Nearly a quarter (24%) of buy-side trading desks plan to implement internal AI technologies for trade execution in the next 12 months, with 15% already doing so, according to a new report from Crisil Coalition Greenwich.
The Great expectations for AI in equity trading study focuses specifically on internally developed AI, excluding third-party tools provided by brokers and vendors. As desks seek to integrate AI into their workflows, the balance between innovation and risk has become a key consideration.
Buy-side traders are placing high expectations on AI. A significant 78% of respondents believe AI will have its greatest impact on real-time algorithm optimization, while 61% point to venue selection and 50% to broker and strategy selection.
But some industry experts caution that these expectations may be premature, especially given the challenges of applying AI in such complex, high-stakes environments.
“Optimizing trading algorithms in real time is the Holy Grail of AI for buy-side traders,” said Jesse Forster, Senior Analyst at Crisil Coalition Greenwich Market Structure & Technology and author of Great expectations for AI in equity trading. “But it is also a risky proposition requiring a significant investment of time and money, and one that only a handful of firms are equipped to handle.”
Indeed, few firms currently have the infrastructure, expertise, or data readiness required to deploy AI for real-time decisions. While using AI to assist with broker or strategy selection is a more manageable starting point—with early mistakes easier to correct—the leap to live optimization introduces significant operational and reputational risk. There’s also the challenge of accountability: unlike traditional rule-based algorithms, AI models often operate as “black boxes,” making decisions that are not easily explainable. For regulators and clients alike, “AI told me to do it” will not be an acceptable answer.
Data remains a central obstacle. Many buy-side desks lack the internal trading data needed to effectively train AI models. The report notes that third-party aggregators like BTON Financial, which pool anonymized data across firms, may be the most viable option for all but the largest desks. While this can help address scale limitations, it also raises questions around data ownership and competitive differentiation
Still, AI may change how traders think about strategy. Some believe that in an AI-enabled environment, the focus will shift away from selecting specific algos—like arrival price or liquidity-seeking—and toward outcome-based execution. AI could guide routing based on real-time market conditions and trade urgency, allowing desks to focus on results rather than the method. As one systematic head put it, traders may soon care “less about which type of algo they use and more about the overall outcome.”
AI’s impact may be more immediately felt in operational areas. The study found 33% of respondents expect AI to influence compliance and surveillance, while 25% see potential in clearing and settlements. These areas are lower risk, and AI can streamline processes, reduce manual errors, and cut costs. Likewise, while only 17% expect AI to enhance post-trade transaction cost analysis (TCA), vendors are developing tools that use AI to better attribute execution outcomes. “It will allow us to better understand the narrative the data is trying to tell,” noted one provider.
Still, the excitement around AI is also being fueled by external pressure. Some desks feel compelled to adopt AI simply because it’s expected by clients or leadership. “The CIO needs an AI stamp,” remarked one trader. Yet not all buy-side firms see a clear return on such investment. As one head of trading noted, AI may be “a timesaver for the buy side and income generator for the sell side,” pointing to the different business models and incentives driving adoption.
Cost dynamics may soon shift as well. Many AI tools are currently subsidized by venture capital and private equity funding, offered at little to no cost. But several professionals warn this won’t last. “What was once free may soon come with a cost,” said one expert, suggesting firms reliant on low-cost solutions should begin reevaluating their budgets and strategies.
Despite the complexities, the long-term outlook remains promising. “With careful planning, investment and collaboration, we believe that AI can play a transformative role in equity trading execution, enabling buy-side traders to make more informed decisions and achieve better outcomes,” Forster concluded.
