How Jim Simons and a legion of mathematicians are solving the market
Updated: May 21
Simple statistical models form the fundamentals of performing stock predictions. This is how the best do it.
Renaissance Technologies LLC, an American hedge fund founded by Jim Simons, a Harvard/ Massachusetts Institute of Technology professor and former Cold War code breaker, has been averaging a 66% annual return before fees, since 1998. These unprecedented results have brought the title of “the most successful hedge fund in the world” to the company. Surprisingly, unlike traditional banks that heavily emphasis on fundamental analysis, Renaissance Tech’s strategies are completely based on quantitative analyses from complicated math and statistical models. Programming’s power in realising mathematically-based strategies have shaken the stock market, leaving many wondering if they could develop in-house trading algorithms hoping that one day it will bring them a fortune.
Stock prediction using data analytics revolves around three main steps:
extracting primary indicators to perform tests on, finding the proper models for testing, and retrieving historical data for backtesting.
Prior to performing tests on different models, key indicators regarding the stock market should be extracted for computers to learn from the underlying patterns. Some popular indicators include previous days’ open/ close and maximum/ minimum prices. These figures, while straightforward, can be used to compute meaningful statistics such as average returns, volatility, and EMA (Exponential Moving Average), which may play crucial roles in the process of stock prediction.
With a set of indicators potentially related to the stock prices in the near future, one may then implement models to approximate the correlations in between. The most frequently adopted model is linear regression, which assumes a linear relationship between these indicators and the futures to find the corresponding weights. More complex models such as RNNs (Recurrent Neural Networks) and LSTMs (Long Short-Term Memories), however, are also being used in recent years as hardware computations catch up significantly. These are families of the neural network methods, where the previous series of outputs are also utilized to compute the next prediction. In other words, for time-series data like stocks, indicators of previous dates can be gathered as inputs to assist in predicting today’s price.
Ultimately, with well-structured models, the process of backtesting, a term used in modeling to refer to testing a predictive model on historical data, assumes an indispensable role to enhance traders’ and analysts’ confidence. The potential risks and profitability of a trading strategy could then be identified before putting any real capital at risk.
However, while previous literature has featured plausible models for stock market prediction, stock prices are volatile in nature and have a high correlation with unknown factors (i.e., inflation, the presence of disrupting technology, economic growth). Even with state-of-the-art models and methods, it is highly unlikely to generate a model with accurate predictions at all times.
That being said, quantitative analysis has still shown to be effective for short-term predictions, bringing profitable returns to investors. The trend has not only ignited a spark of excitement in Silicon Valley, driving engineers to step out of their comfort zone for a “gold rush”, but also brings about new challenges and opportunities to the labour market. Leading banks in the U.S. include J.P. Morgan, Goldman Sachs, and Morgan Stanley have adopted recruitment strategies that are aligned to those quantitative trading firms to secure the most tech-savvy employees around the world. According to a BCG report, this may be one of the most urgent procedures to get hold of the market shares from hedge funds, mutual funds, and other capital market players.
Despite the exponentially increasing demand, the selection process of quant firms is still exceptionally difficult. Having perhaps one of the most concentrated numbers of award winners in competitive mathematics and computer science, advanced degrees in such fields are almost a must in applying for these firms, not to mention the high requirements on mental numerical calculations and ability to analyse data in a timely yet creative manner. Without a relevant degree, one must prepare substantially via job training and experiences as data analysts to stand chances in the mathematically-gifted applicant pool.
For a comprehensive overview of quantitative analysis in markets, feel free to check out “The Man Who Solved the Market: How Jim Simons Launched the Quant Revolution”.
The UCL Finance and Technology Review (UCL FTR) is the official publication of the UCL FinTech Society. We aim to publish opinions from the student body and industry experts with accuracy and journalistic integrity. While every care is taken to ensure that the information posted on this publication is correct, UCL FTR can accept no liability for any consequential loss or damage arising as a result of using the information printed. Opinions expressed in individual articles do not necessarily represent the views of the editorial team, society, Students’ Union UCL or University College London. This applies to all content posted on the UCL FTR website and related social media pages.