Bitcoin Forum

By now many of you will have seen my sale of CDSes on imsaguy. I just wanted to elaborate on why doing this doesn't make me insane.

Basically, you turn into an insurance company, with all the pros and cons that come with that. The CDS itself is a little more nuanced than straight-up insurance, but the risk/return profile of the CDS seller is about the same.

Think about how much you pay to insure your car, or your house, or your health. It's probably a lot of money, right? And the companies selling you the insurance, most of the time, just sit back and enjoy the cashflow. On the other hand, they're exposed to what is typically called "tail risk" in financial lingo, meaning that they occasionally suffer big losses (when claims must be paid).

Life insurance for 70+-year-olds might be a scary business to get into, but health insurance on young healthy people carries low risk (and lower returns, obviously) but provides some nice steady cashflow for doing very little. Want to supplement your income? Sell catastrophic meteorite insurance! :)

I consider imsaguy to be one of the most trustworthy people in the community, so have no qualms taking on some of his default risk. Depending on your risk tolerance, you might want to sell CDSes on much riskier loans (much larger spreads mean much more regular income for you), but that's really up to you and your perception of the creditworthiness of the reference entity.

Anyway, my point is basically that selling CDSes can give good, steady income, at the expense of increased tail risk. Furthermore, you can actually be doing other things with the money you're liable for during the term of the CDS, which wouldn't be the case if you the loan directly. Of course, doing this exposes you to currency risk too (if you lose the coins you're liable for, and your reference entity defaults, you'll need to buy more coins to cover your debt).

From the larger community's perspective, this will allow lenders to reduce their risk by paying someone else to take it on, thus making the market a bit more mature (http://charleshughsmith.blogspot.com/2012/03/in-praise-of-hedging.html) and giving people more ways to protect their investments. It also gives us a way (whose validity is certainly arguable) to quantify the community's perceived risk on loan defaults, which might help in the longer run.

I'm definitely not advocating that anyone do this unless they fully understand the risks involved. Most of the time selling a CDS is just "free money", if you do it right, but you need to be able to cover the losses if things go down the drain. As always, this is not investment advice and all the usual no liability bullshit, but I think it could be a good deal for some people.

So, read up on http://en.wikipedia.org/wiki/Credit_default_swap and feel free to ask questions here about how they work. Surely I can't be the only one with the guts to try this around here, right? :)

tl;dr: no, definitely not. This is stuff you definitely need to understand well before undertaking the risk

I pay you 100 btc and this is all I get? I want a refund.  I'll see you in #bitcoin-court.

I'm shocked, shocked to find that gambling is going on in here!

To save a couple of seconds for anyone stumbling across this thread, the link to the CDS offer referred to by OP:
 - http://bitcointalk.org/index.php?topic=74552.0

And, just in case anyone isn't aware, security-based swaps are an activity that is regulated in certain jurisdictions, such as in the U.S.:
 - http://www.sec.gov/spotlight/dodd-frank/derivatives.shtml

Finally. So BTC are Securities.

So no taxes until they are converted.

What about Spain?  ;)

Who will buy that from me? I'm extremely likely to default on the first claim made ever ...

In other news:

I'm selling some really cheap catastrophic meteorite insurance, contact me for details!

That's not clear, but a security-based credit default swap is a security. Oh...and probably gambling too. ;D

Maybe on higher risk loans?  lol.

People keep throwing the "gambling" term around but I haven't yet seen a formal definition that separates it from what most people call investing. Everything has a tradeoff between risk and return, so is gambling just when you're not sitting on the hypothetical "efficient frontier", perhaps? Or is gambling just when there's a high risk of any sort, regardless of return?

In most cases I refer to any investment where the  probable outcome is unknown, or not knowable, as gambling.

In this particular case, I don't know imsaguy, or you, and so to invest would probably be gambling. I suspect for most people, just investing in a CDS represents many unknowns and is very likely gambling.

On the other hand, once the risks are known, and the probabilities at hand, regardless of the theoretical risk, it may be a safe bet. I don't think that's gambling.

If the level of risk makes me laugh, because everyone agrees it's really risky, but I have confidence,  everyone else will think it's gambling ;D

Since you've posted this, I've already noted possible CDSes I'd be interested in if they were offered. lol.

Huh, I would have thought the opposite.

If I gamble(bet) on an event, I know the exact payout(ROI). i.e. Roulette 22 Black pays 35 to 1 give or take.

When investing, the payout might be expected but by no means predictable. The ROI is not determined until you are no longer invested.

hi,  I'm imsaguy.  I typically troll in the #bitcoin-otc channel on the Freenode IRC network.  Great to meet you!

Insurance is typically considered negative return.  Is that gambling?

You buy a CDS in the same way you buy insurance. 

Would you sell CDS built on "First Pirate Saving" and "TyGrr-whatever" bond ?

Hmm. Good point. My previous statement is obviously incorrect for the reason you state. Perhaps it is the opposite.

Gambling = unsure bet
Investing = (probably) sure bet ;)

Roulette strikes me as pure gambling. Each round has the exact probability of each previous round, so there is no additional insight gained by the "skill" of the player. I would be more likely to place a bet on how much the gentleman in the cowboy hat w/ a scotch is going to lose tonight ;)

Blackjack, unless the deck is shuffled each hand, presents additional potential insight which may increase the player's odds beyond the expected house advantage, as well as rules of play that may increase or decrease the player's odds and are used according to the player's skill. I don't feel like I'm gambling at all while I play blackjack. I have the potential to mitigate risk by means of knowing previous hands and choosing betting strategies, and using tools like dealer blackjack insurance. I rarely do worse than breakeven playing blackjack.

I played blackjack at an online casino once. That felt like gambling. The odds appeared to be stacked against me in an exceptionally unknowable way - there was no apparent way to calculate the odds, and therefore there was no way at all to mitigate risk in the game. No blackjack skills seemed to play any part whatsoever, and the game seemed more like the roulette wheel. The only sure bet in this case was that the casino would remove all my money if I did not leave.

Slot machines? Gambling. Slot machines played at the correct rate to get free drinks? Investment ;D
The slot has the exact same probability on each pull. This (IMO) make this a gamble.
- because the probability is the same on each pull, the ROI is not knowable.
- to argue that if one plays longer, the odds are averaged and the ROI becomes knowable is a Gambler's Fallacy http://en.wikipedia.org/wiki/Gambler's_fallacy  (http://en.wikipedia.org/wiki/Gambler's_fallacy)

However,
The rate at which you'll be served drinks while playing is (basically) constant - it is predictable based on knowledge (of the casino, waitress, business practices, or any such things)
-I can now apply my skill by limiting my pull rate on the slot in order to receive free drinks, and approximate the knowable ROI. This makes this an investment.

In the blackjack example, if I take a beginning player, with only rudimentary knowledge of the game, who has not memorized the odds of each type of hand, and essentially remove the possibility of skill from the game, this now seems like gambling again.

EDIT:
+1 ;D

Here's a more concrete idea for how to make money by selling CDSes.

Say, for example, that like many people around here, you've been trying to get an account with pirateat40 for a while. You have 2000 juicy coins and you'd love to be making 7% a week on them. Unfortunately, pirateat40 either doesn't have the capacity to hold your coins or you simply can't get an account with him. But clearly, you don't think he's going to default, or you wouldn't want to put your coins with him in the first place, right?

Instead of letting your 2000 coins rot in your wallet, you can sell 2000 coins worth of CDSes to other people on the forum for rates that approach pirateat40's. You might ask for 10-20% a month for your spread, thus giving you the ballpark of returns that pirateat40 himself offers, without you actually having to deal with him at all. You'll still be as exposed to his default risk as you would be if you had lent him money directly, and you'll be making similar amounts of money off of your exposure. In an ideal world you might lend to him directly, but in effect you're becoming a "synthetic" lender by selling a CDS on him, and will face commensurate risk and returns.

As always, this is not investment advice :) George Soros thinks naked CDSes are the work of the devil. Deals like the above look very appealing and are easy to become addicted to, simply because you don't actually need to have the coin on hand to back your liability. There's nothing stopping you from claiming you can cover 20000 bitcoins worth of default and making 4000 coins a month on CDS fees. The fact that after 5 months people would have paid you more than the nominal value of the debt isn't a big deal, but would certainly exclude people who aren't compounding their loans. But as tempting as it might be, you need to keep in mind that you are actually liable for 20000 coins (about $100k at current rates) if shit hits the fan. Anyone who tries to sell that much protection should expect a good deal of scrutiny of their reputation and should expect to have to provide real identification and proof of a financial situation to support that kind of liability. Caveat emptor venditor!

Important edit: I most certainly am not advocating being liable for coins you do not have in your possession. This not only means you can lose a lot of money, but exposes you to the wildly fluctuating exchange rates of bitcoin (perhaps you can cover 20000 coins of liability at the current exchange rate, but what if they jump to $30 again and you need to pay up?). I'm not fundamentally opposed to naked CDSes in the sense that I don't care if the buyer of the CDS actually has debt with the reference entity, but I think it's scary/irresponsible to sell insurance when you don't have the cash on hand to pay for it. I realize this happens all the time, but it was also a big part of why the mortgage crisis happened. In theory it's "safe" to be liable for more than you own as long as your liabilities are uncorrelated, but the correlation is impossible to measure, so I don't think it's worth it when the stakes are so high.

The most I'd suggest doing with the coins you're liable for is to lend them out in a "risk-free" manner (i.e., so you aren't making no money off of the coins you're liable for; otherwise why not just lend them to the reference entity directly?). Thinking of it this way leads to one approach to thinking about pricing CDSes, but I'll talk about that some other time.

My point about making 4000 coins a month was to point out how tempting it might be to sell more protection than you can afford to sell (a real slippery slope!), not a suggestion on how to get rich quick. Temptation of that sort is what led many people to financial ruin a few years ago, and it's easy for anyone to fall into that pattern when they're getting what feels like "free money" every month. So, again, do not even think about doing anything of the sort without very carefully evaluating the risks involved. Read a lot about finance before you start, too :)

Will you disclose your potential liability, not only now, but at all times that you have CDSs on issue? It would be a good sanity check on counterparty risk (and would set a good precedent for others following you into the CDS-issuing business (as I am considering doing)). :)

Sure thing!

Last I checked, almost all CDS's were not "naked". Some amount of leverage is utilized, but some collateral is required to secure against a default event.

I'm sure in the bitcoin world you can "write" whatever CDS you want, and anybody can take the other side of the bet, but I sure as hell would want to know you can cover at least some kind of amount in the event of a default.

As you might already know, when an event triggers the CDS, the only amount owed is only that amount after whatever can be recovered from the reference entity (defaulter). As in, after all amounts can be recovered (say .40 BTC per BTC owed), the amount you owe (if you sold the policy) would be the remaining .60BTC/BTC. This usually occurs in a bankruptcy court or adminstrative hearing. Last I checked, no such court or arbiter exists in bitcoin, although escrow services may act in that capacity to some degree. For certain, there isn't much case law to rely upon for bitcoin disputes.

Of course, you can design any CDS you want, but that's how I've seen it done in the past. All things considered, I've been interested in this also.

Yeah, I know, but given how most defaults have happened so far in the lending community, recovery is usually unlikely, and it might simplify things to just make binary CDSes. They're less common in the "real world" due to the pervasive presence of a strong legal system, but I'd imagine they'll be the norm around here.

I also think it's important to distinguish between two meanings of naked CDSes:

• The buyer of the CDS does not own the debt in question, so he is either speculating (betting) on the default or hedging other risks that are assumed to be correlated with the reference entity's default. This is the common definition of a naked CDS, as far as I understand it.
• The seller of the CDS does not have the coins he is liable for, so at best he is gambling on the bitcoin exchange rate (to move more funds into bitcoin if he needs to pay out) and at worst he is simply incapable of ever paying his full liabilities if shit hits the fan. As I mentioned earlier, in theory if you can assume that the probabilities of your loans defaulting are completely uncorrelated, it might make sense to be liable for more money than you have, but in general that's a very risky path to go down, especially when everything in this community is so risky and most people do not have experience with assessing risk.

I consider the second meaning of naked CDS (not a standard one, as far as I know) to be a lot scarier than the first one, and would suggest actively discouraging people from doing that by requiring proof of funds. I wouldn't go as far as using an escrow for the coins, because that would kill part of the incentive to actually write CDSes in the first place (they mean you can get paid interest, like a loan, for coins you own, but you can also use the coins in the mean time; have your cake and eat it too!), but CDS writers should be able to prove that they are not only in a financial situation to cover their liabilities, but that they are also responsible and won't take undue risks with the money in question. Perhaps a monthly statement (signed, using that bitcoin client feature for proving address ownership?) showing where the coins are would be enough.

Anyway, thanks for your input. You seem to know what you're talking about, so I'd appreciate any other comments you have on these contracts. There are a number of unique features about the bitcoin credit market that make it hard to blindly apply CDS terms from elsewhere to it, but I think that if we can come up with common features that we'd like to see in them, and possibly even standardize on them, we might eventually see a real market for these things.

That's certainly an issue to keep in mind. By buying a CDS, you're not actually killing your risk as much as diversifying your exposure. Most people don't like to have a portfolio with only one security in it, because "all your eggs are in one basket". Similarly, if your entire risk exposure is from one borrower, that should make most people uncomfortable. By buying a CDS from someone (e.g., me), you're spreading your risk exposure across multiple people: if you own 1000 coins of debt from imsaguy and buy 500 coins of protection from me, you're now 50/50 exposed to me and imsaguy. Assuming imsaguy and I are uncorrelated (e.g., he isn't my brother or my partner in crime), other things equal, you're safer this way (and are paying me for the reduced risk).

Risk evaluation is a really hairy topic and there are entire departments at most financial institutions dedicated to measuring it and making sure people don't do stupid things. Things like (co)variance and correlation exist as imperfect measures of it, but their degree of imperfection can't really be known until it's too late :) Risky business!

Mind = blown