[Braca93]

Hey everyone, a few years back i used to do some arbitrage betting, but as things got tougher i decided to let it go.
Recently I stumbled upon value betting, and decided to give it a try. Starting with the 24th of January, I placed a total of 1063 valuebets until today.

I used a flat stake of 3€ per bet. Odds up to 3 (for first few hundred bets I was placing bets up to odds of 5). Value is calculated based on Pinnacle juice-free closing odds.
Result:+2,1€
EV (at the time of bet): 196,42€
EV (vs closing price): 163€ (idk how to add this to the graph, excel too hard)
Average valuebet %: 5,24% (6,34% at the time of bet)
Average odds: 2,19
Sports: Football, basketball, hockey, tennis
5 different bookmakers

So, that puts me at ~54 units below expected value. I understand variance is pretty big in value betting, but this seems extremely unlikely.
Anyway, just posting this here hoping for some feedback from more experienced guys, is this just a regular downswing or does this never happen?
Anyone else having a rough period last 2 weeks?

If anyone is interested in the whole excel file, send me a PM (or I can attach it here if it's not against the rules?).

[arbusers]

Before anything else, thank you for sharing your experience and your stats. Much appreciated.
To get a better picture of your project, I would like to know the following:
1. Names of the 5 bookmakers
2. Markets that you used
3. What was the reason of the flat staking?
4. How did you scan Pinnacle's odds?

First thing that comes to my mind, is that you could implement a couple of corrections. For example, change the staking plan according to the odds, or quit from certain markets while giving more rarity to others.
There is something that worries me in your project. I watching several value betting projects and I m noticing an excellent period from 15th of January until 5th of February. Some of these projects are based on Pinnacle lines comparison strictly, in the same way you also operated.
Thank you.

[Braca93]

Hey Arbusers, thanks for the reply.

1. Bookmakers I'm using: WilliamHill, Unibet, Betfair sportsbook, Coral, Partypoker (Bwin clone, sadly got limited to 1€ per bet yesterday).
2. As I said, I bet only on football, basketball, hockey and a bit of tennis. No specific markets, I bet on pretty much everything, but I guess majority of my bets are on more "exotic" leagues, for example in football there's a lot of Argentina, Brazil, Mexico, Australia etc. I'd say for football, most of my bets are Draw no Bet, O/U and asian handicaps. Basketball mostly handicaps and O/U at halftime/1st quarter. Hockey mostly O/U.
3. Since I'm using bookmakers where I'm already limited, I thought it would be easiest if I just use flat staking because 3€ stakes usually go through. If I used Kelly or some other model I might not be able to place the desired amount.
4. I'm using a software to find valuebets, but this site is not affiliated with them so I'm not sure if it's okay to say which one. They compare odds vs Pinnacle's current juice-free odds so that's how they calculate value. After the game has started, you can see how much value was there vs Pinnacle's closing juice-free line. I manually checked some games on Oddsportal to see if the software was correctly displaying the closing price, and they were all correct.

I'm aware that other staking plans might be slightly better, but due to already using limited accounts, I'll stick to flat sizing for the testing period.
Don't you think it would be a bit too early to draw conclusions for specific markets, since the sample size per market would be very small?
Also yesterday had a pretty good day, up close to 10 units, we'll see how it goes further.

[arbusers]

Thanks for the info that you provide.

1. I was wondering how your graph would look like with the addition of bet365. You know that bet365 is limiting in 2 phases so as long as you are in phase 1 with 1% limits then you are still good to go.
2. The leagues that you mentioned are probably with higher margins compared to the non exotic leagues. Why dont you try staying with the big leagues/markets and only close to kickoffs so odds are getting tighter?
3. Yes, you are correct. Now you have a really big sample of bets and you could play with the numbers in paper and see if there is any possible adjustment to be made.
4. If it is not in arbusers...

Maybe it worths using a Value Betting bot that will do the maths for you with out personal bias. 1000 bets are substantial.

[Shawn2k]

Hi Braca93,

Personally I would say 1000 bets is not enough to expect EV = Result. Yes, it's reasonable to expect being up after 1000 bets with +5% compared to Pinnacle juicefree closing line, but variance can still have a huge impact. 5% +\- from EV is probably normal. The good thing is that it could turn around at any point. Theoretically it's the same chance for running 5% above EV, as 5% under.

I would have started to look at different odds rangers, different leagues and sport - See if you find any pattern where you win and\or not. This could also be misleading, since you may have few bets in the different ranges, but it's worth following.

The best thing is that you will learn more and more if you continue your work. Theoretical understanding + Experience are a golden combination here as everywhere else 

[Braca93]

Thanks for the replies guys, really appreciate it.

Thanks for the info that you provide.

1. I was wondering how your graph would look like with the addition of bet365. You know that bet365 is limiting in 2 phases so as long as you are in phase 1 with 1% limits then you are still good to go.
2. The leagues that you mentioned are probably with higher margins compared to the non exotic leagues. Why dont you try staying with the big leagues/markets and only close to kickoffs so odds are getting tighter?
3. Yes, you are correct. Now you have a really big sample of bets and you could play with the numbers in paper and see if there is any possible adjustment to be made.
4. If it is not in arbusers...

Maybe it worths using a Value Betting bot that will do the maths for you with out personal bias. 1000 bets are substantial.

The software I'm using is having some issues getting bet365 odds currently, hopefully it gets fixed soon so I can add bet365 into the mix too.
I don't see many valuebets in big leagues, especially in soccer. I guess I prefer quantity over quality   Most of my bets are placed 1-2hrs before kickoff, but I take high % valuebets which start in <24hrs too.
Can you elaborate more about the value betting bot?

Hi Braca93,

Personally I would say 1000 bets is not enough to expect EV = Result. Yes, it's reasonable to expect being up after 1000 bets with +5% compared to Pinnacle juicefree closing line, but variance can still have a huge impact. 5% +\- from EV is probably normal. The good thing is that it could turn around at any point. Theoretically it's the same chance for running 5% above EV, as 5% under.

I would have started to look at different odds rangers, different leagues and sport - See if you find any pattern where you win and\or not. This could also be misleading, since you may have few bets in the different ranges, but it's worth following.

The best thing is that you will learn more and more if you continue your work. Theoretical understanding + Experience are a golden combination here as everywhere else 

I come from the poker world, so I understand that variance plays a massive role in valuebetting. That's why I'm trying to take as many bets as possible.
I'll definitely start looking into different sports/leagues and report back with my finds.

[Shawn2k]

Bet365 move their lines quicker in the big leagues, but you will find valuebets there.

I'm also from poker. As you probably know a breakeven (and negative) stretch for 100K hands happends even for very good (i.e +5bb\100 hands) players... Patience is extremely important.

[barbero]

I'm curious, and sorry to go a bit off-topic, but what made you guys move from poker to betting? Is poker too hard nowadays?

[Braca93]

Bet365 move their lines quicker in the big leagues, but you will find valuebets there.

I'm also from poker. As you probably know a breakeven (and negative) stretch for 100K hands happends even for very good (i.e +5bb\100 hands) players... Patience is extremely important.

Currently I can't take valuebets on Bet365 because software I'm using is having issues getting the odds from them. I'm willing to try other valuebet softwares, anything you would recommend?
Agree with the poker stuff, same as here I guess, just gotta put in the volume.

I'm curious, and sorry to go a bit off-topic, but what made you guys move from poker to betting? Is poker too hard nowadays?

Everything is getting harder nowadays, there's so much free information on the internet, everyone learns fast. I was never a poker professional tho, still play it from time to time.


WARNING: Long post ahead


So after getting a few hundred more bets in, I tried to analyze my valuebetting a bit. I found some really really interesting stuff. Overall results first:

https://imgur.com/a/WBrpIGC

Flat stakes, 3€
Number of bets: 1286
Profit: +8,63€
EV (at the time of bet): 243,08€
EV (vs closing price): 202,17€
Average valuebet %: 5,49% (6,46% at the time of bet)
Average odds: 2,16

EV goes up, profit goes sideways. Over 60 units below expected value.

So after this I checked a lot of things. Won't post all the graphs, only the most interesting ones.

Results by bookies (only approximates, since I was only checking the graphs):

Bet365: +57€, 142 bets. I only placed bets with Bet365 at the start, because as I already said, I stopped getting valuebets for them quickly because of some bug in software. Hope it gets fixed soon.
Betfair sportsbook: -12€, 280 bets
Coral: -20€, 115 bets
Partypoker (BWIN clone): -5€, 149 bets
Unibet: +60€, 371 bets
WilliamHill: -70€, 229 bets

Still fairly small samples, the only thing that's interesting is -70€ at WilliamHill, EV is at ~50€ so that's a 40 unit difference between result and EV. Anyone else losing at WilliamHill? That seems a lot for 1 bookie but lets just say it's variance for now, I find it hard to believe that WilliamHill is sharper than Pinnacle.

Results by sports;

Basketball: -5€, 472 bets
Hockey: 0, 262 bets
Soccer: +60€, 412 bets
Tennis: -45€, 133 bets

Not much to see from here, but after I filtered them by different odds and different valuebet %, there were some obvious patterns. More on that in a bit, now lets see some graphs.


Overall graph sorted by odds from smallest to largest:

https://imgur.com/aCyouCk

The point where it stops following EV is around odds of 1.91

<1.91 odds

https://imgur.com/ujy1kAd

>=1.91 odds

https://imgur.com/a/fWzwAyf

Pretty obvious, right? But it doesn't make sense, a value bet is a value bet, no matter the odds. Someone else posted his excel sheet with over 2000 bets ( he is also breakeven ) and I looked a bit into it, it's completely different. He does better as odds go higher. We're not using the same software though, maybe there is something to it? Is it still only variance?

Then I started filtering by valuebet %. Sorted by valuebet % from higher to lower:

https://imgur.com/GJVg79P

And again, there's a point where result line starts tanking. It happens at around 5%.
I ended up using 5,4% valuebets as my cut-off point as results were a bit more clear than just rounding it at 5%. Probably doesn't matter too much though.

Only 5,4%+ valuebets:

https://imgur.com/fNuvkIc


0-5,4% valuebets:

https://imgur.com/wjYiNlt

<1.91 odds, >5,4% valuebets

https://imgur.com/ROSKv4N

>=1.91 odds, >5,4% valuebets

https://imgur.com/FioNOi6

<1.91 odds, <=5,4% valuebets

https://imgur.com/sZtfFAu

>=1.91 odds, <5,4% valuebets

https://imgur.com/pO0LLVQ

Out of those valuebets below 5,4%, average valuebet is 4,3%. I still consider that pretty high, yet I am getting absolutely smashed by them. I can't explain that one.

So after this I went back to check the results for each sport/bookie, and it was mostly the same thing happening. Above 5,4% - winning, below - losing. Won't post graphs since this post is already way too long. If anyone actually read through all of that, congrats.

From now on, I'm only gonna bet on valuebets which are higher than 5,4% and report back after I get a decent sample. Still gonna take all odds up to 3. That means less bets, so to compensate I'll add a few more bookmakers, any suggestions which one?


TLDR: Valuebets above 5,4% are winning a lot, valuebets below 5,4% are losing a lot. Lower odds seem to work better than higher odds.

Thanks for reading, feel free to ask any questions.

EDIT: Seems like I'm too dumb to directly post pictures, so just links for now.

[tomkruiz]

Thanks, this is very interesting.
I suspect your result has less to do with the value bet percentage and more to do with the inaccurate calculation of value bets by the software you use.
I think they don't calculate the reference probability correctly and therefore all value bets percentages are incorrect.

[arbusers]

Thanks for the detailed info Braca93, keep us posted.
You might need to try something with Bet365 as it is a pity to park it aside.

[Braca93]

Thanks, this is very interesting.
I suspect your result has less to do with the value bet percentage and more to do with the inaccurate calculation of value bets by the software you use.
I think they don't calculate the reference probability correctly and therefore all value bets percentages are incorrect.

Yes that would be the logical conclusion, but value is calculated based on juice-free pinnacle closing odds. To make sure they aren't displaying value wrongly, I manually compared their displayed pinnacle closing odds with oddsportal. On the majority of bets I got the same value %, there were a few slight differences because odds on oddsportal were off by 0,01-0,02 on some matches. Also on some matches value % was different by 0,1%-0,5% but I assume that's due to how the closing odds are converted to juice-free odds. All in all, value % is correct. Now the only thing that comes to my mind is that pinnacle is maybe not as sharp on these lower-tier leagues...

Thanks for the detailed info Braca93, keep us posted.
You might need to try something with Bet365 as it is a pity to park it aside.

I'm thinking of trying another valuebetting service, any recommendations?

[arbusers]

Currently we can recommend the O4S and WGFP projects that were fostered under our ABVI. We tried them both with our own hard earned money.
We are working on more projects as we speak.

[jonibiza19]

1000 bets is not that much considering you are using a lot of bookmakers.in terms of opportunities bet365 is the best followed by betfair,W.HILL,Unibet .

[Shawn2k]

I'm curious, and sorry to go a bit off-topic, but what made you guys move from poker to betting? Is poker too hard nowadays?

Pokercompanies have become a lot more greedy over the years. That's the short version. Also as Braca93 said, the game has evolved dramatically the last 10 years. 10x harder now than before.