AI Funds Face Dearth of Historical Data
Investment funds that rely on artificial intelligence and machine learning have run out of usable data for their learning processes, according to one asset manager.
“In finance, if you look at the data for the market close for the past 20 years, there are not enough data points,” said Ankit Awasthi, quantitive portfolio manager at Qplum, during the recent Fall 2018 ETP Forum in Midtown Manhattan. “Whereas if you look at natural language processing and image recognition, they have millions of data points. There is a fundamental constraint in how you can learn using these domains.”
To address the dearth of data, Qplum has repurposed high-frequency trading data to simulate the millions of transactions that were not included in the closing data of the past 20 to 30 years.
“There is a new trend happening in the industry called ‘synthetic data,’ which is the creation of time series used to test your models,” added panel moderator Bartt Kellerman, founder of Battle of the Quants. “It is moving to the very futuristic type of technology where data is lacking.”
Awasthi also noted that unlike other industry verticals implementing AI and machine learning, financial services tend to be rather myopic and asset manager will shut down AI- and machine learning-based funds after a single quarter of poor performance.
“There is no long-term vision,” he said. “There is all this computing power and all sort of data available. This is the future.”
However, not all AI and machine learning funds are equal. Many funds that reported to be AI- and machine learning-based often rely on linear regression, according to Awasthi.
“It is like adding ‘.com’ after your name back in the day,” said Kellerman.
Quantitative investors may have no choice whether to integrate synthetic data into their calculation given the evolving equities markets.
“It is not like the same strategies will work all the time,” said Awasthi. “Deep learning and machine learning, in general, offer flexible frameworks to detect these inefficiencies in a systematic fashion that is adopted over time.”
“As we have seen in CTAs and HFT, those sorts of markets have left us,” agreed Kellerman. “The opportunities for big money are no longer there.”
Artificial intelligence and machine learning, powered by cloud, is moving into mainstream use.
Data extraction and integration is the second stage of a digitization process.
Increased electronification has created useable and accessible real-time and historic trade data.
The fintech uses data so institutions can assess the environment impact of their portfolios.
Working with Riskfuel will reduce reliance on slow and expensive financial models.