Tuesday, July 27, 2021
HomeThe CapitalTime Really Matters: How Do Hedge Funds Depend on Data Science?

Time Really Matters: How Do Hedge Funds Depend on Data Science?

If a person does not know the future, can artificial intelligence predict where the market will move?

Photo by Andy Beales on Unsplash

The world’s twenty most successful hedge funds raised $63.5 billion in 2020, setting a decade’s record in volatility as tech stocks rebounded sharply after the pandemic-fueled sell-off, LCH Investments reports.

“The net gains generated by the top 20 managers for their investors of $63.5 billion were the highest in a decade. In that sense, 2020 was the year of the hedge fund,”

– Rick Sopher, LCH’s chairman, said in a statement.

The main goal is to counterbalance the fall of some assets with the growth of others, and in order to achieve it, it is necessary to use a well-thought-out strategy and excellently predict future trends. Technological innovation can just become a tool that will make these processes more accurate and efficient.

Risk-adjusted and time-sensible

If hedge funds had to choose between lower but stable profitability and a possibly greater return on a roller coaster ride, then most of them will opt to the first option.

For example, the “stable non-risky 200% per year” strategy will be more advantageous in relation to the strategy like “500% per year with a greater risk of going into losses”. This is because hedge funds are financially responsible for the investors’ money.

High returns alone are not enough, you need high risk-adjusted returns. Most of the developers of algorithmic trading bots in hedge funds start building the software with that axiom in mind.

If you feel confident in market analysis or are ready to study financial tips, then you can try to create your own algorithmic trading bot. There are specialized platforms on the market for creating trading robots — like Thinkorswim (ThinkorSwim Group, Inc.) or MetaTrader 5 (MetaQuotes Software Corp.). Users can either write bots or advisors by themselves or download / rent / buy any ready-made applications.

Most sophisticated trading bots work with 3 moving parts:

[signal generator] -> [risk allocation] -> [execution]

… so the biggest challenge is how creative and insightful your signal selection and processing strategy will be.

Trading bots have strongly taken root in hedge funds. If you ask why, the answer is time. In the world of finance, even milliseconds can play a role.

A brilliant example is the case of Spread Networks, which built a direct fiber optics cable between the Chicago Mercantile Exchange and the data center of the NASDAQ exchange in New Jersey. This was necessary in order to receive data a fraction of a second earlier than other market participants.

Algorithmic trading bots operate magnitudes faster than a human’s thinking time plus reaction time. Most computer algorithms used by hedge funds to trade mimic what human traders do, just more systematically, faster and cheaper.

Using artificial intelligence

BarclayHedge Survey: “Which part of your investment process is driven by an application of machine learning techniques?”

Priority technologies for hedge funds include artificial intelligence and machine learning. A 2018 study by BarclayHedge found that more than half of market players use these methods to make investment decisions and two-thirds to generate trading ideas and optimize portfolios.

This interest in artificial intelligence is due to the need for hedge funds to quickly calculate trends, look for signals in the news and predict the price of a particular stock going to change. The more accurate the forecasts are, the lower the risks and the higher the profit. A person does not know the future, but artificial intelligence can predict where the market will move, based on historical data and changing external factors.

Artificial intelligence can analyze much larger amounts of data than humans, and its processing speed is much higher. The more data analyzed, the more accurate the forecast will be. Thus, the AI ​​fulfills the tasks that the data scientist has set for it.

A data scientist initially builds a model based on data. Once the model is created, it is tested on the events of the past — this process is called backtesting. For example, there is a model that allows you to make a forecast for tomorrow with a certain behavior of stock prices. This forecast forms the basis for the actions of the hedge fund — and shows whether to buy or sell specific stocks. To assess the accuracy of the forecast, historical data is applied into the model.

Theoretically, you can build models without such tests, but for hedge funds, risks are no less significant than profitability (remember that hedge funds manage investors’ capital in such a way as to increase it with minimal risks).

So backtesting is the best option. During backtesting, we will not incur financial losses by buying shares in accordance with the erroneous forecast, but we will check whether the prediction comes true. For example, you can enter data into the system that was collected ten days ago, make a forecast for the day ahead, and compare it with how stock prices performed nine days ago, thus looking into the future.

Strategy Tester — pic credit: MetaQuotes Software Corp.

Most importantly, AI predictions still need to be verified by humans. Whether it is done in a random order or only the most unexpected predictions are tested, the best option is up to you.

What does a data scientist do at hedge funds?

The challenges that data science can solve in hedge funds are vast. The data analyst selects information, categorizes it, and yet there can be a huge variety of influencing factors — macroeconomic and political events, industry trends, the performance of a single company, its competitors, etc. Remember everyone is talking about the growing volumes of big data — data lake and even data ocean.

Data Lake Meme 🙂 Pic credit: Meme-Arsenal

It is important to understand that the hedge fund industry, and the financial industry in general, values ​​the effectiveness of strategies rather than complexity. You should always try the simplest models first and consider whether you need to complicate them. It is also important to understand the fundamental economic underpinnings of a particular idea. So the use of artificial intelligence is not necessary, but if there are indicators that can be significantly improved using ML methods, technology should be implemented.

Popular Articles