We placed 240,000 bets on BetFury - Limbo failed: 97.33% vs 99.02% RTP

[yahoo62278]

TL;DR
 
We placed 240,000 live bets on BetFury's Originals (Limbo, Dice, Plinko). Two findings:
 
1. Their "Provably Fair" implementation does not let a player verify anything about how results are generated. It fails 6 of 7 requirements of a real provably fair system; the 7th is only partially met.
2. In our tests, Limbo returned 97.33% against the 99.02% its own bet panel encodes — a house edge ~2.7x the advertised rate. Dice (99.39%) and Plinko (99.01%), measured the same way at the same time on the same account, were consistent with their advertised math.
 
Every number below is recomputable from the raw data in the repo. No dependencies, Node.js only, ~15 minutes.
 

This is interesting, just like that other thread of yours I saw. Actual math analysis. Since Betfury runs several signature campaigns on the forum, a representative should respond to this. If we give them the benefit of the doubt and assume this is a mistake—provided that your analysis is correct—they should correct their algorithm. Otherwise, they should explain why the numbers you've presented are incorrect.

I wouldn't mind seeing a representative stop in and comment on this, but I doubt we see it. Betfury may have to admit guilt if this analysis is correct and that's not good for business.

@op send a link to this thread to Betfury and see if you can have them reply. Very curious what they have to say.

[ProvablyFair.org]

Please continue sharing this kind of data, as it helps many gamblers to understand how provably fair works.

I read your previous thread on Stake.com's Original game, Crash, and it actually opened my mind to how these things work. Also, it's good that someone is checking so that these gambling sites know that not everyone is just playing, and they can't take advantage of gamblers by tweaking the house edge or the provably fair percentage. That way, they won't dare manipulate those important numbers that could really affect the winning rate of every gambler out there.

It's like a warning that someone is looking.

Thanks appreciate the kind words.

TL;DR
 
We placed 240,000 live bets on BetFury's Originals (Limbo, Dice, Plinko). Two findings:
 
1. Their "Provably Fair" implementation does not let a player verify anything about how results are generated. It fails 6 of 7 requirements of a real provably fair system; the 7th is only partially met.
2. In our tests, Limbo returned 97.33% against the 99.02% its own bet panel encodes — a house edge ~2.7x the advertised rate. Dice (99.39%) and Plinko (99.01%), measured the same way at the same time on the same account, were consistent with their advertised math.
 
Every number below is recomputable from the raw data in the repo. No dependencies, Node.js only, ~15 minutes.
 

This is interesting, just like that other thread of yours I saw. Actual math analysis. Since Betfury runs several signature campaigns on the forum, a representative should respond to this. If we give them the benefit of the doubt and assume this is a mistake—provided that your analysis is correct—they should correct their algorithm. Otherwise, they should explain why the numbers you've presented are incorrect.

I wouldn't mind seeing a representative stop in and comment on this, but I doubt we see it. Betfury may have to admit guilt if this analysis is correct and that's not good for business.

@op send a link to this thread to Betfury and see if you can have them reply. Very curious what they have to say.

I tagged them on the X post, no response so far: https://x.com/provablyfairorg/status/2079448620424663397?s=20

@twistjack if the config we measured applied to your play: 45,000 × 1.694% = about $760 in extra house take. That's expected cost, not an exact refund, your own variance sits on top, and we can only speak for our capture window. Talk to BetFury and keep your bet history.


@Cointxz the PnL is in the dataset, every row has stake and payout, and RTP = paid ÷ staked, so the deficit comes from actual money flows, not wager counts.

On sample size, the maths says we're past where more bets help. We measured at a super low variance setting so 60k bets has strong statistical power. At the 1.10× setting the normal swing on 60,000 bets is ±0.135 RTP points, we measured 1.694 low which is 12.58 standard deviations, and a second 60k capture at a different target landed in the same place. More bets won't meaningfully strengthen this further.

ProvablyFair.org outstanding work and finally someone who understands statistics. Increasing the sample size from here on does not deliver any further statistical relevance. Or put differently, the statistical relevance of your sample is already significant. I have crunched the numbers myself a while ago and you wouldn't even need 80,000 runs per game to determine whether it is off or not. I like your experiment very much, you prepared the evidence very well and it is a relief that you took the time and put in the effort to present it in a way that everybody should be able to understand.

I like that you circumvent to some degree the RTP variance (seemingly complex assumptions etc.). Just measuring expected wins against actual wins and then measuring expected fluctuation against actual fluctuation is easier to grasp for most people. But as you can see in this topic, there are still people who believe you should increase the sample size. You won't ever be able to convince some of the gamblers here.

I have run experiments like this on two other casinos and both are off. It would be great if you run more experiments, I would be happy to send a small donation for your efforts.

Great work and it is actually a project I wanted to do myself! 
Dzonikg

Thanks, no need for any donations but if you have data on a casino we are happy to review and do our own analysis too. If you want to share ofc

[DanWalker]

from trustpilot- recently many players have complained like this

You know, provably-fair's purpose is not saying "Trust us"
The whole idea of ​​provably fair is "you don't need to trust us, just verify us."

Nice statistics, OP.
However, I'm not ready to declare BetFury guilty right now. If there is something wrong with this report, they should respond point by point with reproducible evidence in same way.

Because OP challenged the variance claim in this report with statistics. So it's not ok to get away with just saying our RNG is certified or it's variance.

If they are innocent then there should be no problem publishing an independent audit of Limbo's RTP, right?

[KTChampions]

I wouldn't mind seeing a representative stop in and comment on this, but I doubt we see it. Betfury may have to admit guilt if this analysis is correct and that's not good for business.

@op send a link to this thread to Betfury and see if you can have them reply. Very curious what they have to say.

Yes, an answer is necessary, but it seems the range of the answer is such that in any case it will have some negativity, since all the options explaining the deviation are bad in image sence (I assume that the OP knows his business and the mistake is not on his side).

I saw the previous thread and thought, "Oh, how boring, the math worked as it should have." I expected the same thing to happen in this thread, but no. I will follow the OP and this topic with interest.

[crwth]

Hmm. From what I have read here, there's a missing piece, like the actual formula disclosure for Limbo. Like how the server seed or client seed hash is processed, and to see if it really does what the game should do. It's necessary IMO to see if they can replay bets and have the actual results be verified.

Once that is done, maybe there's a rounding factor with the calculations or possible wrong multipliers? Maybe in that way, with so many bets done, it's bound not to be equal or have very different odds and settlement to the player.

Let's wait for BetFury to respond about this.

[Unsoldier]

An interesting report. But I'm not familiar with casino provably fair systems, so I can't evaluate how honest this report actually is. After all, it could easily be fabricated by a competing platform. In light of this, I have a question for the OP: do you work together with regulatory bodies or the government? In theory, if you're doing all this voluntarily, no one is obligated to pay attention to your reports, since nobody knows how you do all of this.

[Odohu]

Thank you for this wonderful analysis and for taking time to actually show how to verify the much talked about provably fair games that many casinos pride in. However, I will like to know what you intend to achieve with a post like this, such as if you want to make claims from Betfury or you just want to create public awareness. Secondly, have you tried to check other casinos to confirm that they practice what they preach?

[Comeacross]

Let's wait for BetFury to respond about this.

They are likely not to respond to this. Op said he tagged them on X but they haven't responded yet. They are very active on X platform but does not bother to respond. Their representatives on this forum is not too active because their last post is about a month ago despite their profile shows they recently come online. visit BetFury.com profile

Maybe calling their attention on their Ann thread will help? Since it appears they're avoiding responsibility.

It's fine to lose playing but it's unfair for casino platforms to be dishonest with their system.

[KTChampions]

An interesting report. But I'm not familiar with casino provably fair systems, so I can't evaluate how honest this report actually is. After all, it could easily be fabricated by a competing platform. In light of this, I have a question for the OP: do you work together with regulatory bodies or the government? In theory, if you're doing all this voluntarily, no one is obligated to pay attention to your reports, since nobody knows how you do all of this.

This is a controversial statement, I would even say incorrect. The whole point of the provably fair system is that, theoretically, anyone can verify the casino's integrity, right? If someone does and the result is negative, it's in the casino's best interest to explain why. If there's no explanation, or if it's simply ignored, then this is a negative sign for outside observers.
But technically, you're right, no one is obligated to respond to something like this.

[ProvablyFair.org]

An interesting report. But I'm not familiar with casino provably fair systems, so I can't evaluate how honest this report actually is. After all, it could easily be fabricated by a competing platform. In light of this, I have a question for the OP: do you work together with regulatory bodies or the government? In theory, if you're doing all this voluntarily, no one is obligated to pay attention to your reports, since nobody knows how you do all of this.

There are two parts to the report: 1. the provably fair issue and 2. Limbo not paying the advertised RTP. You don't need to understand provably fair to understand the second one.

On fabrication: the full raw data is public on GitHub (240,000 bets in JSON, every bet with its seed, hash and outcome) plus the analysis scripts: github.com/ProvablyFair-org/betfury-investigation.

Every result carries BetFury's own published hash, and anyone can rerun the scripts in about 15 minutes, or place their own bets and see if the shortfall reproduces. So I disagree it could easily be fabricated, it took many hours of placing bets and analysing the data.

We don't work with regulatory bodies, but I do think casinos building their own provably fair games should be required to have third party audits. Full disclosure: independent audits are a service we offer, and some casinos use us to verify their games are fair. This investigation was not paid for by anyone though. BetFury is not a client and no competitor commissioned it. It came from a player report, and since all the data and scripts are public you don't have to take our word on motives anyway. We investigate player reports as part of bringing awareness to the community, some examples here: https://www.provablyfair.org/insights?category=investigations

Thank you for this wonderful analysis and for taking time to actually show how to verify the much talked about provably fair games that many casinos pride in. However, I will like to know what you intend to achieve with a post like this, such as if you want to make claims from Betfury or you just want to create public awareness. Secondly, have you tried to check other casinos to confirm that they practice what they preach?

A player reported the game to us, so we spent the time investigating it.

No claim against BetFury, we have no financial interest in this and we're not seeking anything from them. Two outcomes would be good: players understanding that not every provably fair system is built correctly, and BetFury either fixing Limbo's payout or correcting the RTP they advertise for it.

Yes, we've checked others. You can see the rest here: https://www.provablyfair.org/insights?category=investigations

We'll keep doing them. Players don't have the tools or the expertise to check these systems themselves, and someone should be.

[twistjack]

OP has betfury replied to you at all. It would be good if they could review players bets

Also I have sent you a dm on x for some other casinos I have suspicions on with details of my bet history if you cna look

Thanks for the good investigations helps us players a lot