initial capital
one more think..
as long as you are betting,as much is possible to see more and more negative streaks and variance.So first time you ll increase your bet after your initial capital goes to 50+ % but second time you ll not increase when you goes to double !Do this with a progressive way
1st time you increase when goes to +50%
2nd time when goes to +150%
3rd time when goes to +300% or so
Cause if you bet always 2% and increase every +50% you will be in a big problem when you ll meet a winless streak of 15 or more bets and after this there are two or more winlless streaks coming after a few winning bets!Then is absolutelly normal to loose a huge part of your initial capital.Maybe more than 50% or maybe all after some thousand bets.Even if your system works.
So better safe..1% and increase slower and slower as time goes.

If you know your stats you can calculate all these
ps
since your system works,invest more money on that so as to bet bigger .Dont bet more than 1% and dont use kelly or other stuff.Just bet flat with safety and money will come easily with no problems and not hart attcaks

best wishes for winnings

[gui_m_p]

I am not sure I can quantify my edge per bet. I haven't backtested my system based on that so I don't believe Kelly is a good choice for me.

Here are the stats of my system's results so far :

2646 bets
Won-Lost-Void : 1248-1373-25
Hitrate : 47.62%
Average odds : 2.18
Yield : 3.8%

I am just not sure which bankroll management to use to maximize my winnings...

I made 1 million simulations of 2646 bets with odds of 2.18 and juice of 2% (which is conservative. The higher the juice, the harder is to obtain your record).

The chance of achieving 47.62% only by luck is a mere 0.2%.

Be conservative and focus on long term. You probably have a good model.

[A4K]

Any updates on your model?

[Mousey]

FWIW; I flat bet and recalculate my unit weekly (Monday morning).

[nash13]

I'm more a fan of recalculating seasonwise. So when your system/selection crteria is working good enough you can balance out eventual swings better. As for 2015 I have made 2100 selections, flat betting would have made 100 Units Profit, while with multiple staking I have about 280 Units Profit. If your selections are good, don't leave money on the table by going for flat bets only.

[Elko]

1% per unit. Why recalc?

[bocajuniorska]

did i see someone mention marathonbet in a thread where the issue is ROI %s and stuff? lol...

[CHUBNUT]

The problem with staking is the same as many other problem involving decision making, it becomes psychological. Nobody knows the future and random losing runs can happen with the best of systems. I've tried daily, weekly, monthly control depending on results and bankroll without any real success.I find a non specific review at any time relaxes the situation where one can make an informed choice rather than a mathematical or mechanical decision. For instance you may have increased your stakes on the back of expected good results only to find you've now hit a bad run. If its a system you believe in then carry on, if however the doubts are of genuine concern then a big drop in stakes may be called for. Like everything else with gabling, you need soe luck.

[KVB]

1% per unit. Why recalc?

Good point if you are making what you want to make from sports.

But to be truthful, raising the bankroll slowly, as suggested by Achilleas is paramount. I can tell you this...if you believe in the ceiling and that you can't do better than 55-57 percent over the long haul, then I would advise changing your bet size as little as possible.

I've posted about this many times but if you change your bet size, you change your break even point. I can safely tell you that with a bankroll of about $1000 building all the way to $150,000 you should change your bet size a little more than a dozen times...and NO MORE.

Achilleas 21 hits it on the head with a pretty realistic and sharp post above.

[tsty]

So all those bets were +EV?

i find that hard to believe ey

[the shadow]

to maximize your winnings ,you should recalculate every second.

[omedo]

Good point if you are making what you want to make from sports.

But to be truthful, raising the bankroll slowly, as suggested by Achilleas is paramount. I can tell you this...if you believe in the ceiling and that you can't do better than 55-57 percent over the long haul, then I would advise changing your bet size as little as possible.

I've posted about this many times but if you change your bet size, you change your break even point. I can safely tell you that with a bankroll of about $1000 building all the way to $150,000 you should change your bet size a little more than a dozen times...and NO MORE.

Achilleas 21 hits it on the head with a pretty realistic and sharp post above.

I allready read you other posts. I dindt did the math, but it wasnt proven that kelly criterion maximize profit?
Imagine a world when you know exactly what's your edge in every bet.


the second thought is, why so few changes from 1000 to 150 000??

Care to explain a litle further?





Edit:

Well, went to explore this forum a litle more and found my answer right below on a post by Ganchrow.

Thank you all.


Expected Growth vs Expected Value

[KVB]

I allready read you other posts. I dindt did the math, but it wasnt proven that kelly criterion maximize profit?
Imagine a world when you know exactly what's your edge in every bet.


the second thought is, why so few changes from 1000 to 150 000??

Care to explain a litle further?





Edit:

Well, went to explore this forum a litle more and found my answer right below on a post by Ganchrow.

Thank you all.


Expected Growth vs Expected Value

Sharp posts by Ganchrow for sure. No long term successful bettor uses full Kelly (it's simply too expensive in the major markets), but at a best a part of Kelly. Every time you change your bet size you change the break even point.

Because of the floating nature of +EV analysis in the major sport markets many bettors will find the edge disappears before they ever realized they had it.

For the majority of players, it would be fine to adjust the bet size as little as possible.

When you are betting as a business, taking a salary and preserving the bankroll, and not just pursuing the valuable hypothetical long term, adjusting the bet size to make it worth the investment and time (a decent ROI) is all I would recommend.

A proper bet size for one's bankroll is not really a subjective issue. I could argue that there is only one bet size appropriate for a given situation, and that's it.

Another issue I've noticed is that the definition of flat betting is muddy.

Betting to win a flat amount is not flat betting, risking the same on each bet is flat betting.

I always recommend keeping your risk flat, especially in a game like baseball where moneylines vary so much, and let the book's risk jump around instead of your own.

[evo34]

"Betting to win a flat amount is not flat betting, risking the same on each bet is flat betting."

Nope. If you risk $100 to win $100 on nine +100 bets and then risk $100 to win $1 on one -10000 bet, do you think that 10th bet is going to have anywhere near the expected value of the first nine? Of course not. Let's say your edge is is 2% on the +100 bets, so your expected win on each $100 bet is $2. Now how do you make your expected win equal $2 on a -10000 bet? Well, even if you assume max possible edge (100% chance of winning), you'd have to risk $200 to make the bet worth $2.

The opposite, of course, is true for 100-1 longshots. You should never risk $100 at those odds if your unit size for even money bets is $100.

If you want to flat bet, try to equalize the expected value of each bet, which will mean risking different amounts depending on the odds.

[KVB]

"Betting to win a flat amount is not flat betting, risking the same on each bet is flat betting."

Nope. If you risk $100 to win $100 on nine +100 bets and then risk $100 to win $1 on one -10000 bet, do you think that 10th bet is going to have anywhere near the expected value of the first nine? Of course not. Let's say your edge is is 2% on the +100 bets, so your expected win on each $100 bet is $2. Now how do you make your expected win equal $2 on a -10000 bet? Well, even if you assume max possible edge (100% chance of winning), you'd have to risk $200 to make the bet worth $2.

The opposite, of course, is true for 100-1 longshots. You should never risk $100 at those odds if your unit size for even money bets is $100.

If you want to flat bet, try to equalize the expected value of each bet, which will mean risking different amounts depending on the odds.

Good post Evo. For the most part, I agree here. I can see the "flatness" in what you are saying but it's really a different definition than what I am saying. It's part of that muddy part.

For this purpose, I would say that flat betting implies risking the same amount each time, without equalizing the expected value of each bet.

What is at risk is the money out there, regardless of the perceived expected value and especially regardless of the odds offered. Once the bet is made, probability does not matter (unless you are adding to the position by live betting.) Once the bet is made, that is the amount at risk. A loss is a loss, you don't lose more, you don't lose less with different outcomes (Asian handicaps can be a bit of an exception.)

I frame it in the sense of a sports betting as a business because that's what I do. It can be very difficult to assess a given edge on a given game because of the long term nature of calculating EV.

For most bettors not seeing this, it will cost far too much in vigorish to adjust. Look at baseball, if a bettor bets $100 on an even money bet (+100) then likes the next games favorite at -125.

When the bettor determines the same edge in both offerings, adjusting to win the same $100 (similar to your $2 example) essentially means the bettor would risk $125.

This happens all the time and many bettors are essentially saying one game is worth 25% more risk than the other. The same example could work for higher moneylines.

If one is really finding games that are worth 25% (or higher) more risk than other games, it may be wise to just bet those games.

After all, those higher bets are the only ones that will matter in the end.

Like I said, I can understand what you are saying by equalizing the expected value of each bet, but it can cause a lot of problems for most bettors. I guess I view it more like teaching new bettors, especially with questions like this one.

Remember, I am recognizing a floating type of EV that changes on bettors as well as not finding the markets as efficient as you do evo. I've seen +EV results over many NFL games but have also seen the same methods yield less than 40% for some of those individual seasons (you can even break up the seasons into parts.)

Because a book lists a team at -200 does not automatically make that book twice as favored mathematically to win over a team at +100. It's just the bookmakers forced opinion based on marketplace and its environment.

Too much perception, even by sharper forecasters, causes inefficiencies all the time. The books lead everyone around by the nose, even the sharper groups.

Sharp money can tighten a line, but only if the books make the move. This doesn't always happen as books often aim to take positions from the open.

Cheers evo, I'm glad you posted.

[omedo]

KVB, can you give 1 or 2 examples of what you mean? Probably it would clear that mud out of the way

[u21c3f6]

... I frame it in the sense of a sports betting as a business because that's what I do. It can be very difficult to assess a given edge on a given game because of the long term nature of calculating EV.

For most bettors not seeing this, it will cost far too much in vigorish to adjust. Look at baseball, if a bettor bets $100 on an even money bet (+100) then likes the next games favorite at -125.

When the bettor determines the same edge in both offerings, adjusting to win the same $100 (similar to your $2 example) essentially means the bettor would risk $125.

This happens all the time and many bettors are essentially saying one game is worth 25% more risk than the other. The same example could work for higher moneylines.

If one is really finding games that are worth 25% (or higher) more risk than other games, it may be wise to just bet those games.

After all, those higher bets are the only ones that will matter in the end. ...

KVB, The reason you can wager 25% more on a -125 than you can on a +100 with the same edge is the fact that the -125 has a higher probability of winning. It is Kelly which is a mathematical fact. The problem is of course is how well one can calculate their edge.

Joe.

[KVB]

KVB, The reason you can wager 25% more on a -125 than you can on a +100 with the same edge is the fact that the -125 has a higher probability of winning. It is Kelly which is a mathematical fact. The problem is of course is how well one can calculate their edge.

Joe.

I agree with the problem about calculating edge, but when I talk about +100 and -125 I am talking about what the books offer. Just because the market sits at -125 does not make it a fact that it has a 25% more chance of winning over a market price at +100. It could be, but it doesn't have to be.

If they were as efficient as some believe, then you make a valid point.

But I have always contended that the books simply aren't that efficient at all times and when we think of the forces that shape it, we really can see why.

I know this is a contentious subject but I see the markets more as moving towards efficiency rather than already being efficient. I also know that inefficiencies are often created purposefully and that this happens often.

I don't question the math facts behind Kelly but I do know that betting amounts at a "full Kelly" level according to your calculated edge through a typical NFL season is not a wise idea.

I seriously doubt any long term gambler is betting full Kelly amounts and surviving. I do know many that use a partial variation. I have separate bankrolls with separate strategies used at separate times.

It’s a bit like having diversification within my own results and some ways the only way I know how to clean up some of the lost vigorish.
But for most, because of the hazards of a typical season or set of seasons, I recommend bettors keep their amount at risk the same until they really have a handle on their situation.

Otherwise they are going to change the breakeven point over even the short term and pay too much in vigorish...the old killer.

The markets and those who shape it are wise to this behavior. It seems like we've hit on this before, maybe a different thread.

[KVB]

KVB, can you give 1 or 2 examples of what you mean? Probably it would clear that mud out of the way

Actually, depending on which examples you are looking for, this is best taught as the markets and season unfolds, real time.

...Because a book lists a team at -200 does not automatically make that book twice as favored mathematically to win over a team at +100. It's just the bookmakers forced opinion based on marketplace and its environment...

Certain times of the year this really stands out. One of the best betting weekends of the year for me is often Super Bowl weekend. The NCAA Hoops markets are loaded with unsophisticated money that most certainly influences the lines to be a bit out of whack.

Anymore, the swell during March Madness is evident but you have to be more patient. If you made it to Super Bowl Saturday, you already were patient.

Remember Michigan State in the NCAA Tourney? That line opened at 17.5 (which was way too high) and it stayed there. The money line was nowhere near representative of reality if each team plays a competitive game. I build a line with a different type of dynamic for outlier events like the Tourney and I knew I had the sharp line with a much, much closer game.

The moneyline was far out of where it should have been and I know the books knew it.

Despite that, it was reported to be one of the most heavily bet games and goes down as one of the biggest upsets in history. It wouldn’t have, had the line been smaller.

Maybe a tangent here, but what makes an upset? Is it because the market has a favorite? What about when the higher ranked team is an underdog…which, again, we see at times like the NCAA Tourney…which one is the upset?

Sometimes broadcasters play it off as one team is the favorite even if the line says otherwise.

What if it’s money? It seems rational to think that if Michigan State had been favored by less, with a smaller moneyline, but not so small to tip the scales, even more bettors and money would have fallen when they lost. But with a smaller line, even though more bettors lost more money, would they have called it a bigger upset?

Not likely.

These muddy definitions are a part of sportsbetting, separate from the math, it’s just the way it is and a part of how you view the markets.

[KVB]

KVB, can you give 1 or 2 examples of what you mean? Probably it would clear that mud out of the way

Or maybe you were thinking an example along these lines.

...Too much perception, even by sharper forecasters, causes inefficiencies all the time. The books lead everyone around by the nose, even the sharper groups.

Sharp money can tighten a line, but only if the books make the move. This doesn't always happen as books often aim to take positions from the open...

This is best shown real time. Last year I chimed in regularly in a thread about the CFL, an easy market for examples. I described some bookmaking techniques and how they were being applied to the betting marketplace by the books.

It was an excellent learn while you earn loaded with examples. I offered a sharp forecast for each game and plays that were verified with both market movement and results. I didn’t always bet with the forecast, but both were successful. The bets against TOTALS were positive money but the forecasts for TOTALS were the weakest of all forecasts. (That was part of a broader trend though).

I was trying to illustrate the value of knowing why a line opens where it does and why it moves where it moves. It was a good thread that needs a final update before getting into the next season. It is basically “rinse and repeat” as long as the forecasts are dynamic.

It was also a glimpse into what it is like trading around a sharp forecast in these give and take markets.

All tracking was done risking the same amount but tracking was done two ways with just betting the issue as well as pulling the trigger multiple times on some issues. Kelly was not used.

I will admit, I do not know that the trading was the most efficient I could have done. Right now any algorithm just has too many variables that have to be determined live for me to figure a perfect betting strategy.

Plenty of rules are followed, but it’s a bit like equity trading. The numbers get you so far, then there’s some judgement to put the pieces together, real time. Quantifying some is a work in progress.

At that point it becomes more of an art, than a science. Like equities, or any markets, it can be what separates top traders from the pack; experience.

[KVB]

I also think that bettors have to be honest with where they are. This is by no means aimed at anyone in this thread, but I think it does bettors good to ask themselves if they should be worrying about the exact Kelly proportion based on their edge or if they should worry about concluding winning bets and understanding why some bets lose.

If mistakes are made, it's best to learn from them.

[omedo]

I also think that bettors have to be honest with where they are. This is by no means aimed at anyone in this thread, but I think it does bettors good to ask themselves if they should be worrying about the exact Kelly proportion based on their edge or if they should worry about concluding winning bets and understanding why some bets lose.

If mistakes are made, it's best to learn from them.

Yeah, you are right for sure. I should not be worring so much on how much to bet.
I should be worrying about capping and trying to bet the least possible until i prove myself to be a profitable bettor.

But to be honest, the reason i had signed in this forum is cuz of that CFL thread. Is like gold for me. I'm still on the first 1/3 but allready changed my point of view of things.

[omedo]

Certain times of the year this really stands out. One of the best betting weekends of the year for me is often Super Bowl weekend. The NCAA Hoops markets are loaded with unsophisticated money that most certainly influences the lines to be a bit out of whack.

Anymore, the swell during March Madness is evident but you have to be more patient. If you made it to Super Bowl Saturday, you already were patient.

Remember Michigan State in the NCAA Tourney? That line opened at 17.5 (which was way too high) and it stayed there. The money line was nowhere near representative of reality if each team plays a competitive game. I build a line with a different type of dynamic for outlier events like the Tourney and I knew I had the sharp line with a much, much closer game.

The moneyline was far out of where it should have been and I know the books knew it.

You are right and i allready saw it in some games.

The books try to capture the public idea and ride it.

I'm not american and i dont follow none of your sports but that happens everywhere.
And i agree with you, you have to have patient to those moments and self confidence to pull the triger.
And i think that the last is the most dificult because you start to doubt yourself when every one seems to think the same way. The very opener odds influencies the view of the public.

But i can understand now why you derailed the question, for you its a no question and it will come to me when i have a more grasp of what's hapening.

Thank you once again