Algorithmic trading types

The below article explores what types of algorithmic trading exist, how they are different from each other, and which of them are used for prop trading and which are offered to clients as financial products.

The main types of algorithmic trading:

  1. HFT
  2. CTA
  3. Short term CTA
  4. Factor trading
  5. ETF rebalance
  6. Arbitrage
  7. Stat arbitrage
  8. Market making
  9. Dark pool trading

HFT (high-frequency trading)

This type of business searches for tiny fluctuations in asset prices and the required response rate is measured in nanoseconds. HFT has never been available to clients and has always been used by leading American firms to trade on their own account. Currently, HFT is dominated by 3-5 companies as a result of a series of mergers and acquisitions.

Competition and the struggle for speed have led to a situation where there is no time to use programming languages ​​and assembling, the program code is immediately soldered into a chip at the factory, and trade servers are located next to the exchange servers with the shortest possible wires. This technology is called hardware acceleration.

HFT strategies have a very high Sharpe ratio: up to 10-20. Strategies need to be re-optimized quite often – usually once a week. Even fairly simple strategies work well, but latency, the speed of receiving prices and sending orders to the exchange, is very important. This business has a natural limitation – the speed of light, so the consolidation of this industry will continue and the profits of individual players will continue to fall.

US HFT market leaders:


Typically, Commodity Trading Advisors (CTA) either manage assets on behalf of clients, or advise by transmitting trade advice in the form of SAAS agreements. Trading is conducted on the futures markets. 

Strategies in 90% of cases are fully automated, robots trade and monitor positions.

CTA strategies are now very much more complicated: modern mathematical and physical methods for studying stochastic processes are used, as well as very complex machine learning and big data technologies. If you don’t have a PhD in math or physics, it’s best not to try to work in that direction.

CTAs are also divided according to the length of position retention: from several days to 2-3 months.


Short term CTA is the most profitable and most technologically advanced type of algorithmic trading in the futures market.

The holding period of the position is from 2 to 5 working days, depending on the volatility of the instrument.

Due to the specifics of training artificial intelligence systems, for better performance, as much data as possible is required.

Short-term strategies use minute data for training, which is about 4 million data points per instrument over 10 years. Whereas longer term CTAs with a retention period of 2-3 months use daily data for training, which gives 260 data points per year or ~ 5000 points in 20 years.

Thus larger, long-term funds are forced to exploit long-term trends which tend not to benefit from systematic analysis.

These are the advantages of short-term strategies in the futures market:

  • Higher flexibility (the rate of absorption of qualitatively new information is higher due to the shorter history of market data used to train trading systems).
  • A large number of initial individual algos, only some of which experience a “fading” of the expected positive performance over time.
  • Great flexibility in choosing the type of algo (trend / counter-trend)
  • Constantly increasing portfolio diversification with the addition of new futures contracts or new models for signaling functions, resulting in more profitable and sustainable portfolios.

As a result, you can achieve large Sharpe ratios of between 1 and 2 for short-term strategies.

Access for free ETF Quant portfolios

arty ETF portfolios are based on AI