[SadZizou]

I just started to try to explore the world of betting. I'm very familiar with concept like +EV, long term (and patience), variance, base math...

I just tried to understang concept like the Kelly Criterion and sports valuebetting. After deciding a decent amount about my broll, I would like to start trying my valuebets using 0.75 Kelly.
Here are some questions...all Tips from experts are very much apprecieted.

_What are the negative sides of valuebetting?
_What site comparison do you use?
_How many sites that makes comparison do you use?
_How many books odds do you need to start comparation?
_Is there something you really don't want to do?
_Is there something you really want do to?

Thanks to all

[SadZizou]

Ok Ok...I imagined more feedbacks than what I got, but...never mind :-)
I really think that online forums are a good way to grow, so even if my questions seemed to bit kind of stupid, I will persevere.

Above all the not answered questions I made in the previous post, I'll add some:

_I tried to start with a strategy that seemed quite simple to me, but theorically efficent: compare odds, finding averages, finding valuebets, betting a proper amount according to Kelly (I use 0,75, with some exceptions if the book limits the bets).

This strategy seemed not to work fine at all in the first 100bets. i'm very conscious this is an high variance strategy, so i'm just supposed to wait, but, since I found my bankroll crippled to 50%, I'm starting to wonder what I am missing, and if there are more complex aspects to consider that I should add to the process.

Thanks all again for the replies.

[u21c3f6]

Please expand on your definition of "valuebetting". Does this mean you are trying to beat a particular (or averaged) no-vig line?

Joe.

[wrongturn]

A detailed example on how you calculate average odds, how to identify your valuebets, and how much you bet, will be easier to get some responses. My quick dumb response is to take the opposites in the next 100 bets..

[SadZizou]

Yes, i was not familiar with no-vig line concept till 30 mins ago (now i think i am), but I think I'm excatly doing that.
After finding the average odd between different bookmakers (the more, the better), I calculate the average payout of the bookies. Then: 1/av.odd1+1/av.odd2, I'm supposed to have the average no-vig line.
This way I always find myself betting on big underdogs. To determine the amount of the bet, I use the Kelly formula. I use 0.75 Kelly to be a little bit more conservative.
If all this is ok (i sum up all my questions):

_What are the negative sides of valuebetting?
_What site comparison do you use?
_How many sites that makes comparison do you use?
_How many books odds do you need to start comparation?
_Is there something you really don't want to do?
_Is there something you really want do to?

This strategy seemed not to work fine at all in the first 100bets. i'm very conscious this is an high variance strategy, so i'm just supposed to wait, but, since I found my bankroll crippled to 50%, I'm starting to wonder what I am missing, and if there are more complex aspects to consider that I should add to the process.


Thanks very much for any advice, hope my english is good enough :-)

[xyz]

I think the problem is in "finding the average odd between different bookmakers (the more, the better)". You need to categorize books into two categories: books that are offering current market prices, and books that are slow to react to market price changes. Here is an example that will get your strategy into trouble. For simplicity, let's assume you just have three books: A, B, and C. A and B are off market, and C offers market prices. You compute the average odd using these three books. The value bets you find will most likely be the ones when A and B are off market. But your computation will tell you to bet at C for those games, since your strategy averages the odds. Your bet at C will most likely be -EV.

[wrongturn]

Do you use percentage of running balance instead of original balance for kelly? Sounds like you are using original balance. 50% decline sounds extreme.

[SadZizou]

Thank you xyz for the answer. I understand the troubles that slow books can bring. I usually try to use site that compares odds, and I never use less than 6 bookies to make comparison. In some cases I havean average between 20-25 books, so I think that even if there are some "old odds", the average odd should be very close to the real one. So I'm not actually really sure this could be the problem. By the way, how do you suggest to avoid this? Is it better to use just 4-5 fast bookies, or let's say, 20, with some of them not so reactive? In the first case, what books do you advice?

Thank you wrongturn. I use percentage of running balance. I mean...if i make 5 bets before the events start, for the second one I consider the bankroll as (Starting Broll-1st bet stake). The fifth bet will consider the Bankroll in kelly as: SB-Stake1-....-Stake5.

In this 100 bets, with an average of 14,94% success each, I won 4 times. Is this just variance? Am I supposed to keep pushing?

Thanks all again

[wrongturn]

Wow, your average bet is like +560 for American odds? I am not sure you get the edge calculation correct, because winning only 4 out of 100 is off target too much.

[SadZizou]

Average % of event: +561.81
Average Odd: +846.25


I think it's unlikely i'm making major mistakes here, I'm not new to this kind of thinking process. This is just the first time I'm applying to sports.

Have I been just so unlucky? Any other thoughts?

[tachi]

whatever tactics -money mgt,Kelly,odds comparision sites,discipline,etc- all this is a bankrupt if you don't have an edge.

If you want to build an edge on the big markets 1x2,AH,O/U -it will take many years and most probably you will never do this.And don't think you can go quickly,unless someone helps you.

It is easier to find an edge towards prop bets or very small sports (par example volleyball).

as you are an arber,the best is to research trading.

En fait,c'est mieux de travailler.

[SadZizou]

You might be right. By the way I'm not completely new with sportsbetting. It's just that I'm focusing on it full time only now.
I also know it may takes a lot of time to get an edge, I've been and I am poker pro, and I know how hard can be getting an edge.
Your post, sadly doesn't say a lot on how my strategy should improve, and there's no idea to reduce the edge the book seems to have on me.

This is what a forum, should be about, I'm just trying to understand and improve.

All I know that the golden rule in any kind of betting is making +EV plays, reducing the risk of bankrupt. I think that my strategy can achieve that, and seems there's no mistake. So why should I lose and lose?

[tachi]

I don't know how you can improve your strategy,car elle est pour la poubelle.

research the trading.Do you know what is trading?

[wrongturn]

How do you get the 14.94% of supposed success rate? I think your calculation for this number is way off target.

[SadZizou]

Of course I know, and I'm not new with scalping.
But I don't know why you should be offensive in french. Say something wise, or shut up please.

[SadZizou]

14.94% is the average of the success of my bets. It's a simple the sum of the percentage divided by the number of bets. There are some bets with 20% of success and other of 8% success.

[dialup_king]

Average % of event: +561.81
Average Odd: +846.25


I think it's unlikely i'm making major mistakes here, I'm not new to this kind of thinking process. This is just the first time I'm applying to sports.

Have I been just so unlucky? Any other thoughts?

You could be making a mistake calculating the no vig market price by going down the middle. If -270 is one side and +220 another, the no vig isn't -245/+245. There is a thread on the first page right now that explains Ganchrow's method of determining the no vig line

[SadZizou]

I checked again and I think I'm making the right math there. By the way, correct me if I'm wrong, if I know the payout of the book in a certain situation (there are site that collect payouts as they collect odds), I don't need to calculate the vig-line, which is basically the same:

no-vig=1/(payout)*100

If payout is 96,5% then no-vig=1/95.6%*100=103.63

Any mistake?

[SadZizou]

I checked out again the math about my simple value betting strategy but I still don't see where i'm wrong. And I'm quite sure I'm not.
But If I'm making correct math, I don't understand why there is no edge here, as someone says.
I mean...if a book underestimate the chance of an underdog to win, and it underestimates enough to beat the no-vig line...I think it should work.
Please I need some ideas by experts! :-)

[u21c3f6]

I am not an expert in this area but here are my thoughts.

You keep repeating that you have done nothing "wrong". I have heard this from many other players before when they try to describe what they do to make their wagering selections. Somewhere during the discussion I will always ask: "Are you profitable"? I get a lot of hemming and hawing with some statistics (usually improperly applied) with a bunch of well if this didn't happen and I really shouldn't have made these particular bets etc, etc then I would have been profitable. To me, all I heard was, they are not profitable. The problem is that they have told their mind that they are really profitable if they just "fix" those "small" inconsistencies. The problem with this is that because they think they are "profitable", they have no reason to change what they are doing and unfortunately for them, they continue in their unprofitable ways.

Now, for your example (and unfortunately I am not truly sure that I understand exactly what you are doing), here are my thoughts.

Something is clearly wrong. There is an extremely small % chance that it is variance but I doubt it. First, your spread: Average % of event: +561.81, Average Odd: +846.25 is tremendous. If this is true, you can make a fortune arbing and/or hedging without the worry of variance. I would need an actual example to be sure that I am interpreting what you have written correctly.

Beating the no-vig line. If someone is actually profitable doing this, my hat is off to you and congratulations. After many conversations, reviewing anecdotal evidence and reading some posters on this and other forums, this doesn't work, or in the least, doesn't work as well as one "expects" (as long as all wagers made are included in the sample and no wagers excluded due to "it shouldn't of have been a wager afterall").

The problem as I see it with the beating the no-vig line concept is that there are no hard probabilities. For example, if we throw one die and I offer 4-1 on any number, it is very easy to calculate that any wager is -EV. Now if I offer 5-1 on the #5, now you can calculate that you should be even or close to even in the long run wagering on number 5 and if I offer 6-1 on #6, then you have a +EV wager and should be profitable in the long run. These hard probabilities do not exist in sportsbetting.

The no-vig line only represents the probabilities over all lines or at least a large enough random sample of lines. The no-vig line really doesn't say much about one individual event. If you were allowed to randomly select a wager to beat the no-vig line, then you would be in business. The problem is that you can only make wagers that beat the no-vig line that the market gives you. This is not random and these wagers may or may not conform to the overall statistics. The only way that you can know if there is an edge doing this is to calculate it from the sample of wagers selected by this method. However, most (as apparently you are finding out) do not show a profit because the statistics for their subset of wagers do not conform to the overall no-vig statistics.

Joe.

[SadZizou]

Thank you a lot Joe for the answer. Please note that i'm saying that I'm not making a mistake pretending to having edge, complaining on bad luck. I'm pretty sure there's a mistake around, I'm just trying to figure out where is it, because I can't see it.
But I also know how variance can influence a small sample of bets like 100, so I'm trying to not consider the poor result and making the theory work. If I would have a 10K amount of bets with this little of result, it would be a clear proof to me. But 100 is just too little.

Let's make and example:

Event A - Event B

This are the odds and relatives payout collected from 7 different books:

5.00 / 1.1 / 90.2%
5.01 / 1.11 / 90.9%
7.25 / 1.06 / 92.5%
5.00 / 1.1 / 90.2%
6.00 / 1.1 / 93.7%
6.5 / 1.13 /96.3%
8.0 / 1.06 /93.6%

Averages are:

6.11 / 1.10 / 92.9%

Odds to beat are supposed:

6.57 / 1.18

In my thoughts, any odd bigger than 6.57 is a value bet on Event A. In this case if i'm betting on the last book at odds 8.00, i'm making a +EV bet.

If there's is a mistake, and probably there's, it should be here. But I can't figure it out.
can someone explain?

THX

[Thremp]

Beating the no-vig line. If someone is actually profitable doing this, my hat is off to you and congratulations. After many conversations, reviewing anecdotal evidence and reading some posters on this and other forums, this doesn't work, or in the least, doesn't work as well as one "expects" (as long as all wagers made are included in the sample and no wagers excluded due to "it shouldn't of have been a wager afterall").

This idea is completely wrong. The information I had before Pinny left the US actually had it being slightly better than expected. Going off my own guesstimates, I'm going to say that I have not just run hot for the last 3 years.

[u21c3f6]

First, where are these odds on the same event? Second, as to your question, trying to come up with the no-vig line with this spread of odds is part of the problem. You have no idea which one of these odds is closer to reality as they spread from an approx 10% chance to an approx 18% and that may not even be correct for this one individual event.

What I see if I am understanding what you wrote correctly is that you have a book where you can get +800 for Event A Yes but that there is also a book where you can get (based on your payout %) about -565 for Event A No. Is this correct?

If so, your answer IMO is to arb and/or hedge and forget about trying to find the +EV wager. If not, please explain.

Joe.

[u21c3f6]

This idea is completely wrong. The information I had before Pinny left the US actually had it being slightly better than expected. Going off my own guesstimates, I'm going to say that I have not just run hot for the last 3 years.

Those that have attempted this in reality and failed would tend to disagree with you.

After the fact statistics tend to give this method some credence because there are no real-time errors. If one can do this successfully, I am sincerely congratulating you. The problem for most is that they can't seem to do it successfully in real-time as sometimes bets that should have been made were not and other bets that shouldn't have been made were made due to changing odds right up to event time. An additional part of the problem that I see is the non-random effect of the market determining which wagers you can get that "beat" the no-vig line which may or may not conform to the stats of the no-vig line in its entirety.

I don't say it is impossible and again, I congratulate anyone that can do this successfully. I am just saying that the empirical evidence seems to suggest that it is not the Holy Grail that it tends to be portrayed as.

Joe.

[wrongturn]

Of course I know, and I'm not new with scalping.
But I don't know why you should be offensive in french. Say something wise, or shut up please.

What did I say to offense you? I simply asked you to how you calculated the supposed win rate because your result was off too much. You did it on post #21 and you should have post that earlier. I am sure you will get more constructive comments on it.

[SadZizou]

What did I say to offense you? I simply asked you to how you calculated the supposed win rate because your result was off too much. You did it on post #21 and you should have post that earlier. I am sure you will get more constructive comments on it.

Sorry, I was not talking with you but with Tachi, who wrote some offensive sentences in french, dont know why

[SadZizou]

First, where are these odds on the same event? Second, as to your question, trying to come up with the no-vig line with this spread of odds is part of the problem. You have no idea which one of these odds is closer to reality as they spread from an approx 10% chance to an approx 18% and that may not even be correct for this one individual event.

What I see if I am understanding what you wrote correctly is that you have a book where you can get +800 for Event A Yes but that there is also a book where you can get (based on your payout %) about -565 for Event A No. Is this correct?

If so, your answer IMO is to arb and/or hedge and forget about trying to find the +EV wager. If not, please explain.

Joe.

Obiouvsly these odds are from the same event.
I' ll make the conversion in US odds

+400 / -1000 / 90.2%
+401 / -909 / 90.9%
+625 / -1666 / 92.5%
+400 / -1000 / 90.2%
+500 / -1000 / 93.7%
+550 / -769 /96.3%
+700 / -1666 /93.6%

Averages:

+511 / -1000

Odds to beat:

+556 / -556

In this case +700 is the wager with +++EV. Right?

It is true that I don't know what odds are closer to reality, but my assumption is that the average odd between 25 book, for example, is quite close to reality. At least closer than just the comparison between 4 books. But may be I'm wrong here.

[u21c3f6]

In this case +700 is the wager with +++EV. Right?

Not the way I see it (and I'm still not sure I understand your posting of these odds). Those are huge spreads.

I would suggest that the line is closer to somewhere between +700 and -769
based on what you posted.

Joe.

[wrongturn]

In your example, either +700 or -769 is likely to be a good value, but I don't think you can decide which by averaging all books, especially from those low payout books because their odds are so distorted.

[SadZizou]

Ok but, let's say that +700 is my bet. Even if some kind of distortion doesn't make +700 to be a full +EV value, this odd should be at least around break even. If it is so, i think my results that caused my bankroll down to 50% after 100 bets, are completely effect of a poor run.
Or am I wrong? Because all bets i made are completely similar to the model of the example, and if +700 is a good value...a large part of my bets should have a good value.

[u21c3f6]

Ok but, let's say that +700 is my bet. Even if some kind of distortion doesn't make +700 to be a full +EV value, this odd should be at least around break even. If it is so, i think my results that caused my bankroll down to 50% after 100 bets, are completely effect of a poor run.
Or am I wrong? Because all bets i made are completely similar to the model of the example, and if +700 is a good value...a large part of my bets should have a good value.

The problem is that there is no evidence that the +700 wager is +EV. In fact, based on your results so far (even though it is only 100 wagers) suggests to me that there is a high probabilty that your wagers are actually -EV.

Joe.

[sharpcat]

It appears to me that you are betting prop bets that do not draw much action therefore this is not an efficient market meaning that you likely can not use the no-vig line as your implied win probability. In a market like this you need your own implied win probability.

[SadZizou]

The problem is that there is no evidence that the +700 wager is +EV. In fact, based on your results so far (even though it is only 100 wagers) suggests to me that there is a high probabilty that your wagers are actually -EV.

Joe.

Ok, but your making this assumption basing your consideration on the result of 100 bets, which is not the good approach, since 100 is a little sample. I have the suspect that a mistake could around, but i would like to see it in the thinking process. Sadly at the moment there's none that found something.

[SadZizou]

It appears to me that you are betting prop bets that do not draw much action therefore this is not an efficient market meaning that you likely can not use the no-vig line as your implied win probability. In a market like this you need your own implied win probability.

I don't know exactly when a market can be considered efficient. By the way, the majority of bets are made with a comparison between 20 to 30 books. Sometimes less, sometimes more.
Can an average of 25 odds be considered an efficient market?
Should I not consider the book with a lower payout, and consider just the major ones? Does this improves the evaluation of the implied win probability?

[sharpcat]

An efficient market would be a market that gets a lot of action like mlb spreads and totals. In a high volume market the market will correct itself to reflect the true win probability, in small markets that do not get a lot of action the market can not adjust itself.

I would focus on certain books which offer low juice and high limits when calculating no-vig lines.