# How Algorithmic Trading really works?

In my final year of the part-time MSc Financial Engineering program that is offered by the World Quant University, things are coming together and the full picture of algorithmic trading and alpha generation starts to make sense. The subjects are super challenging relying heavily on computer programming & algorithm, financial markets knowledge, statistics, and econometrics. Even though I had very limited knowledge previous to starting advanced education on these subjects, now I feel much more confident and have a roadmap on a real-world implementation where only hard work and continues practice is required for concrete results.

This is a short article seeking to simply explain what algorithmic trading is, strategies that can be implemented through the use of algorithms and some of the tools and skills that one needs in order to accomplish this trading process.

In simple terms, algorithmic trading is using computer programs that act upon certain criteria specified by the user and execute trades in the financial markets through a broker or an exchange. These programs are represented by algorithms (mathematical representation of a sequence of instructions set on solving a problem coded on a computer). The combination of financial markets and trading knowledge paired with programming skills and computational power can result in the creation of machines that take advantage of market inefficiencies.

Algorithmic Trading is about finding market inefficiencies and taking advantage of them as well as any market anomalies that may be turned profitable. Market Anomalies or market inefficiencies in the financial markets are deviations from the efficient market hypothesis in terms of prices and rates of return.

The Efficient Market Hypothesis (EMH) is a theory in financial economics that states that asset prices fully reflect all available information. What this basically means is that a trader/investor cannot “beat the market” consistently on a risk-adjusted basis as prices already reflect any new information. This hypothesis is central to finding any assets that do not comply with it and are inefficient.

The Random Walk theory is also embedded in the Efficient markets. The Random Walk theory suggests that stock price changes have the same distribution and are independent of each other hence past movements or trends cannot be used for future prediction. When talking about Algorithmic trading we have to always account for these hypotheses as anything out of their scope would be a potential opportunity or trading idea.

Efficient assets can’t be traded due to unpredictable prices. Inefficient assets can be traded due to predictable prices. Inefficient Markets – turbulent economies and large volatility and uncertainty in the market present us with the anomalies/opportunities we need to implement different trading strategies. These anomalies can relate to fundamental, technical or calendar events.

So Market anomalies are the events and trading ideas looked upon as a problem that we wish to solve through the use of our algorithms. Solving these problems on the basis of the criteria we are looking at would result in successful trades returning positive returns. Once you have created your algorithm exploring a trading idea on the basis of market anomaly you can backtest it on the underlying financial data and see if it actually works. On a cautionary note, even though the algorithm at hand might perform perfectly on a backtest, in reality, it might be a completely different story as there are many other factors influencing it such as transaction and brokerage cost, order types, slippage, etc.

There are a number of ways you can trade with algorithmic strategies that bank on a particular aspect of the financial securities’ inefficiency. These strategies buy into to different aspects of the price movements of securities. Below is one of the most popular main categories of strategies explained briefly:

• Momentum Strategies – With momentum strategies we are simplifying following the trend of the price action. Buying into the strength and along the movement of the security and selling into the weakness or the fall movement. These strategies benefit from market sentiment and big directional market moves. Through the use of moving averages crossovers and convergence as well as support and resistance price breakouts strategies can be implemented.
• Mean-Reversion Strategies – They are banking on the strength and weaknesses of the markets on both ends of the investment spectrum. Mean-reversion strategies rely on the price of the security coming back to its true value – buying into weaknesses that are going to adjust and selling into strengths that have been overly increased. Technical indicators such as Bollinger Bands, Relative Strength Index, and Channel indexing can be implemented into trading strategies for this particular market inefficiency.
• Frequency Strategies – when speaking about Algorithmic Trading, one can always relate to HFT or High-Frequency Trading which is a sub-category that is characterized by very short holding periods, low latency response and high trading volumes. Strategies of varying frequency from Low, High to Ultra can have holding periods of milliseconds though require very advanced market knowledge, market structure, and trading infrastructure.

### Skills and Tools

In this final paragraph of the article, we are going to look at some of the main tools and skills one needs to successfully implement algorithmic trading strategies:

• Technical Skills:
• Programming – research, data mining, analysis, simulation, trading idea implementation, machine learning, numerical optimization, big data analysis with Python, R or C++.
• Finance – deep understanding of financial markets, market structure, products, financial theories, modeling and trading strategies.
• Quantitative Analysis/Statistics – knowledge of statistics, econometrics, quantitative methods.
• Risk Management – risk identification, evaluation, and management frameworks.
• Soft Skills:
• Adaptability and ability to work long hours under stress
• Opportunistic
• Comfortable with failure
• Innovative
• Data Sources – there are many free and paid data sources that you will need in your financial analysis. For example:
• Research Platforms – once you have the data for your trading strategy you will need to backtest it to see if it actually works as expected before you plug it in a brokerage account:
• Brokers – there are also many different brokers to choose from whichever your geographical location might be. Some of the best ones are:
• Interactive Brokers – are a good option for implementing algorithmic trading strategies on the retail level. Great API, low fees, and no hidden fees.
• ThinkorSwim by TD Ameritrade – also a good option with great functionality though the fees are higher.

Algorithmic trading gives computers the ability to make buy and sell trades based on sets of rules provided by the trader. It is an area of quantitative finance and trading that is very complex and interesting at the same time. Ultimately, algorithmic trading is a very complex subject matter and seeking further education to excel in this area is crucial. The MFE degree offered by the World Quant University is one which gives you the set of tools required for successful implementation and uses in this field. It is one that I can personally vouch.