How trading on the exchange actually works, and how it can be improved: A simple algorithm (part 3)

We continue the cycle of articles on how high-frequency trading is organized in simple and understandable language. In the previous two posts, the author described the mechanics of the process, the basic concepts and spoke about the social aspect of HFT (why the race for the speed of reaction in the market has become an end in itself). This time we will talk about the negative consequences of the chase of traders in a race with time and how to level them.

Why do HFT traders compete in speed

Once again: traders who practice high-frequency trading make money on market-making strategies. In simple words, they sell liquidity to speculators at a price equivalent to the distribution of bid / ask. The main rule of such a strategy is to be on top of the order register.

')

SELL(owner=Whiskey , min_price=10.02, quantity=200, time=9:45:00.000) <- Trades third SELL(owner=Echo, min_price=10.01, quantity=100, time=9:45:00.003) <- Trades second SELL(owner=Topher, min_price=10.01, quantity=200, time=9:45:00.001) <- Trades first ------ BUY(owner=Sierra, max_price=10.00, quantity=100)

In this example, Topher will trade first because his order came first at the lowest price. Whiskey made the first bid, but will be the last in the queue because it has offered a higher price. There is a general rule: the more you trade, the higher the profit is theoretically. It is possible that in our case, Echo wanted to offer a more attractive price. Say $ 10,008. However, the current norm of the SEC Rule 612 minimum price increment prohibits it from setting the sale or purchase price in fractions of a cent. Therefore, it can be argued that Topfer was the first, simply because it was faster, and not because of the best price.

Sub-Penny Rule

On US stock exchanges, there is a minimum price increment rule (Sub-Penny rule - similar rules exist on other stock exchanges), which acts as a price lower limit for liquidity. It prohibits selling liquidity at a price below $ 0.01. Suppose Echo wants to offer a better deal, but this provision does not allow her to beat Topher. The net effect of this thing is this: anyone who wants to acquire liquidity must overpay $ 0.02 per share. And all this income goes into Tofer’s pocket.

As in the case of the classic minimum income level, two sides suffer from this. First, the buyer who overpays. Secondly, the seller who is ready to offer a lower price, but it is knocked out of the market.

At a price above $ 0.01, the prices become intermittent. For example, for a $ 0.05 bid / ask distribution you cannot enter the market with a price of $ 0.049 or 0.045. Therefore, players are forced to pay more attention to the reaction rate, not price. Price competition is possible when one of them is willing to offer a price of at least $ 0.01 better than the other. In a real market situation, this is an exceptional case.

And when price competition is impossible, market makers are forced to compete with each other by other methods - in this case, this is the speed at which applications are executed.

HFT traders strive to "rent for rent"

The term economic rent determines the difference between the raw material costs (any factors of production) for the production of any product or service and the price of this product. In a competitive market, rent usually tends to zero.

As a classic example of economic rent, the author cites a newly built city in the American West. The landowner (let her name is Adele) is the one who, as a rule, arrives first. When Adele along with all her belongings appears in a new place, she builds a hotel. Anyone who arrives later than her has to pay her rent for living or living on the street. Thus, Adele is free to set any price. Suppose $ 100 per day.

As the town develops, new hotels are being built, and Adele is eventually forced to reduce the price to $ 50. The difference between the monopoly price of Adel for living and its new price in a competitive market ($ 50) is the economic rent it receives or “alternative income”.

If our Adel is fixated on receiving income from rent, she is in rent seeking (rent seeking). In this strategy, she spends all of her cash resources to get her share in the already existing social wealth, without trying to create something new. Classical economics textbooks offer her in this situation to burn the forests around so that no one builds more hotels. In the current situation, she would have had the opportunity to convince the city council to conduct zoning and use other types of regulation to stop competition.

Another example of economic rent is the minimum income and the creation of trade unions, monopolies, regimes of exclusive licenses, the behavior of corporate managers who refuse to improve production processes in order to protect their seat.

In the case of high-frequency trading, the minimum increment rule gives rent to someone who is faster than others. In our example, Topher collects a rent of $ 0.002 per share. His trading system, an algorithm that shortens the time it takes to receive a response to a request from the server, is an example of rent-oriented behavior. He spends all his resources to win the right to this share, instead of creating something new.

It is not difficult to guess that, in fact, the rent is all $ 0.005 per share. We get this result if we consider the best price on the market as a random distribution variable on the interval [nx $ 0.01, (n + 1) x $ 0.01]. Topher can invest in improving his strategy of receiving an investment from $ 0.0049 per share. It is beneficial for him, even if none of his clients receive income from it.

Cancel Minimum Increment Rule

Throughout the twentieth century, the markets had an increment rate of 1/16 dollar (0.0625). In 2001, the Securities and Exchange Commission ordered the financial markets to reduce the increment step to 0.01, and all prices to tenths. At first such a move seemed very successful, it allowed to significantly reduce trading costs for players, since now the bid / ask distribution was much less than 0.0625. For example, at the time of this writing, the distribution to Bank of America shares was $ 7.93 / 7.94. In the 1/16 system, the best distribution could be $ 7,875 / 7.9375 (or $ 7.9375 / 8.00).

In those days, when traders traded only manually, the minimum price increment served a useful purpose. This helped prevent a war between traders in an attempt to bring down the price for hundredths. Today, these concerns are no longer relevant, since market players have learned how to use computers.

As a result, the author proposes to make the minimum increment even more fractional. Let's say add shares up to $ 0.001. This will diminish the desire of HFT traders to accept rent, and reduce the cost of a slow reaction for others to $ 0.001 per share. In this case, Topher may need to focus on other things. As a result, IT professionals working for Tofer and other similar financiers may leave to work on improving the infrastructure and speed of Facebook or Amazon.

disadvantages

In any case, such global changes in the structure of financial markets should take into account possible risks. Fortunately, changing the minimum increment is not a new thing. We can see its consequences. We have once reduced the increment from 0.0625 to 0.01 without any obvious benefit or visible harm.

In foreign exchange markets, the usual minimum increment is $ 0.0001. At the time of writing, the bid / ask distribution for the euro / dollar pair was $ 1.3027 / 1.3030. For the securities market there is an example of the Bombay Stock Exchange and the Indian National Exchange, where, in terms of a dollar, the increment step is $ 0.001.

There are a couple more objections, apart from the fear of global shocks that are worth mentioning. Someone thinks that the volatility of markets can grow. According to the author, this should not worry. Now, when the rule is $ 0.01, the trader has no incentive to raise the price by 1/1000 cents. Without it, such an incentive appears. The network effect will lead to an increase in the number of price changes, volatility in general will increase, but not as seriously as skeptics suggest. Several small jumps - this is not one big jump. Instability increased in 2001, but then the price of introducing new rules was considered nothing compared to the shortcomings of rule 1/16.

Another objection is based on the fact that it will be more difficult for a trader to assess markets. Instead of just looking at the list of stocks at $ 10.00, he will need to add stock positions at a price of 10,000, 10,003, 10,0015, and so on. This will require a habit and a set of existing tools. For example, the program may display prices "no worse than 10,0029". It is enough for a market participant to see a cumulative price review, not a price tree review.

Benefits

The first advantage of canceling the rule is that the bid / ask distribution is narrowed (in the US, perhaps by half a cent). This will make it easier and cheaper for players to open deals. Although, to be realistic, this is not such a big advantage.

More significant will be the redistribution of forces from the pursuit of the speed of reaction to other goals. When the author was a HFT trader, his colleagues, all as one, were quite clever people, capable of doing meaningful things. It is hardly worth wasting your abilities to make one trader faster than another. Many clever people will be able to focus on other tasks. Perhaps more useful is outside the scope of financial markets. There are a bunch of technology companies. There is a need for talented people in the field of financial risks.

In the context of changing the situation, there was also a proposal to impose a tax on financial transactions. But this is too extreme a measure.

To be continued...