Bitcoin Forum

Hi everyone,

I do frequent cryptocurrency exchange reviews, and as I got to Bitfinex (BFX) my attention was immediately drawn towards their liquidity swaps. They reveal a substantial amount credit risk when using Bitfinex. In fact, the market derived probability of default for the exchange over a one year period can be determined at somewhere between 39.59% and 52.79%. This depends on whether you assume to lose 75% or 100% should the exchange actually default, but in any case a higher recovery rate would not be reasonable given ISDA standards.

For full details how I get to these numbers, check out: http://digiconomist.net/caution_advised_when_using_bitfinex/

To put it differently, the chances of losing at least 75% of your money on Bitfinex is well over 290 times larger than the chance that you will lose as much by simply holding Bitcoins in cold storage. I based that on the current 1 day volatility of Bitcoin, included in the details. Obviously, a default probability of at least 40% does not need comparisons to show that it is pretty bad.

These numbers would only be incorrect if market prices aren't "right." The problem is, markets are always right, unless there is some manipulation involved. If that would be the case, then there would be even more reason to avoid BFX. After all, it wouldn't be more than a scam-platform. I have no evidence that this (manipulation) is happening, just pointing out that it wouldn't improve the conclusion: putting money on BFX should be considered a very high-risk activity. Especially liquidity providers should be aware that by providing liquidity at BFX they are running a bigger risk than by buying into Bitcoin.

Good article with a lot of good info in it. However, I think that the rate the market is willing to lend/borrow funds at is somewhat supply/demand driven (due to market inefficiency??) and not purely risk to default driven. But I might wrong on this one and I'm still thinking about the article. I need sleep :-)

I'm sorry...I can't let this one sit here. Not because I think he is slandering a great institution, which he is, but rather that he is just so wrong on so many points. I appreciate people's ideas, but any idea should stand up to further scrutiny, this one clearly doesn't.

Here is one interesting thing which he failed to notice. IF the return is correlated to solely the risk of default of the exchange, then why would different currencies have different rates?

First off, the efficient market hypothesis was just that, a hypothesis. It has widely been viewed as a flawed model, and not actually practical in the real world. The whole purpose of markets is to discover what something is worth, if we already knew, we would never trade. But let's look at the numbers...

You pointed out that the USD return over 30 days was 2.79%. So, let us look at the BTC market. It's daily rate (according to Bitfinex.com) is 0.0055, and that works out (after their 15% fee) to .14% per 30 days.

So, as you astutely pointed out.
"After all, if the exchange suddenly defaults or disappears, then all money present on the platform might be gone as well. Since there is no counterparty risk on the traders due to the exchange’s system, the full credit spread is caused by counterparty risk on the exchange itself."

So, an exchange with dollars in a bank account has a high chance of disappearing, and yet an exchange with bitcoin in its account, has a very low risk of disappearing. This seems to be pretty weird, given that they are indeed the same institution. If anything, one would imagine that bitcoin losses would be much harder to recover, but we could ask the Mt Gox customers to clarify.

So, two assets, lent by the same business, with widely different rates. It is almost as if the entirety of the rate is not explained by one cause. Perhaps there are other factors at play. Let's explore a bit.

Why would the USD rate be orders of magnitude higher than the BTC rate? Econ 101 says that prices are set by supply and demand, and based on the size of the market, it seems to fit. The USD market is much larger than the BTC market (30 million vs 2.5 million, roughly). Seems like a LOT of demand for USD, and what is that used for? To buy bitcoin. So it would seem that there are a lot of people who are bullish bitcoin, and very few people who are bearish. Perhaps that explains the big difference between the rates?

According to YOUR hypothesis, that rate=risk, if the risk carried by the lenders is ONLY a function of counterparty risk, the risk should be equal for BTC and USD. It is not, therefore I think your hypothesis is flawed.

I think that the market actually functions somewhat as follows. If you think that the price of bitcoin is going to be higher than $594, it makes a lot of sense to take out this loan. If you do not think the price will be higher than that over the next 30 days, it makes a lot of sense to offer this loan. Judging by the size of the markets, it seems a lot of people think the price will be higher than that in 30 days.

Lastly, IF there is a correlation between the risk undertaken by the lender, and the rate that they receive. It would, IMO, be relatively easy to tease that data out by looking at the interest rate on the lowest priced asset, the one in the least demand, and then figure that must be closest to the actual number. People lent that asset at that rate, and since any counterparty risk should be equal over all assets for one exchange, that rate would be, if you agree with his hypothesis, the one closest to the actual compensation for risk.

There is probably some inefficiency, Bitcoin prices are not equal on all exchanges either. But at the same time, Bitcoin prices are also not extremely far off. That happened at Mt.Gox when it was no longer possible to retrieve USD from the exchange. In the case of Bitfinex money is not facing such a barrier, so it would still not be too hard to take advantage of extremely inflated rates. And even if you would cut probabilities in half, you would still be left with some significant numbers.

The numbers might sound more intuitive if you put them in perspective. Around 25-45% is normally associated with a C credit rating (depending on the agency and exact rating). C reflects substantial risks and should be considered extremely speculative. Such a rating wouldn't be odd for a Bitcoin-startup registered on the BVI, not releasing any frequent financial reports and not audited by a professional and independent agency (or regulator).

Absolutely.
Previously discussed here:

https://bitcointalk.org/index.php?topic=667105.0
https://bitcointalk.org/index.php?topic=713722.0

Perfect price requires two things - perfect information and infinite demand & supply.  

There's no perfect information here - how can an average lender compute the chances of default? He doesn't know how bitfinex's finances look like. He doesn't know how the future btc prices look like.  
There's no statistical model to make. Just a mostly irrational hunch.  

The second. Even given perfect information, the price is right only if both the supply and demand side is infinite. In real world mispricings are everywhere, and they only disappear once someone actually takes 'impossible' profit.  

The whole thing about probabilities tells us that no price is perfect, because an event either happens or it doesn't. On a perfect market with perfect information the probability of default today would be exactly 0%, until the actual default day, at which point it would be 100%. If you have other odds it means you don't have perfect information.  

Barring theft, serious incompetence or similar circumstances, the only possible situation in which lenders can lose money is serious, fast price drop where there's not enough demand to close the trades without lenders' loss.  
In this situation, bitfinex can and should just give the lenders bitcoin it can't sell. If the price doesn't rise afterwards, they would lose dollars, but their situation would be either equal to holding bitcoins or better (if bitfinex is able to return part in dollars).  
Stopping trading in this situation, which is what happened already, is exactly that - bitfinex was waiting for enough liquidity.  
Note also that bitfinex can, in a critical situation, sell bitcoins from margin called trades on other exchanges (or just wait for arbitrageurs). So what really matters is the entire bitcoin's potential demand, not just bitfinex's orderbook.  
Thus, lending is less risky than holding bitcoins.  

The reverse situation for btc swaps is theoretically possible but I don't think it's a remotely realistic scenario.

thanks.

Okay, so on comparing BTC to USD this is not just different in terms of exposure, but also the BTC swap market on BFX is completely illiquid. The demand for BTC swaps is literally zero, and you only point out the best offer rate. The worst offer rate is 185 times higher (1%). The USD market is a lot more liquid, and the worst offer is no more than 1.07 times the best offer. This is liquid markets versus no market.

The rest of your story seems to focus on the efficient market hypothesis. Along with:


I'm happy to find that the EMH is well known, but do note that these days it is not about perfect prices really. Maybe in the most strict form, but otherwise it allows for deviations around the true value. And free lunch still does not exist in economics - no return without risk and definitely not for long. For my conclusion to be awfully wrong, you are going to need to argue that there is a significant structural excess return on interest rates. Could markets be terrible inefficient when even Bitcoin hedge funds are entering the market (GABI), and there are no restrictions on money? There have definitely been some inefficiencies when looking at:

http://www.bfxdata.com/swaphistory/usd.php

But in support of my story, there are also clear pressures on the supply side. An argument such as demand is high so prices must be high will at best only apply for a short period of time. If you look at the graph, you can observe this easily. You can see some peaks in volume where rates also rise quickly, but as soon as they stabilize rates revert back. That's exactly what you would expect, because there is no reason to assume credit risk would suddenly change. The additional demand results in an excess return for the lenders, but when demand stabilizes lenders will keep taking advantage of this until the excess is gone. At that point, only the real risk compensation remains. Perhaps not exact, but not 20% off.


Interesting, thanks!

Current interest rates are the proof that your theory is completely wrong. Unless dropping bitcoin prices drastically decrease default risk, which is absurd, because the reverse is true (margin call cascade risk). 

That's probably actually true to some extend. Less valuable Bitcoins are less attractive to steal. But anyway, dropping rates typically do not cause less default risk. But models are typically not perfect either. But is it wrong if it is not completely accurate? Every economist alive will guarantee you that these rates are not without (significant) risks of losing it all or a large part of it. The derived risk was high and it still is, so this hasn't changed much at all.

pure FUD - you ever used bitfinex yourself?!

I would not let most of the current exchanges handle my money. Not just because of these numbers, but also because their compliance standards are low to begin with. If you are lucky you find a public proof of reserves audit, often by performed by non-professionals and leaving any governance unaudited. Making things even less transparent, Bitfinex for example is apparently registered on tax haven the BVI, the best way to operate under a cloak of secrecy - no wonder perceived credit risk is high! And then people wonder why Bitcoin is taking so long to go mainstream.

It's completely wrong on two levels. The first is that very high rates in the past were not due to lenders' risk estimate but because of insufficient supply (the interesting thing is as bitcoin price rises swap supply shrinks because some lenders buy bitcoin, possibly with margin). This is consistent with the fact that rates are dropping for a long time now.  
The second is because the information to evaluate risk isn't available to majority of lenders. Thus the risk as perceived by 'the market' is meaningless. It's obviously riskier than government bonds, sure, but that only sets a very low floor.  

There's also another thing, some people keep money on exchanges to be able to buy bitcoin fast after some event, bank transfers can take even a week from some countries. For these people the rate is not related to risk at all (which they already accept), but to perceived opportunity cost of having to wait (at worst) 2 days to buy bitcoin.  

It's one thing to be an exchange, but they're also offering anonymous interest-bearing short-term deposits. It would require a lot of very costly red tape (like a banking license...) in most countries, probably turn out to be impossible in many.  

Of course there is information available. It's not the first Bitcoin exchange, not the first BVI registered company and also not the first one that does not publish financial statements. There is more than sufficient general (macroeconomic) information on all of this. Also the firm specific information isn't a complete void, just think of system failures, reputation, flash crashes on the platform, etc. There are a ton of variables that relate to credit risk. And of course there are inefficiencies, but this does not make the rate 60-80 times higher than the risk free rate (1 month).

It is just the laughable 'market knows best' notion - now that the rate comes down to below that of most credit card and a lot lower than 8 weeks ago, does that means BFX default risk is now A LOT lower?  Everyone knows there's risk, risk of hack, platform crashing, flashcrash, btc crashing, banks account getting frozen or them suddenly wanting to be crooks or having been cooking the books all along, etc.

but what about the return?  In honest opinion, I find the current 14% to be low relative to the risk involve - esp given there are other non bitcoin investments out there where one may argue gives 14% but has a lower / more quantifiable risk profile.

But what about when the rate was 150% or higher?  Does it justify putting in a few bucks?

This article conflates:

(1)  What rate traders are willing to pay to leverage or short BTC

(2)  The chances of catastrophic exchange bankruptcy

(3)  All relative to an artificially low and manipulated US treasury rate (not free market since 2008)


It's like saying because the Stock Loan Rate for many NYSE stocks is 40% or more...
That the NYSE has a 60-80% chance of going bankrupt relative to the 30 day Treasury rate...
Good for an "F" on your Economics 101 mid-term.

When Bitfinex refers to risk...
They are referring to counterparty risk between traders that borrow and traders that lend...
Which is strictly a function of their margining alogorithms... and probably close to zero.

To analyze Bitfinex bankruptcy risk...
Once must estimate how much money they are making relative to deposits...
And understand how they make most of their money (hint: trading against order flow).

Also, one must understand the Crypto Space one is dealing with...
Comparing anything here to the TBill market is absurd...
It's like trying to analyze junior gold stocks in the Pink Sheets using bond market metrics.

Crypto is a place where securities spike 200% or 300% or 1000% in one day EVERY DAY...
So a DIVERSIFIED Crypto portfolio expects to take a near 100% loss on a position several times per year.

I think a lot of the concerns pointed out by OP is in light of MTGOX incident and incompetency of a few other similar operations.

Most defaults so far have occurred because of incompetency resulting in hacks as Xiaoxiao mentions, Mt.Gox of course being one of the best example.

What can be said on BFX competency? Well, for example, they registered on the BVI, while there are well documented reasons why such a (tax haven) registration increases (credit) risks. Think of a lack of oversight and increased incentives of risky behavior (increased profit margins) among others.

This risk reflects in the lender rates, because whatever way you look at this it only includes market risk + credit risk + some noise. To the lender, the market risk is actually near zero given the margin stop out level. In extreme cases, Bitfinex has covered losses so far. So that leaves something that looks like a corporate bond; credit risk + some noise due to the nature of the actual product. To the lender, this noise results in a risk free return, that automatically has a minimizing effect. So then you would end up with mostly credit risk, for which it makes sense for the number to reflect a high value. It's only a quantification of the obvious.

Dropping bitcoin prices will actually increase the default risk, not decrease it. The vast majority of positions are long leveraged positions on bitfinex.

I am pretty sure that they have the appropriate risk controls installed to prevent potential defaults on accounts as positions will be liquidated prior to going into negative equity.   

? That's exactly what I wrote...

I thought about shorting bitcoins and bitfinex looks like a good thing because of their trading volume. But im not so sure anymore about that.

So far i have read that BTC can be shorted on bitfinex, ICBIT, MPEX and fxopen.com. I didnt find info that its possible on btc china, though i didnt even find an official thread from btc china on BCT.

So is bitfinex the best option or a risky one?