Hi fellas,
I don't know if there's anyone explained this before but here we go...
So after getting few times in a row returned bets playing over/under handicap I realized how to "prevent" possible return and scale it up.
So for example:
Arb is in over/under 3.0 (2,86% profit)
Odd for under 3.0 is 1.8
Odd for over 3.0 is 2.4
Considering (/for easier calculation) that stake is €100
As an classic arber You would place bet like this:
On odd 1.8 - €57,14 (profit: €2,85)
On odd 2.4 - €42,86 (profit: €2,86)
But instead of being classic arber let's move whole profit on over 3.0 and place bets like this:
On odd 1.8 - €55,56 (profit: €0,01) - BE (break-even)
On odd 2.4 - €44,44 (profit: €6,66) - bigger profit
Now the key part:
Watch or track result of the game and then if in any moment are scored exactly 3 goals, place additional bet on live betting that will be no more goals with whole or part of the profit, depending on your style and greediness, that You would made if there were 4 or more goals (€6,66).
For example exactly 3 goals are scored in 1st half and odd on live, for no more goals in the game, is 2,4 You can arb it with "over 3.0" in a way
Profit / 1 + live odd = amount to place on live bet
Formula breakdown in simple way:
Profit - stake = win on over
live odd * stake = win on exact number of goals
--------------------------------------------------------------------
since we are arbing both wins need to be same and based on that:
live odd * stake = Profit - stake
--------------------------------------------
live odd = 2,4
Profit = 6,66
--------------------------------------------
2,4 * stake = 6,66 - stake | + stake
2,4stake + stake = 6,66
3,4stake = 6,66 | / 3.4
stake = 6,66 / 3,4
stake = 1,9588235294117647058823529411765
stake = €1,96
Formula confirmation:
Profit - stake = win on over
6,66 - 1,96 = 4,7 (€4,7)
--------------------------------------------
live odd * stake = win on exact number of goals
2,4 * 1,96 = 4,704 (€4,7)
--------------------------------------------
win on over = win on exact number of goals = €4,7
Scaling up arbs when return is possible
-
Mystic
- Has experience
- Karma: 2
Post
Scaling up arbs when return is possible
When You know how it works then You should know how to break it. 
-
arb12
- Totally Pro
- Karma: 23
Post
You're a young market enthusiast looking for yours way, and let me drop a few words here.
On the one hand, I won't comment in detail on the technique that you mentioned. Every approach that suits your needs is good. On the other hand, I'll try to encourage you to go much further - in the value betting area, provided that you can pre-estimate great risks.
You are afraid of returning the full amount at the quota of An=3 goals scored. But the most equilibrium market in your example is not that O/U 3 goals, but Over/Under 3.25 (hypothetically fractional return of the amount in case of 3 scored goals) and given that the implied probabilities here are much closer to 50:50, you could consider more valuable moves on your part by assuming the O/U 3.25 market - your calculated probabilities may differ from those 50:50 implied by the agencies.
Also, returning your stake is not a bad thing. In my opinion, when a gray area is touched in your preliminary analysis of the O/U & Asian Handicap markets, perhaps the returned stake in the case of a given integer is better than losing when value betting is in effect.
The general hypothesis you've covered is as follows:
(1). Initially, you've chosen to make an equivalent of a free bet, hypothetically profitable when β goals are scored. Here α<β goals scored; α∈{0, A1, A2, ..., An}. The set {0, ..., An} is zero profit, lossless, at that stage only.
(2). At a further stage of the runtime, if the An numbers of goals are scored, you are about to inject profit into the "exactly An goals scored" position through handling a hypothetical equivalent of "against β goals scored" markets, where β∈{An+1, An+2, ..., ∞} and α<β.
When step 2 is completed, already γ scored goals become profitable, where γ∈{An, An+1, An+2, ..., ∞}.
Specific data in your example is as follows:
(1). Your initial operations are completed at a quota An=3. So in a case of 0, 1, 2 or 3 scored goals you're break-even.
Exactly three goals were scored before the half-time. Therefore β∈{4, 5, ..., ∞}.
(2). You have reached the final set of Gamma elements by intervening at the odds of 2.4 in your example.
I see plenty of opportunities for optimization through a couple of Socratic questionings. Some examples could be:
- What's the overall valuation of the long-term application of your method in the next X games, compared to the value betting positions (at a given risk factor θ)?
Here X can be equal to 1000, 10 000 and so on. The Theta coefficient is according to your risk tolerance;
- Same as above, but let's assume the pre-selected "An" position is not reached at all. The missed profits in the next X games when your hedging method is applied, compared to value betting positions at any chosen Theta coefficient are equal to...?
Calculate that assuming X=10 000.
- and so on, and so forth.
Re: Scaling up arbs when return is possible
Hey,
You're a young market enthusiast looking for yours way, and let me drop a few words here.
On the one hand, I won't comment in detail on the technique that you mentioned. Every approach that suits your needs is good. On the other hand, I'll try to encourage you to go much further - in the value betting area, provided that you can pre-estimate great risks.
You are afraid of returning the full amount at the quota of An=3 goals scored. But the most equilibrium market in your example is not that O/U 3 goals, but Over/Under 3.25 (hypothetically fractional return of the amount in case of 3 scored goals) and given that the implied probabilities here are much closer to 50:50, you could consider more valuable moves on your part by assuming the O/U 3.25 market - your calculated probabilities may differ from those 50:50 implied by the agencies.
Also, returning your stake is not a bad thing. In my opinion, when a gray area is touched in your preliminary analysis of the O/U & Asian Handicap markets, perhaps the returned stake in the case of a given integer is better than losing when value betting is in effect.
The general hypothesis you've covered is as follows:
(1). Initially, you've chosen to make an equivalent of a free bet, hypothetically profitable when β goals are scored. Here α<β goals scored; α∈{0, A1, A2, ..., An}. The set {0, ..., An} is zero profit, lossless, at that stage only.
(2). At a further stage of the runtime, if the An numbers of goals are scored, you are about to inject profit into the "exactly An goals scored" position through handling a hypothetical equivalent of "against β goals scored" markets, where β∈{An+1, An+2, ..., ∞} and α<β.
When step 2 is completed, already γ scored goals become profitable, where γ∈{An, An+1, An+2, ..., ∞}.
Specific data in your example is as follows:
(1). Your initial operations are completed at a quota An=3. So in a case of 0, 1, 2 or 3 scored goals you're break-even.
Exactly three goals were scored before the half-time. Therefore β∈{4, 5, ..., ∞}.
(2). You have reached the final set of Gamma elements by intervening at the odds of 2.4 in your example.
I see plenty of opportunities for optimization through a couple of Socratic questionings. Some examples could be:
- What's the overall valuation of the long-term application of your method in the next X games, compared to the value betting positions (at a given risk factor θ)?
Here X can be equal to 1000, 10 000 and so on. The Theta coefficient is according to your risk tolerance;
- Same as above, but let's assume the pre-selected "An" position is not reached at all. The missed profits in the next X games when your hedging method is applied, compared to value betting positions at any chosen Theta coefficient are equal to...?
Calculate that assuming X=10 000.
- and so on, and so forth.
Last edited by arb12 on Thu Dec 14, 2023 12:50 pm, edited 1 time in total.
-
arb12
- Totally Pro
- Karma: 23
Post
Re: Scaling up arbs when return is possible
Another possible optimization task regarding your method:
Due to your mention of live odds of 2.4 for "No more goals", when three goals are scored in the first half, let's assume for simplicity reasons the equivalent market is Under 3.5 goals scored, and the condition "α<β goals scored" still holds. I assumed that those well-chosen desired odds were preselected by you here because of the target value for your working Model.
Let's again see the creation of the final Gamma elements by the final process in which the first β-element is involved.
Observing quite a lot of mixed samples of data that contain some odds provided by Sharps and Softs inside gives us the confidence to say when 3 goals are scored till half-time, the next part will start with a quota of at least 4 (low-medium further goals scoring expectation) or higher (odds>8 were spotted when the extremely high further goals scoring expectation is in effect) - here I mean the offered odds for "Under (the first β-element minus 0.5 goals)" in your set of β elements.
If you analyze a very long mixed sample of your data (again Sharps and Softs' odds are inside), you could notice that in a large percentage of the events played, your targeted odds @2.4 for the "Under 3.5" market have occurred in a short time interval whose midpoint was circa the 68th minute. Further analysis will extract the midpoint in some specific circumstances - for example, when the Favorite has locked up the game after the 1-st half and later only the defensive Department of the Favorite was interested in struggling at a high level (in some cases even earlier than the 62nd minute). And some games in top-notch leagues with an extremely large goal expectancy from the market - that midpoint was spotted even after the 75th minute. Not to mention the influence of hundreds of analyzed factors such as red cards given, "dead rubber", heavy rain when one of the teams is much more technical, and so forth. Every one of these hundreds could trigger a hypothetical value in the dynamic Model, built by you, and in my view, that's much better in the long run than any arb/hedge when the risks are handled.
So, in the 46th minute, when the second half starts, you could immediately create the most profitable Gamma elements by processing the aforementioned Statistically-proven interval of (circa 4 to >8).
Additional possible optimization tasks here would be:
- If you handle the "Middles" very well (as in the NBA & NCAAB Basketball etc), that could help you generate a lot of ideas here - in the 46th min, you could immediately accept much lower odds, offered regarding the equivalent of "Under" the second, third and so on elements of the set {An+1, An+2, An+3, ... , ∞}, which can immediately provide very different gain for the An-element than if handling the (4, ..., 8, ... ) odds interval if you provide for Under (the first β-element minus 0.5 goals). In the end, after a lot of "middling-style" operations on the market, you can achieve δ elements, where δ∈{An, An+1, An+2, ..., ∞} and also γ∈{An, An+1, An+2, ..., ∞}, but while every Gamma element gains equal profit by design, every Delta element is different due to your triggered operations of processing of Beta elements differently!
- Additional various analyses as the aforementioned (whether the value of γ or δ elements would be better for your Model in a given match), but additional Time processing is involved. Why? As you can see, there is one more thing to keep in mind here - the time interval between the halftime (when three goals were scored in your example and the offered odds' interval of (4, ..., 8, ... ) is in effect here), and the midpoint mentioned above (when your target odds could be in effect). Your long-term forecast for the next thousands of X games, how often will the desired by you odds be in effect and you will have equal profit on both legs of that arb, versus all the profit on one side and zero on the other due to scoring goals before the desired odds occur? In other words, does the analysis of multiple X games advise you of keeping more profitable Beta elements and no profitable An-element, which belongs to Alpha elements at this point, or converting them into less profitable Gamma elements - de facto reduced profit Beta elements plus the profitable 'An" element? This analysis can help you better select scenarios for the desired "An" element. Or in case your Model today is tuned up to better value at producing δ elements - exactly which market "middling" operations and when should they be done in order to optimize the required Delta elements?
- And so on, and so on...
Due to your mention of live odds of 2.4 for "No more goals", when three goals are scored in the first half, let's assume for simplicity reasons the equivalent market is Under 3.5 goals scored, and the condition "α<β goals scored" still holds. I assumed that those well-chosen desired odds were preselected by you here because of the target value for your working Model.
Let's again see the creation of the final Gamma elements by the final process in which the first β-element is involved.
Observing quite a lot of mixed samples of data that contain some odds provided by Sharps and Softs inside gives us the confidence to say when 3 goals are scored till half-time, the next part will start with a quota of at least 4 (low-medium further goals scoring expectation) or higher (odds>8 were spotted when the extremely high further goals scoring expectation is in effect) - here I mean the offered odds for "Under (the first β-element minus 0.5 goals)" in your set of β elements.
If you analyze a very long mixed sample of your data (again Sharps and Softs' odds are inside), you could notice that in a large percentage of the events played, your targeted odds @2.4 for the "Under 3.5" market have occurred in a short time interval whose midpoint was circa the 68th minute. Further analysis will extract the midpoint in some specific circumstances - for example, when the Favorite has locked up the game after the 1-st half and later only the defensive Department of the Favorite was interested in struggling at a high level (in some cases even earlier than the 62nd minute). And some games in top-notch leagues with an extremely large goal expectancy from the market - that midpoint was spotted even after the 75th minute. Not to mention the influence of hundreds of analyzed factors such as red cards given, "dead rubber", heavy rain when one of the teams is much more technical, and so forth. Every one of these hundreds could trigger a hypothetical value in the dynamic Model, built by you, and in my view, that's much better in the long run than any arb/hedge when the risks are handled.
So, in the 46th minute, when the second half starts, you could immediately create the most profitable Gamma elements by processing the aforementioned Statistically-proven interval of (circa 4 to >8).
Additional possible optimization tasks here would be:
- If you handle the "Middles" very well (as in the NBA & NCAAB Basketball etc), that could help you generate a lot of ideas here - in the 46th min, you could immediately accept much lower odds, offered regarding the equivalent of "Under" the second, third and so on elements of the set {An+1, An+2, An+3, ... , ∞}, which can immediately provide very different gain for the An-element than if handling the (4, ..., 8, ... ) odds interval if you provide for Under (the first β-element minus 0.5 goals). In the end, after a lot of "middling-style" operations on the market, you can achieve δ elements, where δ∈{An, An+1, An+2, ..., ∞} and also γ∈{An, An+1, An+2, ..., ∞}, but while every Gamma element gains equal profit by design, every Delta element is different due to your triggered operations of processing of Beta elements differently!
- Additional various analyses as the aforementioned (whether the value of γ or δ elements would be better for your Model in a given match), but additional Time processing is involved. Why? As you can see, there is one more thing to keep in mind here - the time interval between the halftime (when three goals were scored in your example and the offered odds' interval of (4, ..., 8, ... ) is in effect here), and the midpoint mentioned above (when your target odds could be in effect). Your long-term forecast for the next thousands of X games, how often will the desired by you odds be in effect and you will have equal profit on both legs of that arb, versus all the profit on one side and zero on the other due to scoring goals before the desired odds occur? In other words, does the analysis of multiple X games advise you of keeping more profitable Beta elements and no profitable An-element, which belongs to Alpha elements at this point, or converting them into less profitable Gamma elements - de facto reduced profit Beta elements plus the profitable 'An" element? This analysis can help you better select scenarios for the desired "An" element. Or in case your Model today is tuned up to better value at producing δ elements - exactly which market "middling" operations and when should they be done in order to optimize the required Delta elements?
- And so on, and so on...
-
arb12
- Totally Pro
- Karma: 23
Post
Re: Scaling up arbs when return is possible
In your example, you've used that technique for hedging purposes, but one possible Socratic questioning is as follows - is there a possibility to switch to value betting techniques, that are more profitable in the very long run for the next X games, considering taking the additional reasonable risks (determined by your Theta) by your side?
Let me present to you examples from my recent practice.
I have to stress that for simplicity reasons only a few factors are estimated here in my further examples, while in reality, the estimated by my Model factors are hundreds...
Low-risk position, Europe.
AEK Athens vs. Brighton & Hove.
There were many magnificent trading and value betting markets here, but I'll only mention the Asian handicap attack of (0.75), while a lot of agencies suggested AH (0.25) and AH (0.5). In the end, it was half profit, half void.
Reasons: A lot of weight was given to the importance of Vida's play in the defensive department of the AEK team as a whole (and from time to time, he attacked on the right flank). From that point on, I noticed different nuances in the defensive behavior when the team competed in domestic competitions and when the team competed in internationals. Vida's performance in AEK vs. Giannina and vs. Lamia was awesome, but vs. Marseille away, and later at home wasn't the same quality, and also I think the defensive department's stability in general deteriorates by XXX points when Vida is not on his maximum.
AH (0.75) odds were offered over my calculated odds, so the hypothetical value shined.
Self-criticism here: While modeling that clash, among the leading hypotheses was the following: a given likelihood that one of the possible scenarios is 1:1 after the 70th min, red card given and the home team is looking for a second goal. At that point (possibility of 1:1 and 2:1) my AH would be nonsense.
But that working hypothesis came out of the top highly likely outcomes and later I attacked Brighton's trailing by (0.75) AH instead of (0.5) due
to possibility of De Zerbi's attacking explosiveness at the end (Arsenal vs Brighton, May 2023, Man United vs Brighton, September 2023).
Let me present to you examples from my recent practice.
I have to stress that for simplicity reasons only a few factors are estimated here in my further examples, while in reality, the estimated by my Model factors are hundreds...
Low-risk position, Europe.
AEK Athens vs. Brighton & Hove.
There were many magnificent trading and value betting markets here, but I'll only mention the Asian handicap attack of (0.75), while a lot of agencies suggested AH (0.25) and AH (0.5). In the end, it was half profit, half void.
Reasons: A lot of weight was given to the importance of Vida's play in the defensive department of the AEK team as a whole (and from time to time, he attacked on the right flank). From that point on, I noticed different nuances in the defensive behavior when the team competed in domestic competitions and when the team competed in internationals. Vida's performance in AEK vs. Giannina and vs. Lamia was awesome, but vs. Marseille away, and later at home wasn't the same quality, and also I think the defensive department's stability in general deteriorates by XXX points when Vida is not on his maximum.
AH (0.75) odds were offered over my calculated odds, so the hypothetical value shined.
Self-criticism here: While modeling that clash, among the leading hypotheses was the following: a given likelihood that one of the possible scenarios is 1:1 after the 70th min, red card given and the home team is looking for a second goal. At that point (possibility of 1:1 and 2:1) my AH would be nonsense.
But that working hypothesis came out of the top highly likely outcomes and later I attacked Brighton's trailing by (0.75) AH instead of (0.5) due
to possibility of De Zerbi's attacking explosiveness at the end (Arsenal vs Brighton, May 2023, Man United vs Brighton, September 2023).
-
arb12
- Totally Pro
- Karma: 23
Post
Re: Scaling up arbs when return is possible
High-risk position in Europe.
Some leagues could provide value opportunities, due to odds-makers' errors - spotted in Romania, Greece, Türkiye...
Sometimes, value amplification is possible at a bigger level of risk, provided that the safe zone of the implied value position has been crossed.
Gençlerbirliği vs Göztepe. New Göztepe's coach Stoilov is known for practicing mid-directional attacking and pressing style of soccer, using a maximum of four (often three) dedicated defenders. I remember Astana Kazakhstan under Stoilov - draws vs Atlético de Madrid, Benfica and Galatasaray, victory over Slavia in Prague... Levski under Stoilov in 2022 eliminated PAOK after 2:0 in Sofia and 1:1 in Thessaloniki and even as a big pre-match underdog in Thessaloniki, he didn't order a defensive style, but ordered a saturated midfield and a small number of dedicated defenders, playing away from their own penalty area.
The game away vs the classy Gençlerbirliği team in Ankara had to be the first serious test away for Stoilov, who started to transform the defensive specialist Göztepe. In my opinion, too many players are specialized in defense and have a no-press style of play under the previous coach. Additionally, when the starting line-ups were announced, the best Göztepe player Nukan wasn't there again, don't know why.
On the other hand, I think Gençlerbirliği has serious goal-scoring potential for the 1Lig competition, despite scoring very few of them.
A very important thing was to build probabilities for a successful transformation of Göztepe from defensive into a mid-directional attacking team through some risky disruption of previous interactions in the best defensive department in 1Lig so far. Other pre-assessments were about the serious, but low-efficiency attacking potential for Gençlerbirliği so far (and also their mediocre defensive department so far which has cost them a lot of points). That's why my Model triggered high risk to attack the big value against the Asian Handicap of (0.75), while the agencies proposed as main lines the Asian Handicaps (0.0) and (0.25).
Self-criticism here: If Gençlerbirligi had scored the penalty, and if their player hadn't been sent off (at 0:0), and if their goal hadn't been disallowed by VAR at 0:1, then probably the Göztepe team wouldn't score three goals and probably my handicap would be lost. Only the right pre-estimation assigned to the home attacking and defensive dept (and also the late perfect game of the visitors) saved the handicap...
Extremely high-risk position in South America.
Atlético Mineiro vs Grêmio.
The Model triggered a risky possible attack about a BTTS market position. Why?
- Recently Suárez scored three vs Botafogo away;
- Perfect Suárez's goal-scoring form as a whole;
- The observations pinpointed an extremely high dependence of Grêmio's attacking department on Suárez's moves, let's say estimated as XXY points.
- However, Marcelo Bielsa put Suárez on the bench vs Argentina away;
- Bielsa gave Suárez only 20 playing minutes vs Bolivia and the level wasn't as usual...
- Overall, Scolari didn't overcome most of Sampaoli's level at Atlético (say YZ points);
- And many more vectors, of course.
All of this helped me build up the likelihood of Atlético not conceding a goal against Grêmio's atomic bombardiers, trying to predict exactly how Scolari would disrupt the interplay between Suárez and Co. I'm pretty sure the experienced Scolari had studied carefully how Suárez & Grêmio turned Botafogo around - from 3:1 to 3:4 within minutes.
Self-criticism: Building a forecast against the best player in Brasileirão Série A 2023 Luis Suárez is more than madness, indeed.
I personally enjoy watching one of the last triads of old-school romantic soccer: Uruguay, Bielsa and Luis Suárez. I also remember the great Uruguayan style under Óscar Tabárez...
Also, when I have enough free time, I'm about to start modelling some working hypotheses about Brasil under Carlo Ancelotti's future governance...
Some leagues could provide value opportunities, due to odds-makers' errors - spotted in Romania, Greece, Türkiye...
Sometimes, value amplification is possible at a bigger level of risk, provided that the safe zone of the implied value position has been crossed.
Gençlerbirliği vs Göztepe. New Göztepe's coach Stoilov is known for practicing mid-directional attacking and pressing style of soccer, using a maximum of four (often three) dedicated defenders. I remember Astana Kazakhstan under Stoilov - draws vs Atlético de Madrid, Benfica and Galatasaray, victory over Slavia in Prague... Levski under Stoilov in 2022 eliminated PAOK after 2:0 in Sofia and 1:1 in Thessaloniki and even as a big pre-match underdog in Thessaloniki, he didn't order a defensive style, but ordered a saturated midfield and a small number of dedicated defenders, playing away from their own penalty area.
The game away vs the classy Gençlerbirliği team in Ankara had to be the first serious test away for Stoilov, who started to transform the defensive specialist Göztepe. In my opinion, too many players are specialized in defense and have a no-press style of play under the previous coach. Additionally, when the starting line-ups were announced, the best Göztepe player Nukan wasn't there again, don't know why.
On the other hand, I think Gençlerbirliği has serious goal-scoring potential for the 1Lig competition, despite scoring very few of them.
A very important thing was to build probabilities for a successful transformation of Göztepe from defensive into a mid-directional attacking team through some risky disruption of previous interactions in the best defensive department in 1Lig so far. Other pre-assessments were about the serious, but low-efficiency attacking potential for Gençlerbirliği so far (and also their mediocre defensive department so far which has cost them a lot of points). That's why my Model triggered high risk to attack the big value against the Asian Handicap of (0.75), while the agencies proposed as main lines the Asian Handicaps (0.0) and (0.25).
Self-criticism here: If Gençlerbirligi had scored the penalty, and if their player hadn't been sent off (at 0:0), and if their goal hadn't been disallowed by VAR at 0:1, then probably the Göztepe team wouldn't score three goals and probably my handicap would be lost. Only the right pre-estimation assigned to the home attacking and defensive dept (and also the late perfect game of the visitors) saved the handicap...
Extremely high-risk position in South America.
Atlético Mineiro vs Grêmio.
The Model triggered a risky possible attack about a BTTS market position. Why?
- Recently Suárez scored three vs Botafogo away;
- Perfect Suárez's goal-scoring form as a whole;
- The observations pinpointed an extremely high dependence of Grêmio's attacking department on Suárez's moves, let's say estimated as XXY points.
- However, Marcelo Bielsa put Suárez on the bench vs Argentina away;
- Bielsa gave Suárez only 20 playing minutes vs Bolivia and the level wasn't as usual...
- Overall, Scolari didn't overcome most of Sampaoli's level at Atlético (say YZ points);
- And many more vectors, of course.
All of this helped me build up the likelihood of Atlético not conceding a goal against Grêmio's atomic bombardiers, trying to predict exactly how Scolari would disrupt the interplay between Suárez and Co. I'm pretty sure the experienced Scolari had studied carefully how Suárez & Grêmio turned Botafogo around - from 3:1 to 3:4 within minutes.
Self-criticism: Building a forecast against the best player in Brasileirão Série A 2023 Luis Suárez is more than madness, indeed.
I personally enjoy watching one of the last triads of old-school romantic soccer: Uruguay, Bielsa and Luis Suárez. I also remember the great Uruguayan style under Óscar Tabárez...
Also, when I have enough free time, I'm about to start modelling some working hypotheses about Brasil under Carlo Ancelotti's future governance...
-
