Trading Machines

Dominating the financial markets

Originally published in the March 2012 issue

Beginning with the 2008 crisis in the financial markets, we began to notice phenomena that have not been observed previously over a fairly long history of finance and, in particular, in the futures markets. These phenomena include increased volatility, as well as the fact that market participants ignore some fundamental factors in making their trading decisions. It is becoming increasingly apparent that fundamental analysis and a focus on fundamentals have ceased to play any significant role in trading decisions. Such changes suggest that most of these decisions are being made not by people but by mechanical trading programmes.

Although in earlier times, the volume of these machine operations represented only an insignificant market share, currently their actions are a major factor in defining the direction and speed of movement of the prices of various financial assets. Of course, even before the appearance of trading programmes, sudden speculative price spikes or slumps occurred that were in conflict with fundamental factors. However, there were quite a number of arbitrageurs, investors, and commercial players who, by their actions, quickly restored the status quo in the market, returning prices to normal ranges.

At various levels, including governmental ones, the increasing role of high-frequency trading programmes, and ways to limit the damage of their impact, are being discussed. However, I would like to propose a new idea by stating that the market is not only being affected in the short term by the actions of these machines, but also that trading machines currently play a dominant role in financial markets in general.The best examples, the essence of the phenomena, are shown by spreads and exchange rates. In both cases we are talking about relative value. In the case of spreads, this refers to the difference between the prices of related goods or financial instruments. In the case of a currency, it means the purchasing power of one currency over another. In contrast to the outright prices, the impact of fundamentals on spreads and currency pairs has always been much higher.

Prior to 2008, spreads movements that were contrary to the fundamental factors rarely occurred; however, after the 2008 financial crisis, such phenomena have often been observed. Here are several examples that demonstrate the essence of the problem. Most recently, the €/CHF rate (see Figure 1) has created a lot of noise in the press. Contrary to all logic, within six months, the value of the franc has strengthened by nearly 30%. In other words, the purchasing power of the CHF rose by 30% relative to the purchasing power of the Euro. The reason for this phenomenon may be related to the following fundamental factors:

• The sharp increase in Switzerland’s GDP
• The contraction of the Euro area economy to a similar size
• The simultaneousexpansion of the Swiss economy and the contraction of the Euro area economy
• The rise in the Switzerland bank’s refinancing rate

However, nothing like that stated above has yet been observed. The GDP growth rate in the Euro zone and Switzerland are at about the same level, and the refinancing rate in the Euro area is even higher than in Switzerland. However, during an extremely short time, the exchange rate of €/CHF almost reached parity, which contradicts any common sense. It cannot be ruled out that the strengthening of the franc would continue if not for a radical statement of the Swiss bank management to establish the minimum values of the exchange rate at 1,200.

Another example is the pair $/Yen (see Fig.2).

It is no secret that Japan’s economy is much more indebted than that of the US. The US GDP growth rates are higher than similar ones in Japan, and the refinancing rate in Japan for many years was close to zero, unlike the US, where this happened only after the 2008 crisis. However, recently we have been witnessing a process of continuous strengthening of the yen against the dollar. Moreover, this is happening despite the constant threat that the Bank of Japan may make an intervention. There is no fundamental factor that can explain the behavior of this currency pair. This confirms once again that such decisions are being made not by people but by trading programmes that operate huge sums of money.

Here are a few examples in support of this idea. Although they are not known very well to the general public, participants in metals and energy futures markets are aware of these three spreads. Those three spreads are spread platinum/gold, crack spread heating oil/crude oil, and spread Brent/WTI.

Consider the spread platinum/gold. For platinum, a precious metal much more rare than gold, the production volume is approximately 30 times less than that of gold. Platinum is widely used in jewelry and for industrial purposes. The main demand is coming from the automotive industry, where it is used for the manufacture of catalytic converters. Industrial demand for platinum is inelastic, since there are practically no alternatives to it, except for palladium, which belongs to the same group of metals.

Platinum supply, unlike gold, is also inelastic, since there are no large reserves of platinum in the warehouses. In addition, 80% of the world production of platinum is concentrated in South Africa—a country that is not a model of stability. The production cost of platinum is significantly higher than that of gold. Taking into account all these factors, we can say that the price of platinum should significantly exceed the price of gold, as was observed for many years when platinum was worth an average of 20% to 30% more than gold (see Fig.3).

If we look at the history of the past 10 years, only during the crisis of 2008 was the price of platinum at the same level as the price of gold, and for a short time it was even $20 cheaper than gold. However, this situation did not last long, and the spread began rapidly widening, which led to the restoration of the traditional differences in prices.

Over the last two months we have been witnessing a situation in which platinum is $120 to $180 cheaper than gold. This is a clear anomaly that the market is not even trying to correct, because the value spread has stabilized at these values.

The spread Brent/WTI is actively traded by participants in the oil market. This spread is based on the difference in prices between Brent oil futures, which are traded on the ICE (London), and WTI oil futures traded on the NYMEX (New York). WTI predominantly serves as the benchmark for the US market, while Brent is the pricing benchmark for most of the European and Asian countries.

Generally, these two grades of oil are essentially similar and have very little difference in chemical composition. Both contracts are based on 1000 barrels of oil, and both are quoted in US dollars. Theoretically, WTI should sell at a premium to Brent. This premium is attributable to two factors: a somewhat lower content of heavy sulfur compounds in WTI and the difference in freight costs between Europe and the USA; that is, if we assume that oil is shipped from Saudi Arabia, then the per barrel shipping cost to the US would be higher than that to Rotterdam. However, the actual size of the spread often differs from its theoretical values.

It should be noted that until 2008 the value of this spread steadily fluctuated in the range of +/- $5, with rare short-lived spikes (see Fig.4). In the midst of the 2008 crisis, a widening of the spread was observed, when the value of the difference between the two benchmarks reached $12–$13. But within a few days, the market quickly returned the spread to its traditional range. Since January 2011, we have seen a steady widening of this spread, which in October reached nearly $28 per barrel. In other words, at a cost of WTI at $80 a barrel, close by the chemical composition, Brent is almost 35% more expensive. Even to a person unfamiliar with market spreads, this difference may seem absurd. However, the market has held a spread value of more than $20 for the past four months. This difference in price cannot be supported by any reasonable, fundamentally significant explanation.

Another spread I would like to discuss is the heating oil/crude oil spread. This spread may be categorized as a processing spread but, as distinct from a full crack spread, it has a truncated nature and essentially reflects the margin earned by petroleum product manufacturers on heating oil. As opposed to a more complex full crack spread, this spread is easier to assess and trace. Historically, this spread rarely exceeds 10% to 20% of the price of oil (see Fig 5).

In 2008, for the first time ever, the value of this spread reached a level of $40. However, the value of the spread at these levels lasted only a few days, and then the spread returned to the traditional boundaries. Note that the spread value of $40 was achieved when the price of oil was at $145 per barrel; that is consistent with the value of a spread of approximately 27% of the cost of barrels of oil.

Since the summer of 2011, the value of the spread has remained stable at $35 to $42, and the ratio of spread to the cost of barrels of oil reached on some days approximately 50%. Such spread values seem absurd to any veteran of the energy market, as there are no fundamental factors to explain such a magnitude of spread movements or how they maintain such high levels for such a long period of time.

All of the above examples suggest that the market is short of participants who are oriented to the fundamentals and able to bring it back to normal. Insistence on making trading decisions that are contrary to sound logic can be attributed only to the work of programmes that operate on very different parameters than that of the human fundamental analysts. Because of huge leveraging of financial markets, the impact of trading machines has been exacerbated, allowing them to play a defining and dominant role in the markets.
This is what can be expected in the near future, if the number of trading programmes continue to increase:

Giant volatility: Intraday variations of 5% to 10% will no longer be deemed a rarity, and they will instead become the norm.

Negative Price Values: As strange as this idea sounds, the futures markets will observe short-term effects of negative price values, which means there will be some financial assets that will be worth less than zero. In principle, this should not be surprising, since the programme selling anything over zero, followed by the purchase of something below zero, is not something that is contrary to the logic of the machine. We are now witnessing the phenomenon of negative returns on short-term US bonds.

Large Drawdowns: There will be large drawdowns of hedge funds, basing their work on the use of trading machines. That huge volatility, which is a result of the trading machine actions, will also have a detrimental effect on these funds.

It remains to be hoped that after quite a long period of four or five years of trading programmes dominating the financial markets, people will begin to reconsider their trading approaches and return to the situation where humans made major trading decisions.

Kirill Perchanok is the author of Futures Spreads: Classification, Analysis, Trading available on Amazon. His research focuses on futures spreads and technical analysis.