Not only traders but also mathematicians and programmers work with stock markets. Director of Engineering at Luxoft Artem Sosulnikov tells about data, which specialists of quantitative hedge funds work with, things they pay attention to, and conditions in such companies.
What Is a Quant Hedge Fund?
Quant hedge funds differ from classic ones in terms of financial analysis. Such companies use a quantitative (or quant) investment approach involving data science, statistical methods, and machine learning. In classic hedge funds, a trader analyzes facts, evaluates indicators, communicates with company representatives, and studies public information. Almost the same is done by algorithms: they look for market inefficiencies by processing large amounts of data in a short time, monitor any mentions of companies in the media, social networks or even in public speeches and use this info in complicated algorithms. Another advantage of the quantitative approach is that automated systems do not gamble — they do what the algorithm says and do not have a behavioural factor, which reduces risks. In quant funds, the main work lies on the shoulders of data scientists, mathematicians, and developers, not portfolio managers or traders.
Quantitative funds have been growing over the past decade, holding the highest share of trading volume on US stock exchanges.
According to Barclay’s data for 2017, capital flows to quantitative funds commanded 29%, or $114 billion, of the total amount transferred to hedge funds. Financial firms continue to pile into quant strategies of investment, with some creating separate quant divisions.
How Do Quant Funds Work?
Quant funds create so-called „signals“ (alphas) based on different sources for making decisions on buying or selling shares.
In addition to classic market data, prices and trading volumes, such sources can be:
Phone geolocation data, based on which the attendance of chain coffee shops and fast-food cafes is monitored to predict increases or decreases in revenues and, subsequently, the stock price of the companies.
Satellite images for evaluating the number of cars parked near chain stores or cafes to predict sales and stock prices.
Another example: searching for all the oil tankers in the images, using the length of the shadow on the water and the time of the image, one can estimate the draft of the ship and, accordingly, the quantity of oil in the tanker, which provides data on the transportation of oil by sea around the world. Now empty tankers are usually filled with random amounts of water, so this signal is not applicable anymore but, for a while, it has been on target.
The calendar of national holidays in countries where certain precious metals are traditionally given on specific days helps to analyze their value, taking into account the dates.
Transactions of card providers (VISA, MasterCard, etc.). Trillions of transactions (if not more) are parsed with machine learning to summarize transactions of all companies represented on exchanges. This data gives access to approximate accumulated revenues so the fund may understand the situation in each particular company before quarterly reports are published.
There may be more than 10 million such signals for analytics in the fund, which means terabytes of input data daily. Most often, the signals are developed by researchers, quant engineers with expertise in finance and algorithms.
And What About Programmers?
IT specialists create an infrastructure for processing, converting, and downloading data, develop and test trading applications. Therefore, hedge funds are looking for seniors with knowledge of Python, Java, C++, expertise in data science, machine learning, cloud, data lakes, and big data. Technology stack: Kafka, Cassandra, Spark, Django, Tornado, etc. Not the least of these is an upper intermediate level of English.
The funds hire high-class pros, experts. Many mathematicians are working there, so you need to be well-versed in algorithms, probability theory, and number theory.
An experienced coder in a quant fund takes the place of an experienced trader: a perfect algorithm can gain an advantage for the fund in the market. According to the Wall Street Journal, hedge funds battle against Silicon Valley for the very best mathematicians and programmers. Hedge funds try different ways to attract candidates and offer more benefits than tech companies: they give as many vacations as necessary, offer instant bonuses to those whose work has brought profit, organize hackathons, open quantitative research laboratories for Oxford students.
What to Do: History Data, NLP, Reporting
Different teams work on different projects. For example, it could be:
building data lakes and big data storage using HDFS, Spark, and Java;
writing data managers in C++ to convert data into one format, and developing plugins to convert data from different formats into a single one;
writing a history-based market behaviour simulator, a software package, which simulates the entire market. You can load a strategy into it, and it will go through the market at a rapid pace, while the fund’s specialists will analyze the results: whether the strategy advises buying or selling something correctly;
configuring the infrastructure using AWS, K8S, and Docker virtualization and clustering;
building distributed systems based on Kafka and Cassandra;
solving data science problems using NLP (natural language processing), NumPy, Pandas, NLTK, and Python;
infrastructure projects related to revenue calculation, reporting to regulators, internal reporting in Django, Tornado, and Flask.
A Little Bit About Working Conditions
Globality
Hedge funds hunt for developers (who isn’t?), so geography is not limited to the country of presence. Besides the U. S., where most of the funds are located, they are looking for talent in Asia, Russia, and Europe.
Remote and Almost Remote
Due to the peculiarities of information and financial security, there were practically no remote specialists in the projects of quant funds for a long time. The pandemic has made its adjustments: now it is possible to find remote or temporary remote offers. Some companies prefer a hybrid format of „flexible share desk“ where some days the employee works from home and others spend in the office.
Salaries and Bonuses
A senior engineer in a quant fund can get wages comparable to European ones. Some companies give additional bonuses. For example, a major American hedge fund we are working with has set up a prize fund for our employees. It is about $100,000 a year for six champions and six top performers, who will receive $10,000 and $5,000 respectively.
How to Get In?
The easiest option is to monitor the vacancies of specific companies; another one is to get a job in a partner company of a quant fund. But please, keep in mind that they select a minimum percentage of the received CV’s.
If you are interested in working in quant funds but aren’t experienced enough yet, you can try internships. They are most often paid but require coming to another country. The Two Sigma and Citadel founds, for example, have such offers, with training in the United States and China. You can also get training online on a data science module.
Pros:
respectable wages, 20-25% above the market average;
real big data: up to 10 petabytes of new data per day;
modern technologies: funds immediately apply new language standards.
Cons:
often you will have to rewrite or even completely replace the modules you have written;
you need to be competent not only in IT but also in finance;
don’t forget that the stock market is risky.
Reading
Heard on the Street: Quantitative Questions from Wall Street Job Interview
Cracking the Coding Interview by Gayle Laakmann McDowell
The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Trevor Hastie, Robert Tibshirani, Jerome Friedman
Advances in Financial Machine Learning by Marcos Lopez de Prado
High-Frequency Trading. A Practical Guide to Algorithmic Strategies and Trading Systems by Irene Aldridge
Learn more about the Luxoft Quantum Fund project and open positions here.