Data Science meets Cryptocurrency Trading — more than Just Friends
By CoinDCX on The Capital
How does a crypto exchange function and how is data science shaping these transactions? To start with let's understand what is cryptocurrency trading and how these transactions are executed. Cryptocurrency trading is the act of speculating on cryptocurrency price fluctuations via a CFD ( Contract for Differences) trading account or trading in the underlying coins via an exchange.
CFDs trading are derivatives that provide the option to speculate on cryptocurrency price fluctuations without actually taking ownership of the underlying coins. CFDs trading has the potential to execute both long (buy) and short (sell) options with the cryptocurrency price movements. Both are leveraged products, which means only a small deposit known as margin is required to gain complete exposure to the underlying market. The profit and loss are calculated with respect to the full size of position, hence leverage magnifies both profits and losses.
On the other hand, while trading in cryptocurrencies via an exchange, purchasing the coins itself is required. An exchange account is created, full value of the asset to open a position is maintained and cryptocurrency tokens are stored until the investor wants to sell them. Cryptocurrencies exist merely as a shared digital record of ownership, stored on a blockchain, unlike traditional currencies. The transfer of cryptocurrency from one user to another happens via the users' digital wallets. For the transaction to be considered final verification and addition to blockchain in mandatory, the process is called mining. This is how new tokens are created in the ecosystem.
Though cryptocurrency markets are decentralised, i.e. they are not issued or backed by a central authority such as a government, but the exchange can be centralised or decentralised (DEx). Centralised Exchanges involve the role of a third party between buyers and sellers to execute the trade, whereas in a decentralised exchange the trade is facilitated via blockchain eliminating the need of a third party. Both have different characteristics in the sense though centralised systems are exposed to hacking frauds but a user can recover an account password if forgotten, the opposite is true for decentralised exchanges. When using centralized exchanges the users send their funds to wallet under controlled conditions, managed by one entity known as the exchange. While on decentralised exchanges, digital signatures are relied on to directly authorize the transactions. This is precisely the reason behind slow working of decentralised exchanges vis a vis the centralised exchanges. Decentralised exchanges permit the payments in cryptocurrencies alone whereas centralised ones allow the use of traditional payments as well. Along with serving as a trading platform centralised exchanges store private information as well. Another additive feature of centralised exchanges is the possibility of exchange of FIAT into a cryptocurrency and vice versa.
Most analysts in any trading market use highly speculative models which aren’t necessarily based on numbers and statistics. There is a growing movement within the crypto and blockchain community that is striving to actively use a scientific approach to creating useful crypto investment strategies.
There is a buzz in the tech community with Artificial Intelligence and the Internet of Things becoming fully functional realities. Both Data Science and Blockchain Technology play an important role in this new digital revolution. There is a growing intersection between Data Science and Blockchain Technology, especially in the use of the former to conceptualize certain elements pertaining to cryptocurrencies.
The crypto market faces high volatility and price fluctuations with a lot of speculative activity. Price charts over time show a lot of sharp peaks and troughs indicative of an unstable price environment. Analysts are using the opportunity to use a data-driven approach to make sense of the cryptocurrency market. The following are some of the data-driven approaches to crypto investments.
Bitcoin Price Prediction
Python plays a big role in the field of Data Science, Big Data, and Machine Learning. Using basic Python scripts, analysts are able to develop algorithms that study the price of Bitcoin and predict the future price of the cryptocurrency. Bitcoin is the most valuable cryptocurrency in the market and it is no surprise that it is a focus of crypto investment data analysts. A large number of historical Bitcoin price data pulled from numerous exchanges is utilized for this analysis.
Depending on the programming skill of the analysts, they can choose to use an Anaconda interface due to the high level of dependencies that are involved in the study. Anaconda is one of the more popular Python data analysis tools. When all the price data has been collected and coalesced into a single dataframe, the behavior of the Bitcoin is then studied. The next step involves a probabilistic extrapolation that is used to predict the future price of Bitcoin. This same procedure can be used for Ethereum and other altcoins.
Altcoin Price Correlation
With more than 1,300 different altcoins present in the market, price correlation is an important parameter when developing cryptocurrency investment strategies. Price correlation refers to the relationship between the prices of two commodities in the same market. It is a useful tool for predictive analysis models that are frequently used in investment analysis. Price correlation answers the question, “is there any relationship between these two assets?” By studying how the price of one asset behaves vis-à-vis another, the analyst can determine the relationship between the two assets.
Investors like to diversify their portfolio across different asset classes and the crypto market is no exception. Large hedge funds are beginning to enter the crypto market, spreading considerable investment sums across a number of altcoins. This increases the level of speculation on these altcoins and causes their prices to behave differently. With price correlation data, analysts can begin to pinpoint which altcoin pairs behave alike in terms of price volatility. Using a number of computer programs and algorithms capable of handling big data, these price correlations can be obtained. The effect is that similar investments are being seen in specific altcoin pairs as crypto price correlation data is becoming more refined.
Using social data to predict consumer behavior
For the conventional monetary system, using big data to predict consumer behavior could be tricky. Most financial instruments depend on various factors which make it difficult to predict the direction in which the market will move. However, for the cryptocurrency, this is not the case because demand depends solely on supply.
Big data gathered from social media profiles, especially twitter can gauge the market sentiment by reflecting a clear picture of people’s feelings towards the current state of the cryptocurrency market, latest events which concern cryptocurrency, etc. The demographics related to social media and cryptocurrency trade are the same, the trade relies on individuals more than on large companies, and all the events that can affect cryptocurrency are predominantly first, and in the largest scale published on social networks.
No matter how secure a system may be, it can still leak information or suffer a hack attack. Cryptocurrencies, especially Bitcoin, are very secure and provide a limited amount of public data; however, with the rising tide of data-based hacking and quantum computer technology, the risk of losing all your hard-earned cryptocurrency is very real. In order to identify the potential leaks and security hazards, security analysts use big data analysis so they could improve the overall safety and prevent theft.
The Road Ahead
It would be extremely premature to say that these data-driven approaches to crypto investments have successfully been able to understand the market. A lot of it is still theoretical with a great deal of testing and finetuning required. Analysts are still coming to grips with certain nuances that are unique to the crypto market and creating contingencies for such in their programming models. Also centralised exchange platform poses challenges for programming and majority of the daily trade takes place on centralised exchanges. If successful, these scientific approaches to crypto investment could help investors become more successful in the market.
Using these technical Indicators becomes all the more important as Bitcoin is a currency and not a company with balance sheet and other financials to reflect on future performance.
About the Author:
Ambika Sachdeva is an MBA student from IIM Lucknow. A commerce graduate who has worked in the financial sector and any new developments in the area triggers a curiosity in her. She is passionate about expressing her thoughts via various mediums.
Data Science meets Cryptocurrency Trading — more than Just Friends was originally published in The Capital on Medium, where people are continuing the conversation by highlighting and responding to this story.
Post fetched from this article