Written by Jane Clarkson
As artificial intelligence (AI) weaves its way into the fabric of fintech, firms and retail investors alike are re-evaluating investment opportunities that were once considered too labor-intensive or complex—areas now ripe for innovation. Bhavdeep Sethi, the founding architect behind Frec‘s direct indexing algorithm and former Twitter adtech lead, provides a deep dive into how these technologies are reshaping investment strategies and discusses their broader implications for the future of investing.
Passive Investing Optimization
One path into passive investing, direct indexing has been revitalized by machine learning (ML). “It’s an excellent use case. The amount of tracking and portfolio tuning once needed a dedicated advisor,” says Sethi. Traditionally, direct indexing presented high barriers to entry, making it a resource-heavy option reserved for high-net-worth investors. Now, with each stock’s cost basis assessed in real-time, identifying opportunities for tax-loss harvesting (TLH) and executing trades is accessible to any retail investor. “The granularity of asset allocation means more room to personalize your portfolio,” Sethi adds. “These algorithms can be adjusted to an investor’s unique financial goals, whether that’s ESG or a focus on digital assets. You aren’t locked into an ETF or any of its potential downsides.”
Deep learning mechanisms are also being applied to tax regulation, ensuring compliance while optimizing trades, with criteria-based rules such as avoiding wash sales. Sethi envisions a future where these technologies can tailor tax strategies to each investor’s income, risk level, and investment window. “Continuous learning and adaptation have always driven quantitative investing,” he explains. “The difference now is that you no longer need to be a large firm to access it.” This new depth of portfolio adjustment options enables investors to maximize returns while minimizing liabilities.
Algorithmic Trading
Over 50% of US. stock market trading volume is estimated to come from high frequency trading (HFT), also known as algorithmic trading. Since the introduction of commercial computation, these “algos” have been executing trades based on minute market inefficiencies. “For example,” Sethi explains, “HFT is used to identify arbitrage opportunities—where a security is priced differently in two markets—and execute trades across them within milliseconds.” This speed is essential in HFT, where delays mean the difference between profit and loss, and firms have gone as far as to relocate closer to trading centers. As the technology advances, the speed and efficiency of these algorithms continues to grow. “The other side of the coin here,” Sethi adds, “is that any investor not leveraging algorithmic trading is at a disadvantage, especially as the line between passive and active investing blurs.”
Beyond just increasing speed in fintech, language models can process unstructured data from various sources, such as breaking news articles, posts on social media, and economic reports. By applying natural language processing (NLP) techniques, it becomes possible to gauge market sentiment and predict potential price movements, enabling traders—or their algorithmic proxies—to position themselves advantageously. “This has always raised concerns about fairness in trading,” Sethi notes, “so companies that are willing to democratize the technology will find significant support among retail investors.”
Risk Management
Traditional stress tests apply broad market shocks to entire portfolios, yielding generalized results that might not reflect the nuances of a personalized portfolio. With the increased complexity offered by deep learning, it’s possible to incorporate sector-specific risks, geopolitical events, company-level data, and more. “AI allows us to simulate highly specific conditions,” says Sethi. “We can model how a sudden interest rate hike, for example, impacts different portfolios, giving investors a much better sense of how they would be individually affected.”
AI is also enhancing risk management through real-time volatility monitoring and automated risk mitigation. Labor-intensive periodic reviews and adjustments once left portfolios exposed during rapid market changes. “Today’s models can spot volatility spikes before they even fully materialize,” Sethi explains. This allows the system to automatically adjust portfolio positions, reducing exposure to high-risk assets, and reallocating to defensive alternatives. For instance, if unusual price movements are detected, the algorithm can increase cash holdings or shift investments to less volatile sectors. “Or, it can just provide you with enough information to make the decision yourself,” adds Sethi.
The Future of AI/ML in Investing
If history is any indication, AI and machine learning will continue to make their way into every facet of investing. Fintech leaders have hailed them as the ultimate scaling mechanism, evident in their ability to revolutionize everything from personalized portfolio optimization to risk management. Bhavdeep Sethi agrees but emphasizes that the democratization of these tools is key to the future of investing. “As we make these technologies more accessible, they’ll attract more investors and foster a healthier market. A rising tide lifts all boats.”
