re: Are Markets run by Human Need or Algorithmic Machines ?

Rodrigo Diaz — Sun, 16 May 2010 21:51:00 GMT

Are Markets run by Human Need or Algorithmic Machines ?

I recntly watched an interview in Awe and Shock, at www.ft.com in the Trading room Video section. It is an interview about the growing trend of High Frequency Trading and its growing popularity for Market Making in Asia and Europe.

The interview is Tittled "May 11 : Does speed Kill ? " with Hirender Misra from Algo Technologies, putting a fight in favour of High Frequency Trading... (Naturally).

Here is my take on this. Markets are meant fundementally for buying and selling, It naturally has to be driven by a genuine need for the underlying that is being traded. This genuine needs stems from human need for capital goods and services as much as his own desire to invest and grow his capital. A simple example is if a person who is in need of money in short term wants to sell his securities he might get a buyer who takes a long term view on the security and is willing to buy the same at market bid price. Here there is a fundemental and genuine need and they play their small part in the overalll market price discovery .Now replace this scenario with a Market where most of Trading volume is coming from computer algorithms that are artificially creating huge volume of Buy-Sell simple based on some algorithms which in turn take decision based on inputs given such as general market trend for the day, possible variance in the prices and any particular news specific to the security and the market itself.

These algorithms assign weight factors that eventually come up with a decision to buy/sell the stock based on current trend. Note here that, no one is looking at fundementals of the security itself, i.e., its

PE ratio, price to book value, dividend history, etc., etc. If markets start ignoring fundementals altogether and their sole purpose of trading becomes that small spread due to volitility then In the long

run we will destory the capability and the fundemental responsibility of markets i.e., providing a platform for price discovery.

High Frequency Trading is largely unethical although it can not be proven in a court room. Below are some key issues that proliferation of High Frequency Trading will create.

Volume Distortion:

It will distort volume of trading done on a particular security, index or commodity. In a free market, increasing volume and decreasing volume on trading indicates real demand/supply mechanics from the economic agents involved. It helps figure out any constraints of supply and the supply chain which can be fixed to provide a price stability. For Instance, if some one was trading mango futures in India during summer using High Frequency Trading, it would artificially create more volume of trading and mislead business to think that there is more demand for mangoes. This will lead to rise in price of mangoes and eventually more capital will inflow into mango futures for speculation trading which otherwise would not have happened in a free market if only humans were allowed to Trade.

Imagine a large portion of capital freed from a business undertaking getting diverted into trading mango futures. More capital chasing speculation on mangoes will drive the prices up eventually.

Put in simple words, we trade for a purpose and genuine need. High Frequency Trading defies that. Mr. Misra says we would be in the dark ages without technological innovation. I completely disagree with that argument. 1000 years ago people traded mangoes/rice/any commodity for their basic need of food and even today we do that. Some fundamental things in this world don't change.

A general level of increase in volatility in markets:

High Frequency Trading will lead to a jump in the general level of the average volitility of markets. Weather it is individual securities, commodities or indexes. This is because algorithms will push large volume orders, draw them back or cancel them in a flash of seconds creating an artificial demand/supply in the market that will induce changes in the behaviour of prices. Add to this that there are companies working on introducing a News Scanner coupled with High Frequency Algorithms, that will scan/read all news particular to a company and decide what is to be done.

All in a flash of seconds... imagine a huge sell order at the beginning of a market that will drive the price down and same order can be broken into small demand/buy side pieces to buy the security at lower prices later.

Effect on Market Jumps and Diffusion:

Let us assume that we have a hypothetical algorithm that combines news sacanning and algorithmic techniques to trade, in the middle of a trading day on Apple shares. Let us also assume that Apple is going to announce its quarterly results that day. The quarterly results beat expectations but there is a sudden news that the CEO of Apple has suffered an accident and is in serious condition. Now what decision is the algorithm going to take? What is it influenced by? How much weight should it give to company making a profit vs. the accident which ultimately should have very little affect on the security. If the algorithm decides to push in a huge margin order of sell, It could easily create a stampede on the stock and finally turn into a rout on the stock. What if some hours later we come to know that the accident was not that serious and actually the CEO is doing good? Can the algorithm

now do precisely the reverse to bring the price back to a fair level? No it can't. The market participants would be scared to buy the security because of uncertainty.

Take a sample Example :

Yesterday a spanish newspaper announced that Sarkozy was planning to pull out of Euro. What route could this cause on FX markets if we have an algorithmic trading coupled with a news scanner? Of course EUR is already getting battered.

Make a note that most investment banks have quantitative models to predict and catch jumps in markets. They model it using "Jump Diffusion Algorithms" ..This can lead to a route in markets as these algorithms will push huge sell orders to benefit from a falling market.

The epistemological question of how humans think and make decisions is beyond the level of modelling using computers or statistical algorithms. It simple is not possible.

And if we replace genuine human action in markets by computers it will spell a disater in the long run. I think regulators should have to curb High Frequency Trading to a certain level of volume for major market participants.

Unfair game of machines vs. humans:

A High frequency Trading also requires a direct market access (DMA) via a client API. A normal trader or investor from the public certainly does not have access to this kind of facility. He either depends on a broker or does it himself. In either case there is simply no comarison at all. It is very much likely that this will ultimately lead to a fight of common man vs. the machine, i.e., common man vs. the hedge funds/Wall St./high net worth clients. This is simply unacceptable.

Conclusion:

High frequency Trading is largely unethical and can have devastating affects on markets and a bad day if these algorithms go wrong. I am sure they will. I am a computer programmer myself and I know. No matter how well you write your code there will always be bugs. Look at the latest example of the misterious DOW collpase over 1000 points. This is just the beginning. The regulators should wake up and understand these basic issues otherwise there is very strong possibility that people will loose belief in the markets on day.

anand312 - All Comments

re: Are Markets run by Human Need or Algorithmic Machines ?