1. What a Quant Does

A quant’s job is to design and implement mathematical models for finance, including derivative pricing, risk valuation, and predicting market behavior.

2. Types of Quants

• Desk quant, who develops pricing models used directly by traders. The advantage is being close to the money and the opportunities encountered in trading. The disadvantage is very high pressure.
• Model validating quant, who independently develops pricing models, but for the purpose of confirming the correctness of the models the desk quant built. The advantage is that it is more relaxed, with less pressure. The disadvantage is that such a group tends to be less influential and sits far from the money.
• Research quant, who tries to invent new pricing formulas and models, and sometimes carries out blue-sky research (not quite sure what that is). The advantage is that it is more interesting (for those who like such work), and you learn a lot. The disadvantage is that sometimes it is hard to prove that you exist (like a scientist — without major results, nobody notices you).
• Quant developer, which is really a glorified programmer, but the pay is quite good and it is easy to find a job. This kind of work varies a lot. It may mean writing code all the time, or debugging other people’s large systems.
• Statistical arbitrage quant, who looks for patterns of automated trading systems (that is, arbitrage systems) in data. This technique is very different from derivative pricing, and it is mainly used in hedge funds. Moreover, the rewards of this position are extremely volatile.
• Capital quant, who builds banks’ credit and capital models. Compared with work related to derivative pricing, it is less appealing, but with the arrival of the Basel II banking accord it has become increasingly important. You get decent pay (though not a lot), less pressure, and fewer working hours.

3. How to Become a Competent Quant

As a competent quant, you cannot do without any one of the three: mathematics, programming, and finance. Mathematics is the thinking, finance is the principles, and programming is the means; the union of the three is what we call financial engineering.

Mathematics: Quants split into two major schools: P quant and Q quant. Q refers to the risk-neutral measure. Risk-neutral mainly means that historical data cannot help you predict future movements, so your decisions carry no risk compensation. The models obtained this way can have beautiful mathematical properties, and can yield certain conclusions even when data is scarce. The mathematical techniques involved are mainly stochastic processes, partial differential equations, and the like. P refers to the real probability measure. “Real” here mainly means that the probability distribution the model relies on is estimated from historical data. This approach depends mainly on data, and the larger the dataset, the better the estimation. The techniques involved are mainly time series (ARIMA, GARCH, and the like), Bayesian methods, and currently popular methods such as machine learning.

Book recommendations: Stein’s Fourier Analysis, Complex Analysis, Real Analysis, and Functional Analysis. Kai Lai Chung’s Elementary Probability Theory. Steven Shreve’s Stochastic Calculus for Finance. Øksendal’s Stochastic Differential Equations. Duffie’s Dynamic Asset Pricing Model.

Finance: basic financial engineering knowledge — pricing derivatives, designing arbitrage strategies, risk management, and so on.

Book recommendations: John Hull’s Options, Futures, and Other Derivatives. Exotic Options, An Option Greeks Primer, Active Portfolio Management. Bodie’s Investments. Paul Glasserman’s Monte Carlo Methods in Financial Engineering. Malkiel’s A Random Walk Down Wall Street. Curtis Faith’s The Way of the Turtle. Robert Pardo’s The Evaluation and Optimization of Trading Strategies. Ernest Chan’s Quantitative Trading: How to Build Your Own Algorithmic Trading Business.

Programming: basic data structures and algorithms — the principles of and differences between data structures such as arrays, linked lists, and hash tables; being able to implement some basic search and sort algorithms yourself; being able to correctly estimate a program’s running time and the memory and other resources it needs; and being able to analyze performance bottlenecks on your own when they appear. You should have a basic understanding of some inter-process communication mechanisms, such as files, sockets, or shared memory; this will let you combine different tools (for example Excel and C++) to achieve complex functionality, and often these small combinations get twice the result with half the effort. As for languages, I recommend learning Python and C++.

Book recommendations: any book on the Python and C++ languages and data processing will do. Python Data Analysis Primer: A NumPy Learning Guide, Python for Data Analysis.

4. The Outlook for Quants

China’s market lags far behind foreign markets in high-frequency trading, hedging, market openness, and other respects, and many of the quant’s traditional or currently cutting-edge fields cannot be applied here. The large number of companies stockpiling quants is in fact stockpiling research results that cannot yet bring their power into play.

In the process of a market shifting from retail-investor dominance to institutional dominance, the rise of quantitative strategies and quantitative risk control is an inevitable trend.

At present, quantitative trading accounts for roughly 20% of the trading volume in China’s domestic financial market (a relatively early figure), and it is increasing every year. Judging by the situation abroad, 70% of financial market trading volume is carried out by programmatic trading, while domestically it has only just begun.

Quant positions are concentrated mainly in investment banks, hedge funds, commercial banks, and financial institutions. Overall the work is relatively demanding, and the pay is much higher than in other industries. Take the Quant Developer as an example: although the actual work is not essentially different from that of a programmer in another industry, not only is the pay high, but it is also very easy to find a job.