Bitcoin Market Update 4/4

Hi Everyone,

Since my last update on 3/31, markets have been quite volatile trading between $420/BTC and $500/BTC.

While third party-payment processors in China have been disallowed from servicing bitcoin exchanges there, there has not been any official ban of bitcoin by the central bank and there have been no exchanges cut off from payment processing through their respective banking partners.  This seems to be more of a crackdown on third-party payment processors than bitcoin itself.

Given the IRS rulings on bitcoin as "property" some big winners in the bitcoin markets are undoubtably selling off part of their stake to pay out taxes owed before the filing deadline in April.

Another interesting idea I read from Vinny Lingham, CEO of Gyft, is that the bitcoin price is currently weak because Bitcoin merchant adoption is outpacing consumer adoption. When merchants cash out of the bitcoins their customers give them, this depresses the price faster than normal broad-based demand for bitcoin can buy it up.  This makes a lot of sense and could play a big part in explaining why the price has dropped and stayed depressed recently.  You can read Vinny's full article here:

I was originally planning to talk more about the more tactical aspects of orderbook dynamics today, picking up where we left off last time but I'll do that next week instead.  This week I wanted to address the new Michael Lewis book, Flash Boys: A Wall Street Revolt and the recent New York Times article about it which you can find here:

The premise is that HFT shops were "front-running" orders on the market by reading orderflow on one exchange and issuing orders to other exchanges faster than the rest of the market.  For example, Brad sends an order to buy 100k shares of ABC through his trading platform and routing service for execution on BATS (50k shares) and NYSE (50k shares).  BATS is located closer to Brad's trading desk so BATS receives his order first.  An HFT shop sees Brad's order on BATS and quickly sends an order to NYSE to buy some shares of ABC and repost them for sale at a penny or few pennies higher.  By the time Brad's order reaches the NYSE, he can no longer buy 50 shares at the price he intended to since the HFT shop already bought them and now has to pay a penny or a few pennies higher to the HFT shop to get his 50 shares.  The HFT shop has effectively taxed Brad a few pennies per share.  Rinse and repeat throughout the day and the HFT shop reaps in the profits.  Or so the story is told.

Unfortunately, the story is much more complicated than Michael Lewis suggests.  Lewis would have you believe that the markets are rigged and that the big guys are out to get the little guys.  While it's true that the "retail guy" on average loses money to the "professional", his narrative is disingenuous.  In the capital markets, it's not that one side is AGAINST any other side; it's that every side is FOR themselves (I'll come back to this distinction later once we get to the topic of microwave and laser communication networks).  It just so happens that retail flow and clunky institutional flow are the least sophisticated participants who end up paying out a premium to their faster and shrewder counterparts.  Some of this premium is simply the price for execution, the price for liquidity.  Whereas once upon a time, tick sizes were 1/8 and 1/16 of a dollar, and retail flow had to pay up to the human market makers, human prop traders, and retail brokers to express their opinion, now tick sizes are a penny, spreads are smaller and retail flow plus the rest of the gang, to some extent, have to pay up to HFT shops.  In other words, retail flow has always been getting "screwed" in the sense that they have to pay someone.  It's just that now, that "someone" has changed and that "someone" also collects a few premiums from the other players.  Also, back in the day, a firm could pay an exchange a large sum of money for a seat which would give them direct market access, and trade much faster than any retail person could through their broker.  Now, a firm can pay a telecoms company a large sum of money to use their fiber optics line to the exchange which is a few milliseconds faster than the next guy and trade faster than the rest of the market.  It's gotten to the point where fiber optics (fast, high bandwidth, not affected by the weather) are too slow and companies have been building out microwave dish networks (faster, low bandwidth, affected by the weather) and laser grids (faster, high bandwidth, not affected by the weather) to go even faster (See  At that point, it's just a bunch of HFT firms trying to outdo each other.  In other words, it's not that the HFT firms are against retail folks or the rest of the market, they are effectively attacking each other and trying to eat each other's pie too.  In fact, the real winners of this whole market shift toward speed are the telecoms and technology companies building out these maximum speed communication networks.  Effectively, these telecoms companies are "extorting" HFT firms in the sense that they can approach HFT firms and say, "Hey why don't you pay us to use our network which is slightly faster than anything else that exists?  Oh, by the way, we are also renting it out to all of your competitors.  If you don't use us, you will lose to your competitors and get pushed out of all your speed-sensitive strategies."  This leads to an arms race to the bottom where the arms dealer has all of the leverage.  Also note that the game here isn't about being faster than retail flow since everyone has always been faster than retail even with fiber optics, it's about being faster than each other.  As such, profit margins for HFTs have been shrinking and in a free and open market, profits will continue to be driven down to near zero as they crowd each other out and pay out operating costs to the telecoms companies which own the communications networks.

Also, there is a problem with Lewis' examples of "front-running" an order on one exchange by getting to the next exchange faster.  Market-makers by their natural behavior almost always quote more than the size they are actually willing to do, distributed over multiple exchanges with the intention of canceling their excess orders once their desired size is filled assuming they don't get filled everywhere at the same time.  The reason they do this is to ensure they do not miss an opportunity to trade regardless of on which exchange an opposing order arrives.  For example, Joe is a market maker and is willing to sell 100 shares of ABC.  He puts 70 shares for sale on BATS and 70 shares for sale on NYSE.  Once he gets hit for 70 shares on BATS, he cancels part of his ask on the NYSE so that it now only shows 30 shares.  This is very normal behavior, does not involve any type of "front-running", and would result in vanishing liquidity on one exchange after an order fills on another, the same outcome as what Lewis says is the result of malicious activity.

There's another problem with Lewis' example: how do the HFT firms which "front-run" orders know the full size of the order?  If an HFT firm sees Jim buy 50 shares of ABC on BATS, how does he know if Jim intends to buy 50 more shares at the same price on NYSE or 20 shares or 100 shares or no shares?  By not knowing Jim's full size, the HFT firm cannot with certainty know how much to buy on NYSE and repost at a higher price.  All that being said, of course the HFT firm still has an advantage in that he can buy up some amount on NYSE and repost it and it would be positive EV even though it exposes the HFT firm to some risk.

All that being said, I am not strongly in favor of HFT.  I'm simply making an observation that the question at hand is more complex than at first glance.  To keep it fair, here's one argument against HFT.  From an efficiency standpoint, HFT is wasteful.  Like I mentioned earlier, HFT firms are in an arms race to the bottom as they pay up more and more to stay on the save footing as each other.  They are all trapped in the Nash equilibrium defect-defect outcome of a Prisoner Dilemma situation.  Everyone would be better off if everyone was slower but each individual player has an incentive to be faster which causes everyone to try to be faster which causes everyone to be worse off.  If there was some enforcement mechanism whereby each party would be slower and be guaranteed that everyone else was as slow as him, everyone wins.  Unfortunately, in practice, such enforcement is difficult.  Government regulations are not hard restrictions in the same sense that the laws of physics are hard restrictions.  People will always find a way to game the system to be slightly faster.  For example, even if the government mandated that exchanges could only accept orders through a copper wire, distance to the exchange still matters.  If all HFT servers were mandated to be equidistant from the exchange servers by a 100 meter wire, individual server processing speed still matters (how about ASICs and FPGAs?) and, in a minor and pedantic way, even the coiling/straightness, temperature, and quality of the wires matter.  It's hard to guarantee a completely fair and unbreakable system in general and especially through regulation.

I'd like to end with an analogy to sports.  Most baseball fans are rather indifferent to the use of steroids in the sport.  I've even heard the argument that it makes for a more exciting and watchable game.  But much beyond that, virtually no one complains about the fact that professional baseball players can buy better gloves, shoes, and bats than other professional baseball players.  No one complains that a tennis player's racket costs thousands of dollars or a golfer's club is too well-designed, precise, and light-weight.  Trading is also a game where those with better equipment are at an advantage.  So long as the rules are clear and no one is violating them, all is fair in the game.  We do have the option of changing the rules of the trading game, which I am not against, but if we do, we should take care to lay out the rules carefully and with every consideration taken into account lest we create loopholes which make things worse and also ensure that those rules are enforceable in as precise and all-encompassing a way as possible.

Kevin & Team Buttercoin
Bitcoin Trading Made Easy | 

Buttercoin currently in Private Beta, if you'd like access right now apply here

    P.S.: Whether trading should be called a game or not is purely a semantic distinction. I don't mean that trading is a game in the frivolous sense; I mean that it's a game in the game-theoretic sense.  And if an activity or system can be aptly represented in a game-theoretic way, we should consider the incentives and payoffs of its participants and in its architecture and keep them in mind should we try to design a better system.  In game theory, "mechanism design" and "implementation" are the sub-fields which study these systems and their construction.