Thursday, April 11, 2013

Bitcoin -- aftermath

At this writing, Bitcoin's price appears to have stabilized to around $170, but realistically who knows what will happen next?  I spent entirely too much time today watching the action, or at least what of it I could watch., which gathers market data from the various exchanges, was down for much of the day, apparently due to overload.  MtGox, the major exchange for Bitcoin, was running slow all day and printing all over the place.  $120 one minute, $180 the next.  It was hard to tell what was going on, which must have been extremely frustrating if you wanted to buy or sell (Disclosure: I have no position in Bitcoin, nor do I plan to take one).

The price chart, which is like none I've ever seen, sheds a little light on what was going on.  Up to about noon (this and all times UTC) on the 10th, BTC is continuing its steady climb.  It then starts to break back downwards.  This looks like a normal correction for a while.  Dropping back from 260 to 240 or 220 after a tremendous run-up doesn't seem unreasonable.  When the price drops below 210, though, things start to go haywire.

BTC starts to oscillate rapidly between a steadily dropping floor price and a series of higher levels, first 210, then 200, 180 and eventually 150.  One possibility is that someone is trying to support it by putting out a bid of, say, 200, but when that order is filled, the next highest bid is considerably lower (and dropping) and it fills before the next bid of 200 goes out.  After a while of this, a bid of 200 starts to look too expensive to maintain, so the bid goes down to 180, etc.

One thing to keep in mind is that volume here is quite thin.  Even when MtGox got back on its feet, trades were on the order of a dozen or so Bitcoin, and sometimes as small as a single Bitcoin (or smaller; Bitcoins can be subdivided very finely).  It's not out of the question that some of the higher-level trades were someone putting out a high asking price from one account and then filling it from a second account whenever their sell order was the only one showing.  I'm not claiming this is what happened, but such things can happen in thin markets.

Overall volume for the 10th was around 200,000, or about one Bitcoin changing hands every second.  This volume is about 2% of the total pool of 10 million Bitcoin currently in circulation (but bear in mind the same Bitcoin can change hands multiple times in a day).  This looks to be about three times average volume.  Turning over on the order of a percent on an average day is not unusual.  It's the total float that's smallish.  If Bitcoin were a stock, it would be considered small-cap (it was briefly into mid-cap territory at today's peak).

Back to the chart: Between about 6 and 8 pm, it would be fair to say that the price of Bitcoin was not well defined.  I couldn't bring up the order book on MtGox, but when there's that much fluctuation in a price, with the trade price jumping multiple percentage points from trade to trade, the bid/ask spread is effectively quite large.  Large spreads mean that a market maker should be able to make good money by buying at the low end and selling at the high end.  Again, it's not clear that anyone actually was doing that, but the opportunity was there on paper.

What is clear is that if you were trying to trade Bitcoin during that interval, there was a sizable risk that you would end up with a price far off of what you were aiming for.  This is a classic hallmark of a market crash, which, in any reasonable analysis, is what we had today.  Not a healthy correction.  A full-on crash.  There's not really anything else you can legitimately call a 50%+ price drop in a few hours with a huge spread for most of the duration.

After about 8 the spread narrowed down to more normal levels.  When the price crossed back above 150, it settled into a longer-scale damped oscillation (which appears actually to have started around the time the price dropped below 210 and the short-scale fluctuations began to grow).  Again, this is not something that one usually sees in heavily traded instruments, but it seems to have died down now.

I should mention that MtGox claims that its market has been manipulated by DDOS attacks, in particular that hackers have been swamping the exchange so that legitimate traders can't get through, causing said traders to panic and sell at low prices.  The hackers then back off, goes the theory, to let the price recover so they can sell high.

This seems a little implausible to me.   Even if it's true, it's not particularly reassuring.  I would not rest easily if I knew that a small group of hackers could cause my bank account to lose, say, 10% of its value whenever they liked.  It seems more likely, though, that under circumstances like today's the exchange will get much more web traffic in general, since people want to see what's going on, and at the same time the spread will widen because the price is moving quickly and the market is small and not particularly liquid.  But I don't have their server logs.  It's quite possible that Bitcoin is trading erratically for perfectly good financial reasons, and MtGox is getting DDOSed.

[Update: MtGox is now saying that, while they're nearly always under DDOS attack, today's problems were the result of there being more legitimate traders than the system can handle.  That seems more plausible.  It's not clear how many trades MtGox can handle per second.  As far as I can tell it's not a large number, but I can imagine any number of reasons the system might melt down anyway.]

[Another note: It's also possible that some of the funny-looking price action is due to lag in MtGox's servers.  If the price is dropping steadily but the ticker is printing old trades interleaved with more recent ones that might explain some of the small-scale jitter in the chart.  I can't quite come up with a scenario of that sort that would explain the particular pattern on the chart, but I find the general idea that the technical characteristics of the servers can have their own effects on the reported price action pretty plausible.  MtGox seems to be saying basically that.]

[And finally (I hope): Watching the order book on Bitfloor, it's pretty clear what was happening.  Someone (or ones) was sitting on the ask at (say) 120, either trying to support the price, or just trying to get out at the best price they could.  Meanwhile, the bid kept dropping.  Most of the time, someone would go ahead and sell at the low bid price.  Every so often, someone would go ahead and take the higher price, either deliberately or by mistake.  When that dried up, the ask moved down.  That looks adequate to explain the short-term fluctuations without appealing to technical glitches.  The sinusoidal movement towards 160 or so is a different matter.]

So far, I've been describing price action as for any random stock or commodity, but let's consider what this means for Bitcoin as a currency.  Suppose I were to offer you a Bitcoin certificate of deposit.  You give me 100 Bitcoin today, and 90 days from now I give you back 101 Bitcoin.  That's just over 4% annual return, compounded, not too bad these days.  Personally, I wouldn't touch such a thing with a 10-foot pole.  I have no idea what 100 Bitcoin will be worth 90 days from now.  Maybe $100, maybe $100,000.  Who knows?

That's a problem for a currency.  Stability is good for a currency.  Falling by half in short order is not, obviously, but neither is doubling in short order.  Why should I pay for anything in Bitcoin if everyone believes that Bitcoin is going to be worth twice as much in a week?  Bitcoin's price was a problem on the way up, not just on the way down.  Deflation -- which is built in to Bitcoin's fundamental model -- is every bit as troublesome as inflation, and in some ways more so.

Neither inflation nor deflation, though, is nearly as big a problem as volatility.  When the British pound dropped from 2.8 Deutschmarks to 2.4, about a 15% drop, the Conservative party's poll numbers plummeted and did not recover for over a decade.  Though there were other factors, of course, "Black Wednesday", when the government's efforts to prop up the pound failed, is widely considered to be one of the major ones.

By contrast, Bitcoin has settled down to the point where it has been bouncing back and forth in a window of 150-180, about 16%, and still a third or more off its high, while I've been writing this post.

Bitcoin may yet become a stable currency.  The boom and bust of the Dutch tulip mania wasn't the end of tulip farming as a business.   Today's gyrations may not be the end of Bitcoin as a unit of exchange, but if Bitcoin wants to become a serious currency, as opposed to a means of speculation, the recent run-up and general volatility are not how it will happen.

No comments: