I guess my whole point about probability is that it is a contingent concept and hence can be replaced (it is historically dated). Its main weakness is the identification and delimitation of states of the world to which the probability distribution applies. In case of roulette, or dice, or marbles in a jar, this is not a problem as the possible states (or draws) are clearly defined. In “massive reality”, however, or in the market (which is also a massive reality), it is not so clear that it is even legitimate to discern and identify possible states. 

For instance, you are tempted to identify the prices of the option’s underlying as only states, however, the mere fact that options trade in their own market leads to you to also “sample” different volatility levels as other states; this, in turn, is not enough because barrier options (or exotic options of payoffs more complex than the vanillas) also simultaneously trade independently of the vanillas and their prices may not be explainable except in an even higher-level model where not only volatility is stochastic but its own volatility is stochastic, etc. 

In sum, if we define a market (of contingent claims) as a place where contingent claims of every level of complexity trade simultaneously, at prices that are not redundant which each other, then you will find that you can never capture this in a picture with defined states. People commonly think that there is such a picture, only it changes and expands over time. This is exactly what I dispute, for, in my mind, the definition of market is instantaneous (all contingent claims must instantly trade non redundantly).

The more I thought about this problem, the more it appeared to me that the picture (contingent claims, prices) should completely replace the traditional picture (states of the world, probabilities). Think of the market (or at least, of an idealized picture thereof such as I offer in the book) as a whole new logic, which replaces the logic of probability. Instead of abstract metaphysical states, let us adopt written and material contingent claims. Instead of probability trees and probabilistic transitions, let us adopt the massive exchange place and the coexistence of all prices. 

As a matter of fact, derivatives practitioners very commonly “infer” the stochastic process of the underlying from the instant constellation of prices of its derivatives. They very seldom try to infer it from the history of prices of the underlying! My whole philosophy is to try to radicalise this alternative and to no longer believe in temporal processes (or generally in probability) but only in markets and prices.

— Elie Ayache

datavisualizations:

This visual offers a primer on alternative assets (hedge funds) and the role that shadow accounting plays in the industry.

Shadow accounting isn’t as shady as it sounds. Fortisbank’s definition is:

According to IFRS 4 an insurer is permitted, but not required, to change its accounting policies so that a recognised but unrealised gain or loss on an asset affects the measurement of the insurance liabilities. The related deferred adjustment to the insurance liability (or deferred acquisition costs or intangible assets) is recognised in equity only if the unrealised gains or losses are recognised directly in equity.

In plain English:

Shadow accounting is the generic term to describe an accounting system where two sets of books are kept. In theory both books should show the same values, differences can only come out of a mistake.

It doesn’t have to be two identical set of books, it’s good enough that they show the same values. The most well known system are bank accounts. Both the bank and the holder/client do their own accounting. Sometimes “shadow accounting” is used to make certain calculations, this is particular true when it comes to the accounting of insurance companies.

An integral accident could have happened with the flash financial crash of 2010. At 2:40pm on May 6th , 2010 the overall prices of US shares, and of index futures contracts that bet on those prices, fell by a staggering 6 percent in just five minutes. A fall of 1-2 percent a day is normal. Overall prices recovered quickly, but some very peculiar price fluctuations happened with some individual shares. The global consultancy firm Accenture’s shares fell from $40.50 to a single cent, and those of the auction house Sotheby’s jumped from $34 to $99,999.99.

These two extreme prices provide the clue that this was a crash caused by algorithms because one cent and $99,999.99 are the lowest and highest possible prices that can be entered into electronic trading systems. Thus what happened between 2.40 and 3pm was an accident of the breakdown of computer trading systems and the interaction of software bots on the stock market.

What happened was that:

One algorithm would sell futures to another algorithm, which in its turn would try to sell them again… In the 14-second period following 2.45 and 13 seconds, more than 27,000 futures contracts were bought and sold by high-frequency algorithms… By 2.45 and 27 seconds, the price of index futures had declined by more than 5 per cent from its level four and a half minutes earlier. The market had entered a potentially catastrophic self-feeding downward spiral… At 2.45 and 28 seconds, the price falls triggered Globex’s Stop Logic Functionality [which is]…designed to interrupt self-feeding crashes and upward price spikes] (MacKenzie,18)

The Stop Logic Functionality imposed a five-second pause in trading. It worked. The downward spiral stopped, and by 3pm the financial market was more or less back to normal. Alison Crosthwait, an analyst of electronic trading, argued in an internet discussion forum that the 5 second pause “provided ample time for market participants to consider their positions and return to the market… [It] allowed market participants [i.e. algorithms] to regain confidence”.

— Kjøsen, A. (2012). “Logistics II,” pp. 11-2