Recently, Blackrock announced plans to replace a small portion of its portfolio managers with machines to pick stocks. Ever since Vanguard popularized passive investing, where index funds track the market instead of actively picking and trading stocks, human portfolio managers have been under increasing pressure to justify their high management fees. Now, Blackrock is making a case that machines can outperform human judgment in picking stocks and evaluating companies.
For over a decade now, hedge funds like DE Shaw and proprietary trading firms have used algorithms to identify and transact arbitrage opportunities, often within a few milliseconds. While these quantitative methods and trading bots have dominated high frequency trading, human portfolio managers still dominate long-term value-investing, with human judgment, not algorithms, evaluating a company’s fundamentals, management, and environment. Since at least the late 1980s, AI researchers have written about the potential for neural networks and genetic algorithms in replicating and outperforming human judgment in stock-picking. Early papers built neural network models to identify high-return stocks using recent macroeconomic and company financial data. Some researchers also theorized that neural networks could use data on strategic plans, new products, confidence, optimism, and other features to classify firms.
Instead of a precise set of instructions on how to evaluate companies, these neural networks and genetic algorithms organically discover a large number of ‘rules’ that perform well at prediction or classification.
With this academic foundation, Blackrock is replacing 30 people in its active-equity group with machine learning algorithms. While this represents a small portion of its overall assets under management, Blackrock hopes to demonstrate the superiority of an AI-driven stock-picking strategy to both its in-house human portfolio managers, as well as the market. Across the industry, active investors are increasingly looking to machine learning models to supplement or replace their investing process. For instance, Euclidean, a hedge fund, uses Naïve Bayes and Support Vector Machines to predict long-term stock prices. Betterment and Wealthfront, robo-advisers for consumer savings, use machine learning to recommend optimal asset allocation.
While Blackrock has not specified or shared any details of its new product, there are several fascinating directions AI-investing to go in. For instance, portfolio managers currently spend considerable time listening to quarterly earnings calls and attending investor presentations. In addition to obtaining information to update their models, human investors are also evaluating senior management and forming judgments on the veracity, confidence, and optimism of executives. Machine learning models can perform sentiment analysis (e.g., tone classification and facial emotion detection) on public appearances by executives to predict stock performance. Similarly, AI investors can train models on a company’s patent application to predict the impact of future product and technology development on stock prices. In addition to entirely replacing portfolio managers with AI, some of these models could also provide specialized inputs to augment human judgment, potentially beating a only-human and only-machine strategy.
Sources:
https://www.wsj.com/articles/can-robo-advisers-replace-human-financial-advisers-1456715553
https://www.wired.com/2016/01/the-rise-of-the-artificially-intelligent-hedge-fund/
http://www.euclidean.com/unknowable-future
Swales, George S., and Young Yoon. “Applying Artificial Neural Networks to Investment Analysis.” Financial Analysts Journal 48, no. 5 (1992): 78-80. doi:10.2469/faj.v48.n5.78.
Kryzanowski, Lawrence, Michael Galler, and David W. Wright. “Using Artificial Neural Networks to Pick Stocks.” Financial Analysts Journal 49.4 (1993): 21-27. Web.
http://www.nytimes.com/2006/11/24/business/24trading.html
https://news.ycombinator.com/item?id=13990167
By The Terminators (Maayan Aharon, Aanchal Bindal, Aditya Bindal, Youngeun Kim, Eran Lewis, Angela Lin)
=)