In the past few years, the financial markets have observed an advancement in computing which has disrupted the conventional methods of trading. The machine learning, statistical and pricing models have kickstarted the revolution of buying and selling assets. We are entering an era of technologies taking over manual programs to make trading more efficient, faster and profitable.

One such market that has emerged with this evolving tech-age is the cryptocurrency market. This financial market is one of its kind!

The digital assets show dual characteristics – partly currency and partly equity. 

In 2017, when cryptocurrencies were on the climb, the financial world witnessed a whole new market that ran on decentralized networks – peak its age. The rise brought in multiple opportunities, manual arbitrage trading being one of them. Now that the market has matured, more opportunities not only manual but even automated have risen. 

Even though the cryptocurrency market has matured, it is still in its nascent stages. With high volatility of its assets, unpredictable tail events and spurious correlation it’s extremely difficult to use the same strategies that are implied on the traditional markets. Here, these assets need comparatively more complex mathematical strategies based on in-depth analysis, statistics and uncorrelated techniques. 

This is achieved through Quantitative trading. It is the implementation of trading strategies with a disciplined and systematic approach. These trading strategies are developed through rigorous empirical research and advanced mathematical computations. 

Getting started with the basics

We can characterize trading strategies pursued by a wide range of traders into two general classifications as directional and non directional trading strategies. Both require various kinds of approaches, various degrees of market information and distinctive trading prerequisites. 

Directional exchanging techniques include taking long and short positions in the market. Traders gain profits when the costs of digital instruments in which they’ve taken long positions rises and when the costs of digital instruments where they’ve taken short positions drops. The majority of the trading methodologies practiced by general traders are directional. Some regular instances of directional practices are trend trading strategies, breakout systems, moving average, cross over trading and pattern recognition practices.

On the other hand, non directional trading techniques are market neutral strategies. Here, traders don’t take any long or short positions; rather they coordinate their positions insightfully. Most non-directional trading practices are intricate frameworks which require excellent computerization and predetermined trading rules. These techniques are used by professional traders and enormous players of the industry.

These primary trading techniques are accompanied by two highly important techniques which at alltimeshavetobe crafted according to the market conditions of the cryptospace. First being the High/low/macro frequency followed by Statistical Arbitrage opportunities.

High frequency trading

The long history and empirical evidence of the financial market has proved that particular market vulnerabilities if exploited in the correct manner, consistently, results to trading success. Quantitative trading constantly takes advantage of these opportunities to gain profits that are not necessarily correlated to the market direction and protects traders from the heavy volatility of cryptocurrencies. High volatility is seen as the biggest problem in the cryptospace, while it is the ideal condition for high-frequency traders to use quant and conduct rapid bulk trades. Even a small change in crypto prices can enable high-frequency traders to book profits through massive orders. HFT also gives quant traders more leverage over ordinary traders as they are exposed to higher risk and lower recovery chances.

Medium frequency trading [MFT]

Medium frequency trading methodologies incorporate all trading exercises, that do not require market microstructure analysis on one side and fundamentally depend on market impact on the opposite side. The most significant distinction it has from high frequency trading is that it can automate any strategy from the automation point of view. Using the factor models it can even automate fundamental investing to analyse as well as implementing the algorithm. Portfolio management in this case is the dynamic process, combination of signal [alpha] revelation and optimal execution on the level of trading scheduling. 

Low Frequency trading [LFT]

Low frequency trades mean that very few and comparatively less trades are taken over a monthly cycle, usually because these trades are constructed on long term charts [such as the daily charts], and take more time to advance yet wind up conveying better levels of profitability. The primary benefits of low-frequency trading are the reduction of trading costs, the minimization of taxable events and minimized risks. This type of trading works the best for the hodlers of cryptospace.

Statistical Arbitrage

Statistical Arbitrage approach fares in the cryptocurrency space on minute, hourly, weekly, monthly and yearly binned data of cryptocurrencies. The most accurate for arbitrage being the minute margins.

You can collect data from cryptocompare or other price comparison data sites with their open source APIs. Once that’s done, accumulate data for each minute, collect Open, High, Low, Close, Volumefrom, Volumeto, and Timestamp data. Open, High, Low, and Close denote the first, highest, lowest, and last price paid for a coin A in minute T, respectively. Volumefrom and Volumeto quantify the volume of coins being traded during that period of time and the equivalent value in USD.

Proceeding forward, the next step is to generate training and trading set. Of the data available for each coin, the first two thirds of the time-series are used as training data [in-sample] while the remaining third makes up the trading period [out-of-sample]. The training and trading sets are strictly non-overlapping to ensure that no look-ahead bias is introduced.

The models and strategies used for quantitative trading are typically coded in Java, C++, MATLAB, R and Python.

These techniques are often combined with the market movements, trader psychology and market exit plans. There are four main segments to work on while framing quantitative strategies are as follows:

  • Strategy Identification – Finding a strategy, exploiting an edge and deciding the trading frequency
  • Strategy Backtesting – Obtaining data, analysing strategy performance and removing biases [ADD] 
  • Execution System – Automating the trading and minimising transaction costs, connecting to an intermediary for execution.
  • Risk Management – Optimizing capital allocation, and designing your strategy to understand trading psychology

The atypical characteristics of cryptocurrency arrangement make models depending on momentary data more viable than customary methodologies dependent on technical pointers. Furthermore, due to the high volatility of the series profits are typically unevenly distributed across different assets. Which makes trading multiple cryptocurrencies at the same time with the aid of machine learning algorithms particularly accurate.

Additionally, quantitative trading is the new buzz of the crypto town. Recently in an article covered by Coindesk, Huobi – one of the biggest cryptocurrency exchange platforms spoke about its users opting for quantitative trading as the ditch manual. The exchange spoke about how its quant clients make trades 70 to 100 times faster than other users. Adding that, “One of our clients makes about 800,000 trades a day, and there are more and more such clients.”

Machine learning in finance has discovered a multitude of applications in the development and advancement of quantitative trading systems. These frameworks exploit supervised models trained on historical data and automates the buy/sell signals in the cryptocurrency market. 


Hudson, R., & Urquhart, A. (2019). Technical trading and cryptocurrencies. Retrieved from

Günter Fischer , T., Krauss , C., & Deinert, A. (2019). Statistical Arbitrage in Cryptocurrency Markets. Retrieved from

A, G., C, L., G, P., & B, E. (2019). Quantitative cryptocurrency trading: exploring the use of machine learning techniques. Retrieved from

Skachkov, I. (2013). Statistical Arbitrage: Medium Frequency Portfolio Trading. Retrieved from

S, R. (2018, February 2). Machine learning in algorithmic trading strategy optimization – implementation and efficiency, by Przemysław Ryś; Robert Ślepaczuk. Retrieved December 6, 2019, from

Y, D., & Z, Q. (2000). JSTOR: Access Check. Retrieved December 6, , from

Financial Applications of Machine Learning. (2015). Retrieved December 6, 2019, from

McKie, S. (2019, April 15). A Guide to Algorithmic Trading in Crypto. Retrieved December 6, 2019, from

Jayaraj, D. (2019, February 19). Why algorithmic trading on cryptocurrencies is a great alternative to mining, especially for a… Retrieved December 6, 2019, from

Trudeau, M. (2019, July 27). Crypto Exchanges Are Benefiting from Algorithmic Trading: Here’s How. Retrieved December 6, 2019, from

Unknown author. (2019). Retrieved December 6, 2019, from

Playle, J. (2018, April 27). Algorithmic Trading in the Cryptocurrency Markets. Retrieved December 6, 2019, from

Leave a comment

Podcast suggestions of the week

TechMerge © 2019 . All rights reserved.