Bitcoin Markets Blog

Hi Everyone,

This week markets fell from about $645/BTC to $612/BTC.  This likely suggests that the market thinks Tim Draper overpaid for his 30k BTC.  Bitcoin warnings out of Italy and bitcoin regulation out of Argentina could also have dropped the price a few ticks.

In other news:

• Litecoin botnet Lecpetex was taken down by Facebook and Greek local authorities: http://bit.ly/1kMAQNd.
  
• Gox will auction off bitcoins.com and give proceeds to creditors: http://bit.ly/W4Y8sl
  
• Next version of Bitcoin Core will have floating fees depending on your inputs' coin-age and average transaction confirmation time on the network: http://bit.ly/1n8buhu.
  
• BitXBay, an Ebay-esque decentralized and anonymous marketplace was released this week: http://bit.ly/1twXMcE.  They use collateral multisig escrow payments to keep parties honest.  I'm not convinced that this incentive system actually works.
  
• 3.5 BTC puzzle with 32 page solution solved: http://bit.ly/U69JG5.
  
• Grooveshark and XHamster now accept bitcoin.
  

First, let's ask ourselves how hard is it to win money in the markets?  Assuming that the market is a zero-sum game (i.e. someone must hold a losing position for someone else to win; zero-sum in money but positive-sum in utility since not all participants enter the market for profit only (e.g. buy weather futures to mitigate crop failure risk)), it seems that by being a better predictor of an asset's future market value than half of the players in the market, you will win.  But that isn't actually the case.  Even in this simplified model, you would need to be better than half the capital, not half the people.  Since capital tends to centralize toward the better players (i.e. the winners) in this sort of game, you likely need to be better than 90%-95% of the other players.  Adding on frictional costs like broker fees and exchange fees makes the game negative-sum and ups the threshold to beat. So when is it possible for you, as one person, to be right where many people are wrong?

In the super-short term, profits are divided according to the power law (i.e. 1st takes most, 2nd takes some, 3rd takes crumbs, etc.) to those firms with the best and fastest technology.  Strategies are often simple although the computer hardware and routing/communication technology is very sophisticated.  Speed is valued above all else.

In the short-medium term, in the absence of private and material information, you will likely lose in your analysis of public information to the teams of Math/Stats/Physics PhD's, veteran traders, and other experts at top hedge funds and prop shops.  On top of that, you are also competing against people who actually do have insider information so you will be left picking up crumbs at best and losing at worst.

In the long term, a singular powerful insight not obvious to most people can net you a huge payday.  This is where having a better mental representation of the world can make one person right where hundreds/thousands of people are wrong.  Being able to predict the subprime mortgage crisis would compensate you for thousands/millions of missed short and medium term opportunities.  At this frequency level, being perceptive and meta-reflective matters more than anything since the market, itself, is highly reflexive.  Primary effects cause secondary effects which cause tertiary effects.  All parts of the market system update to new states simultaneously with complex feedback loops and general interconnectedness.

What about market efficiency?  How efficient is the market?  The wisdom of crowds is a phenomenon whereby the aggregate opinion of a large population tends toward the truth even if most members of the population are uninformed.  For example, if a group of people in a room are shown a jar with some number of marbles in it and told to guess the number, the average guess of the group is usually very close to the actual number of marbles in the jar.  Does price discovery in the market follow this same process?  No, while some market participants are trying to guess the intrinsic value of the investment asset, others are trying to guess, at a meta-level, what that aggregate guess will be.  In this way price discovery is more reflexive than the discovery process of the number of marbles in a jar.  Another difference between the marbles example and the markets is that with guessing the number of marbles in the jar, there is no real emotional investment into whether there are 42 marbles versus 47 marbles but with the markets real money is at stake so greed and fear play in integral role in players' behaviors and the determination of price.  This also adds to stronger market reflexivity.

Consider a second experiment (Keynesian beauty contest) where a group of people are told to pick a number between 0 and 100 such that it is 2/3rds of the average guess of the group.  A person might think about choosing 66 and before realizing that should everyone else guess 66, he should guess 44.  But if everyone else guesses 44, he should guess two-thirds of that and so on.  Under perfect rationality on the part of all players, the optimal guess converges to 0.  However, in actual experiments where this game is played out, the winning value is often somewhere between 10 and 25.  Taking this further, we can think of this winning value as a quantification of the meta-rationality of a group.  Without a functional form, we can make ordinal comparisons by saying a group which has a winning value of 10 is more meta-rational than a group which has a winning value of 20 but we cannot compare absolute distance in any way (group-10 is more meta-rational than group-20 by the same amount that group-20 is more meta-rational than group-10).  Nevertheless, it should suffice as a tool to think about market efficiency.  Any market comprised of a population which does not yield a winning value of 0 in this experiment cannot be considered an perfectly efficient market.  How inefficient a particular market is can be quantified by how far it's population would deviate from 0 in it's winning guess should they be subject to this experiment.

The market is filled with a diverse set of players, some more meta-reflective or meta-rational than others.  Predicting the future price of an asset, then, can simply be thought of as an exercise in being slightly more meta-rational than 50% of the capital (95% of the players) in that asset (game).