Arbusers

arbusers wrote:

VidaBlue wrote: 34.9 % ROI! What a week!

And here is a good question for you and everyone else: Would you lower your stake after a record performance week?

After a huge pay day, I would gladly take the Win:Win situation here,Which would be lowering the stakes. I will just not stake the recent ROI.

Say ROI 25%, $250 profit from a $1000 stake. I will remove this $250 from the stake and bet with $1000 for today. True I'm giving up on higher earnings, but as a value better bank protection takes the highest most priority.I would rather risk wining more, than losing more.

Congrats to Vidablue for the most excellent ROI.

So some of you guys claim that lowering and higering stakes would also afect on rulete good one!

I'd agree with @Wolfie here. Value betting is based on the probability theory and the higher than average ROI during the observed period would lead me to expect a somewhat lower than average ROI in the following period. So why not lower the stakes.

tomkruiz wrote: I'd agree with @Wolfie here. Value betting is based on the probability theory and the higher than average ROI during the observed period would lead me to expect a somewhat lower than average ROI in the following period. So why not lower the stakes.

i will copy this: expect a somewhat lower than average ROI in the following period. So why not lower the stakes.

You speak nonsence. Lower ROI is also +0,01%. And if you have positive ROI, its smart to bet still as much as posible. But like some of you speak I asume you expect negative ROI, but at same time you say you expeckt lower ROI (lower RIO can be positive). So first define what you wish to say. I asume you wish to say that you expect negative ROI.
No one can predict when negative ROI will come, only if you have time machine, or you are out of this world. MY ROI on first week of january was 35% insted of 5%. So some of you would consider to lower the stakes. .And till end of february my ROI didnt droped uner 15%. Also march never had a negative week. So with lowering stakes some of you would probably  left  money on table, becouse you would speculate only on ROI past performance.
Matematic is exact science, that can be proved. Based on matematic people beat bookies (some guys have thair model inside of thair brains). You can full matematic, if you can, than  you should get  a nobel price..

Great thread with great contributions

Wolfie wrote: Theoretically speaking the more bets you place the more you go toward expected value.

blackjack wrote: Any sharp deviation above the expected ROI should be followed by a lower staking plan.

I generally agree with these arguments when stakes are continuously adjusted according to the ROI in the observed period.

But it would be clearer if we were very certain of how much "expected value" or "expected ROI" really is. Then you could say that if you're basing your stake on a 8% observed ROI KNOWING that it should be 4%, it would be a good idea to lower it. The thing is, that this is very unclear as it depends on numerous variables, traditional sharps are not always so sharp. I am now quite convinced that I have some patterns where the expected ROI is indeed very high with 2 digits and other patterns with close to negative ROI. The sum of everything gives a stable single digit ROI. During these discoveries, via trial and error, I think it would make sense to increase the stakes in those very profitable markets even though the OVERALL observed ROI has moved above the expected ROI, because the overall ROI is still way below the observed ROI for these particular markets. The same goes for the opposite situation, lowering stakes in losing markets after a losing streak even though the overall observed ROI is less than expected overall ROI.

@4lexy4
One question. If you have a coin and toss it heads or tail. And 10 times in a row it goes head. In the eleventh time whats the probability for head and whats for tail ? Still 50% each ?
Probability is calculated based in information. And you have one information more. The information is that it was head 10 times in a row. So you have an advantage here.
Lets take another example. Real Madrid vs second hand team. Why would real madrid win ? Maybe it was just luck that made Real Madrid a big team and winning champions league. There is no mathematics that can back up why Real Madrid is a big team. Its just experience. And according to experience there is more probability for Real Madrid to win than the second hand team to win. This is what value bet is. Its not some fixed mathematical formula. If it were all people would win and bookies would go bankrupt.

4lexy4 wrote: So with lowering stakes some of you would probably  left  money on table, becouse you would speculate only on ROI past performance.

But what would you do if ROI past performance is all you got?

I think that many answers here assume a scenario where ROI is to be based upon one's actual betting history alone. Could it be that you base your answer upon something else which is not your own, such as pinnacle/asian lean?

In this case, obviously perceptions will be very different.

Wolfie wrote: If you have a coin and toss it heads or tail. And 10 times in a row it goes head. In the eleventh time whats the probability for head and whats for tail ? Still 50% each ?

Excellent example very transferable to value betting. What actually happened here, was that the coin used had a very displaced center of gravity making head much more likely than tails. It took ages and thousands of tosses to realize this, because the world got so fixated on the fact that a coin toss is always a 50/50 probability. Meanwhile someone took advantage of these early observations.

Wolfie wrote: @4lexy4
One question. If you have a coin and toss it heads or tail. And 10 times in a row it goes head. In the eleventh time whats the probability for head and whats for tail ? Still 50% each ?
Probability is calculated based in information. And you have one information more. The information is that it was head 10 times in a row. So you have an advantage here.

You found how to beat roulette now, I am out of this conversation

Alright @4lexy4, chill buddy, we're just making a conversation here, anyone is free to deduce how ever they want. You are right in a sense that I haven't phrased myself correctly before.

What I wanted to say that, when deciding whether or not to amend the stake size where ROI is the only factor, one should need to observe the overall ROI. If the betting during the previous period yielded the ROI higher than average, depending on the total observed period length, it's possible that the overall ROI was affected by this period too. Shorter the observed period, higher the impact on the overall ROI, obviously. If your great ROI in the previous period had a big impact on your overall ROI and you came above the expected value line, I'd say it's safe to lower the stake then.

However, if you are already in the middle of a long run, this might be a micro-optimisation. Because in the of the day, if you have got your EV formula and staking plan straight, it's likely your ROI is steady around the line you expected it to be.

If you lower your stakes every time the ROI goes over the expected one, what happens if it goes under the expected one? Do you up your stakes, or just bring them back as they were planned? I don't think you'd up them, because that would be gambling.

I think adjusting stake size based solely on previous outcomes without causation is incorrect. If you identify the reason then changing your stake is appropriate; "the coin used had a very displaced center of gravity making head much more likely than tails".

Value betting performance week 44 2019 (oct 28 - nov 3)



Period
# of bets
ROI
ROI (flat stake)


week 44
[align=center]106[/align]
20.0 %
[align=right]14.3 %[/align]


2019
[align=center]3,328[/align]
7.4 %
[align=right]6.4 %[/align]



Top sports (sorted list based upon turnover)


Sport
# of bets
ROI
ROI (flat stake)


Hockey
[align=center]29[/align]
25.6 %
[align=right]19.4 %[/align]


Soccer
[align=center]24[/align]
19.3 %
[align=right]18.3 %[/align]


Basket
[align=center]16[/align]
14.7 %
[align=right]-3.4 %[/align]


Esports
[align=center]23[/align]
22.2 %
[align=right]19.8 %[/align]


*ROI (flat stake) is just the hypothetical result had a flat stake been used

Still on the lucky side of variance!

Thank you for all the contributions on adjusting the stakes last week. I took the sum of all opinions (including my own) and did the following: Plus 10% factor on the stake size for those very profitable markets (sport / type of bet). Minus 10% factor in general. This is not necessarily equal to the actual change in stake size, since there are other contributors (ev, bankroll, etc.), but it is just a factor that is now considered. It is a very rewarding feeling, when I compare the ROIs above with the ROIs had I been using a simple flat stake strategy. All would have been lower, extreme case in basket last week, 14.7 % ROI vs -3.4 ROI. Staking is very important.

These statistics are only on shop value betting based upon soft vs softs, soft vs sharps, odds fluctuations, etc ... Other things, such as value betting on sharps and value betting on value traps are not depicted here.

Feel free to comment or start a discussion from this post. The discussion last week was a good educational process for me and maybe for others as well.

    W44 (106) 20.0%    W43 (111) 34.9%    W42 (167 bets) 4,0%    W41 (211 bets) 4,3%    W40 (195 bets) 4,1%    W39 (135 bets) 15,9%    W38 (108 bets) -6,1%    W37 (105 bets) 2,9%    W36 (122 bets) 0,3%    W35 (132 bets) 4,8%    W34 (163 bets) 5,2%    W33 (110 bets) 18,4%    W32 (132 bets) 2,4%    W31 (101 bets) 2,1%    W30 (59 bets) 6,8%    W29 (85 bets) 12,4%    W28 (2 bets) -100%    W27 (6 bets) 39,7%    W26 (31 bets) -4%    W25 (49 bets) 0,4%    W24 (61 bets) -27%    W23 (22 bets) -62,5%    W22 (31 bets) -53,3%    W21 (85 bets) 12,9%    W20 (68 bets) 33,5%    W19 (44 bets) -8,4%    W18 (19 bets) 3,9%    W17 (119 bets) 5,9%    W16 (41 bets) 16,3%    W15 (80 bets) 20,7%    W14 (72 bets) -5,7%    W13 (40 bets) 18,4%    W12 (20 bets) 34,2%    W11 (23 bets) -41,4%    W10 (0 bets) 0%    W9 (0 bets) 0%    W8 (0 bets) 0%    W7 (17 bets) -58,1%    W6 (50 bets) 21,6%    W5 (77 bets) 20,6%    W4 (76 bets) 23,7%    W3 (97 bets) 8,9%    W2 (100 bets) 2,5%    W1 (68 bets) 25,9%

Thank you for sharing.

Compared to the last week's, it seems that this week's performance went down to a large extent.
Of course, we do not have more data about the odds etc, but the general picture as it shows here is that after a frenzied week a more humble week followed.
Is this the case or am I missing something?
Because if indeed this is the case, then my hypothesis and suggestion to lower stake might be proved correct. I m using the word ''might'' because before anything else, we must calculate the long term performance of your project.

We have seen in a number of simulations, that flat staking is not the best way to practice Value Betting. Out of the most common and popular staking plans, proportional betting seems to be the most resilient. But in my opinion, a proportional staking plan could be further developed with the analysis of stats and markets in order to get to the optimum level for the most favorable situation or level for growth, or the golden mean.

Yes, this week's performance went down compared to week before that. But those two weeks are not typical weeks. It went from "frenzy week" to "awesome week". 20.0 % ROI is still way above general expectations, which are close to 7% ROI - I think the term "humble" is an understatement. I would love to only have weeks as those as this would mean early retirement On such a short term, variance must be the biggest contributor, it is a very low sample size, but still it is useful to talk about this week in the context of your hypothesis.

Based on last week's discussion, I think the hypothesis makes sense, so therefore I applied a general correction of minus 10% to the stake size. However, I did not do that in very special cases, where I so far have had huge ROIs. In those special cases, I applied a plus 10% correction factor instead. So first a general minus 10% correction, followed by another +20% correction. It is shop betting, I do not bring a calculator, it is not possible to split a dollar bill in half, there are other contributors to staking as well, etc. - so it is just a mental awareness that I use during action. It definitely means that stake sizes have changed, but it is not accurate.

I have taken last week's performance and categorized it:



[align=left]bet type[/align][align=center]# of bets[/align][align=center]    av. odds[/align][align=center]    ROI[/align]


[align=left]normal vb[/align][align=center]86[/align][align=center]2.24[/align][align=center]3.9 %[/align]


[align=left]special vb[/align][align=center]19[/align][align=center]2.63[/align][align=center]65.0 %[/align]


[align=left]vt bet*[/align][align=center]1[/align][align=center]3.25[/align][align=center]225.0 %[/align]


*This is a bet based upon value trap assumption. It was unintentionally included in the previous table, so therefore it is included here as well

First of all, I have been lucky with the 19 special vbs. ROI for special vbs should be in the 10-25 % regime and not 65 %, but it is such a small sample size. The ROI for the normal vbs seems more in line with the expected. Overall ROI this year is around 7% but if this is a combination of normal vbs and special vbs, it may seem fair that the ROI for normal vbs is be a bit lower, like 4%.

I believe in the hypothesis, but I don't think these statistics convincingly prove it, mainly because:

• sample size is small
• this was not an average week
• high ROI special vbs are noise

In order to get closer to the answer, I think that more data should be added and the special vbs should be left out.

And just to clarify: I have always used a proportional staking plan. These discussions around the staking size are exactly the endeavors in the search for a golden mean.

Value betting performance week 45 2019 (nov 4 - nov 10)



Period
# of bets
ROI
ROI (flat stake)*


week 45
[align=center]92[/align]
8.4 %
[align=right]3.7 %[/align]


2019
[align=center]3,420[/align]
7.4 %
[align=right]6.3 %[/align]


*ROI (flat stake) is the calculated result if a simple flat staking strategy had been used

Top sports (sorted list based upon turnover)


Sport
# of bets
ROI
ROI (flat stake)


Basket
[align=center]28[/align]
6.1 %
[align=right]-18.2 %[/align]


Soccer
[align=center]18[/align]
45.1 %
[align=right]37.8 %[/align]


Esports
[align=center]24[/align]
2.3 %
[align=right]10.2 %[/align]


Hockey
[align=center]7[/align]
-28.7 %
[align=right]-21.7 %[/align]



Decent week. Continuing with the adjustments to the staking plan that were added 2 weeks ago. I had 20 special value bets this week (significantly high ROI bets with additional 10% added in stake size) that achieved an ROI of 13.2 %. It is nice to see that simple flat staking is consistently inferior to the actual staking plan.

    W45 (92) 8.4%    W44 (106) 20.0%    W43 (111) 34.9%    W42 (167 bets) 4.0%    W41 (211 bets) 4.3%    W40 (195 bets) 4.1%    W39 (135 bets) 15.9%    W38 (108 bets) -6.1%    W37 (105 bets) 2.9%    W36 (122 bets) 0.3%    W35 (132 bets) 4.8%    W34 (163 bets) 5.2%    W33 (110 bets) 18.4%    W32 (132 bets) 2.4%    W31 (101 bets) 2.1%    W30 (59 bets) 6.8%    W29 (85 bets) 12.4%    W28 (2 bets) -100%    W27 (6 bets) 39.7%    W26 (31 bets) -4%    W25 (49 bets) 0.4%    W24 (61 bets) -27%    W23 (22 bets) -62.5%    W22 (31 bets) -53.3%    W21 (85 bets) 12.9%    W20 (68 bets) 33.5%    W19 (44 bets) -8.4%    W18 (19 bets) 3.9%    W17 (119 bets) 5.9%    W16 (41 bets) 16.3%    W15 (80 bets) 20.7%    W14 (72 bets) -5.7%    W13 (40 bets) 18.4%    W12 (20 bets) 34.2%    W11 (23 bets) -41.4%    W10 (0 bets) 0%    W9 (0 bets) 0%    W8 (0 bets) 0%    W7 (17 bets) -58.1%    W6 (50 bets) 21.6%    W5 (77 bets) 20.6%    W4 (76 bets) 23.7%    W3 (97 bets) 8.9%    W2 (100 bets) 2.5%    W1 (68 bets) 25.9%