Competition to build faster and smarter algorithms with new technologies to come out on top in buy-side algo wheels has surged among banks and brokers, with those not keeping up at risk of being left behind.
According to a report from TABB Group, artificial intelligence has become a ‘must-have’ in sell-side algorithms to compete in algo wheels, although effective use of the technology requires large investment in big data, quants, storage capacity and computing power.
“Consolidation in the algo services space is expected, just as we are seeing in the sell-side industry in general. In the next few years, only a select few sell-side firms will be able to keep up with the technology required to stay ahead in the survival of the fittest race for best execution,” the report said, authored by Michael Mollemans, senior equity analyst at TABB Group.
“Sell-side firms that have greater order flow with which to test and learn from AI models have a big advantage over small- and mid-size firms. There is a big opportunity for large sell-side firms to create a trading performance gap with investments in AI technology.”
In terms of challenges for the sell-side developing AI algorithms, a poll found that the biggest problem for a majority of 84% is out of sample predictability, which relates to building a model based on sample data.
The second biggest challenge according to 76% of sell-side participants polled is explaining the AI decisions to buy-side counterparts and clients. Sales traders told TABB Group that often the AI algorithm can be too complex to understand, with 32% adding that it is unrealistic to expect sales traders to explain AI algorithms.
The most common application of AI in sell-side equity algorithms, the report added, is to calculate the optimal child slice scheduling and volume prediction, particularly for closing auction. Using AI in venue routing decisions is a more recent trend, although TABB Group said it appears to lead to more promising performance results.