Recent graduates, postgraduates and those in early-career positions with a technical background are now faced with a wide choice of exciting and well-compensated career paths in a diverse set of industries.
Quantitative finance remains an attractive option but the competition for top talent is growing from technology firms outside of the financial industry. The chance to "have an impact", be associated with firms that are household names and still receive attractive compensation is a strong pull for technical talent.
Investment managers globally are experiencing significant headwinds. After years of sub-par returns, institutional and retail investors are questioning the value of active management and shifting funds to low fee passive products and managers. In parallel, growing regulatory obligations consume ever increasing buy-side headcount and budget. After eight years of zero interest rates and overly restrictive investment policies that limited exposure to risk assets, insurers and pension schemes are struggling to close record mismatches between assets and liabilities.
The low rate regime driven by central bank induced liquidity is often blamed for the one-way markets in equities and rates since 2009, rendering many stat-arb strategies unusable, and making consistently profitable stock picking far more difficult. Significant reductions in the ability of hedge funds to generate consistent alpha has resulted in record fund closures and shrinking AUM as institutions and high net-worth investors reduce allocations to alternative asset classes.
Interestingly, buy-side head count is at an all-time high, largely to staff compliance departments saddled with an increasing array of national and supra-national regulations such as MiFID II. With mounting compliance staffing costs and fines for non-compliance, a shrinking budget is left to fund quant-driven alpha discovery initiatives.
This backdrop, and the growing prevalence of Masters in Financial Engineering courses, is causing some readers to question whether the market for quant/FE talent is becoming saturated. In this article we take a look at machine and deep learning opportunities in industries outside of finance and capital markets, and alternatives to working as a quant or financial engineer while leveraging existing expertise.
Quants and engineers also find well-compensated opportunities in the booming financial technology sector, developing and deploying applications and services for investors, institutional and retail investment managers, and banks. FinTech firms are increasingly eclipsing their asset management clients in media coverage and perceived "coolness" factor from an employee perspective.
BlackRock, the world's largest asset manager, recently made the provocative announcement that their Aladdin risk platform will account for 30% of revenues within five years, from 7% currently. Blackrock is unique in that it is both an asset manager and technology provider. The announcement is a tacit admission that asset management fees will continue to be pressured.
FinTech firms are working on a number of important and challenging use cases involving machine learning, cognitive computing and deep learning:
The stock prices of Facebook, Amazon, Netflix and Google have enjoyed spectacular outperformance in recent years, leading to the "FANG" meme. Not surprisingly, these four firms are also leading employers of data scientists and machine & deep learning researchers. The "researcher" title is accurate, though limiting. Team members are also required to collaborate and support implementation, validation, and deployment of their applications.
Quants and engineers accustomed to the interview process at a fund or asset management firm stand well prepared for interviews at FANG and similar organizations. Anecdotal salary and compensation data suggest rough equivalence with jobs in Finance and Capital Markets, based on experience and education. Bonuses, stock options and employee ownership plans considerably muddle compensation comparisons. Hours are long, and explain why the FANG campuses provide so many amenities. Quants leaving behind 13 hour Wall Street days might find themselves working even longer, with multiple on-campus sleepovers.
Launch of the USD 1M Netflix Prize in 2006 introduced machine learning to a mass audience, even before "Big Data" became a buzzword. Recommendation engines on Netflix, Amazon and other B2C sites drive increased sales, and improve the user experience for online consumers making complex purchases. This is also an area of well-funded and intensive research. As recommendations become increasingly aligned with, and predictive of, human behavior the "autonomous shopper" is becoming a reality.
Amazon's career page for machine learning mentions three key initiatives:
The 265 open positions on Amazon's Machine Learning Science career page underscores the shortfall in talent required to meet growing demand for machine learning and data science expertise. Quants and FE's looking to shift industries have significant FANG-based employment opportunities.
Google's TensorFlow, released in 2015, has become a popular Python environment for researchers, engineers, and developers building applications that incorporate AI and machine learning. Google recently released TensorFlowLite to help developers deploy small-footprint versions of their AI-based applications in mobile environments.
Facebook Research encompasses Applied Machine Learning, Data Science, Natural Language Processing & Speech, and Computer Vision. Facebook Artificial Intelligence Researchers (FAIR) articulate the audacious goal of developing systems with "human level intelligence". Emerging applications include real-time translation of conversations between individuals with no understanding of each other's language, and algorithms capable of recognizing a static image such as a face in a short snippet of video.
For quants and financial engineers contemplating a position as an AI, machine or deep learning researcher, a number of industries are applying machine and deep learning to solving systemically important problems. Breakthroughs in disease detection and diagnosis, improved allocation of scarce resources, and anticipating and minimizing risk will drive a golden age of innovation across multiple industries.
Ideal For: Those with a strong statistical machine learning background with extensive programming and software development skills. Very suitable for computer science graduates/researchers and those willing to relocate to the major tech hubs such as Silicon Valley, Seattle, LA and NYC.
Jet engines, wind turbines, delivery vehicles, and assembly lines produce massive quantities of time-stamped, geo-coded sensor and machine data. Predictive asset maintenance, forecasting supply chain disruptions, and optimizing manufacturing processes are leading use cases. Due mostly to physical limitations of sensors, this data tends to be noisy, requiring expertise in time series smoothing and validation, and synchronizing multiple sensor data feeds before machine learning can begin. Building a multi-exchange real time order book is a highly useful quant project akin to dealing with sensor data series.
Once drone-based package delivery services gain regulatory approval, a wave of data will be created and captured for machine and deep learning algorithms to leverage. Generating efficient drone delivery paths, informed by real time weather forecasts and altitudinal traffic conditions are an active area of research. Closely related, autonomous vehicle research, prototyping and testing is receiving significant R&D budget from Google, Apple and Tesla, car hire services Uber and Lyft, and traditional car companies like Ford and Toyota. Browsing the engineering and data science career pages of these firms underscores the growth potential of autonomous vehicles, whether airborne or ground-based. While many industrial analytics startups exist, IBM and GE enjoy significant market share for this growing class of applications and use cases.
Ideal For: Those with an aeronautical or mechanical engineering background, as well as quants, who have utilised both traditional time series methods and novel machine learning anomaly detection in their work.
Providing cost-effective and evidence-based healthcare to aging populations, predicting the likelihood of hospital readmissions, detecting and preventing the spread of disease vectors, and discovering innovative new drugs and therapies are key areas of research for pharmaceutical companies, non-profit institutions, and universities. Machine learning and cognitive computing are being applied to these and other related problems. With healthcare consuming a growing amount of government budgets, the potential payoffs for researchers and data scientists are significant.
Image analysis is a prevalent machine learning application in medical diagnostics, drug discovery, and clinical trials. Image analysis that distinguishes between benign and malignant tumors, or techniques that quantify bone and tissue growth resulting from an experimental drug treatment are two example use cases. Experience with geo-coded data is also helpful. Example applications include forecasting the spread of a virus by geographic coordinates and understanding proximate environmental factors related to "cancer clusters".
Machine learning is having a major impact on genomics as well, allowing scientists to move beyond merely finding correlations between genetic patterns and diseases. These insights establish causal interpretations of genetic variation that help clinicians diagnose conditions with greater speed and confidence, and enable molecular biologists to design highly targeted gene-based therapies for a growing number of diseases.
Ideal For: Those with a scientific background (physics, chemistry, biochemistry) that have applied machine learning to their research problems. Also applicable to applied mathematicians, statisticians and computer scientists who want to directly apply their theoretical knowledge to an extremely important, and intellectually rewarding domain.
Detecting and deterring existential threats is arguably the ultimate machine and deep learning challenge. Use cases include detecting ballistic missile launches, differentiating actual missiles from decoys, and forecasting adverse weather, and eventually seismic events.
Like healthcare, image analysis is a key skill, in this case applied to satellite imagery and CCTV video to identify faces and detect threatening objects. Overlapping with industrial applications, this chemical sensors mounted at the entrance to tunnels and other vulnerable infrastructure produce massive volumes of data that must be analyzed in real-time to detect needle-in-a-haystack anomalies. Understandably, this is the most opaque and secretive class of machine learning use cases and IP. National defense agencies and their contractors are employers of quantitative and engineering talent, as are commercial ventures such as Palantir.
Ideal For: Quants/engineers with a background in object recognition, such as those working with funds utilising alternative data, e.g. satellite feeds to model global oil supply and demand.
As intelligent systems make cities, governments, organizations and corporations more efficient, risk tolerant and innovative; machine learning and AI will deliver a golden era of innovation across multiple industries. Quantitative finance likewise has much room for innovation and discovery. That doesn't necessarily entail getting a traditional bank, hedge fund or institutional asset manager job. Opportunities abound for quants and engineers open to exploring other industries and use cases.comments powered by Disqus
You'll get instant access to a free 10-part email course packed with hints and tips to help you get started in quantitative trading!
Every week I'll send you a wrap of all activity on QuantStart so you'll never miss a post again.
Real, actionable quant trading tips with no nonsense.