Types of Algorithmic Trading

Part 3, final

Today, we finish our trilogy about Types of Algorithmic Trading. We counted 9 types of Algorithmic Trading in our initial blog post. We have already described 5 of them: HFT, CTA, Short term CTA, Factor Trading, ETF rebalance. Now for the final instalment: Arbitrage, Statistical arbitrage, Market-making, Dark pool trading.


High Frequency Arbitrage Strategies have a fairly simple idea: buy something on one exchange and sell it on another (SPATIAL ARBITRATION).

There are even more sophisticated strategies, for example – an index against a basket of stocks from which it is composed.

Since the idea of ​​arbitrage is simple, it is difficult to make money – all such strategies are now automated and are completely managed by robots.

There have been cases in history where a group of entrepreneurs laid cable from Chicago to New York City to compete for signal speed while competitors used towers with signal lights. The film “The Hummingbird Project” was made about this. There is also an interesting book Flash Boys, which describes this.

It should be noted that arbitrage strategies equalize prices on different exchanges, bring liquidity, and, most importantly, transfer economic information.

Due to the great competition and simplicity of the idea, the profit from arbitrage is very small, and this business is available only to very large players with large capital.

We could compare arbitrage robots with sherpas who carry cargo for climbers – climbing to the “Alpha Summits”.


Statistical arbitrage is a very technical type of algorithmic trading, which involves not only the presence of a sophisticated high-frequency IT platform but also a large team of mathematicians and physicists.

It is essentially a combination of high frequency trading and arbitrage strategies. Statistical arbitrage, unlike conventional arbitrage, works with multiple assets and/or linear combinations of them. Assets (or combinations thereof) are used that are either strongly positively correlated with each other or strongly negatively correlated. For example, gold is correlated with 10-year US bonds, and the Canadian dollar against oil, or the S&P 500 index, is negatively correlated with the VIX volatility index.

The robot measures a certain equilibrium value from the price differences of several instruments, and trades are executed when this difference deviates relatively strongly in one direction or another. Transactions are made at once with all assets at the same time.

It should be noted that since there are quite a few quantum funds in the world that operate this type of trading, its margin has dropped dramatically, and it is recommended to use statistical arbitrage strategies together with trend and counter-trend models when ensemble trading systems.


Since the early days of exchanges, there have been brokers who were always ready to buy and sell goods. They held the so-called spread: offering buy and sell orders at the same time. These brokers are called market makers.

They are very much needed by both exchanges and traders, as they create liquidity and help smooth out large price fluctuations.

With the advent of electronic exchanges, the situation has not changed, only now robots play a greater role.

This business is also heavily monopolized, and the markets are dominated by only a few players, for example, Citadel and Virtu. These companies list up to 100,000+ instruments simultaneously on many exchanges around the world. The influence of Citadel is enormous in the stock markets. The fact that Ben Bernanke, after leaving his post as Fed Chairman, moved to work at Citadel speaks volumes.


A dark pool is a private, over-the-counter securities trading platform.

There are about 30 darkpools in the United States, which account for about 15% of the turnover of American stocks.

Initially, the idea of ​​dark pools arose in the ’90s when there were so many securities under the management of investment banks that it was easier to find a buyer or a seller within the ecosystem (or darkpool) of the bank rather than list transactions on the exchange.

With the advent of high-frequency traders, banks realized that it was profitable to sell them access to their darkpool for a fraction of the profits generated by trading with their darkpool clients. This, of course, was already on the verge of a conflict of interest, but somehow it worked, and the regulators could not catch hold of it.

Basically, each client, when buying or selling a stock, gives a portion of the money to a high-frequency trader admitted to the darkpool.

Thus began the era of robotic DARK POOL TRADING. This is well described in the book Flash Boys.

Furthermore, with the advent of discount (IB, Schwab) and free (Robinhood) brokers, they also began to sell access to their flow to HFT traders.

Brokers in their reports openly reflect income from the sale of client orders to HFT traders (mainly Citadel and Virtu), which amount to up to 70% of the proceeds. An example can be found in the Robinhood report.

Access for free ETF Quant portfolios

arty ETF portfolios are based on AI