Original article can be found here (source): artificial intelligence
Artificial intelligence and systematic trading strategies are natural bedfellows, and many investment managers of machine-based strategies have begun to incorporate AI into their models in recent years. AI evangelists claim that the technology – or, rather, the intelligence – will provide an edge in the never ending quest for alpha.
One of these firms is a French investment technology company, Ai Square Connect, which has developed an AI-based systematic trading strategy, Ai For Capital, that sits on hedge fund hosting provider RCube’s fund platform.
The program trades futures contracts on all the major exchanges on an intra-day basis and closes out all of the positions before the end of the trading day.
AI Square Connect’s CEO, Béatrice Guez, explains why the firm chose to focus on intra-day strategies. “First of all, predictions are more reliable in the short term than the long term. Secondly, our strategies capture intra-day opportunities that are de-facto de-correlated from long term market trends. Lastly, closing our positions every night avoids any gap risk and offers daily liquidity to investors”, she said.
Compared to traditional CTAs, Ai Square Connect applies a Machine Learning overlay to identify good trades and filter bad trades based on features that are different and independent from technical indicators used in the base strategies that the firm has developed; adding this overlay has led to performance improvements of between 20 to 30%. The overlay considers a range of inputs such as the time of day, presence of economic figures (non-farm payroll, ISM, Fed Interest Rate decision, etc.), surprise, the fear index, volume, volatility, central bank meetings and more, according to Guez.
“In February, most managers suffered due to the volatility in global markets, but we have made gains. We feel that this reflects the robustness of our strategy in difficult market conditions”
“In February, most managers suffered due to the volatility in global markets, but we have made gains. We feel that this reflects the robustness of our strategy in difficult market conditions”, she said.
As with all systematic investment programs, managed futures or otherwise, the program is in a perpetual state of refinement.
“We’re looking at new strategies to capture trends. Up to now, our strategies, which are mainly based on mean-reversion, were waiting for a rebound and were not capturing the bullish trend of 2019. Our new long and short strategy brings a strong diversification within our portfolio and minimizes its volatility.”, said Guez.
The product – which is available exclusively in a managed account format – was launched to outside investors in the summer of 2019 and clients can select their target leverage which is designed to deliver between 5 and 15% target returns. The firm is already seeing solid interest, according to Guez.
“The base strategy targets returns of 4-5% with volatility of 4%, so a Sharpe ratio of a little more than 1. Most of our clients who came on at launch were high net worth people coming from the founders’ network. In the past few months, we’ve seen increased interest from corporate treasury departments, who want to use their cash reserves more effectively, and larger family offices in Europe”, she said.
RCube, the owner of the regulatory platform which AI for Capital’s program sits on, is registered with the French regulator the Autorité des Marchés Financiers and with the National Futures Association in the United States, providing access to clients the world’s largest economy.
“Americans are of course very familiar with the CTA model and we think our program will appeal to investors there. It was important when we were selecting a partner that they had access to the U.S. It’s an important market for us going forward” says Guez.
Another avenue that Ai Square Connect is exploring is supplying their machine learning technology to other investment managers. They are currently working with two firms which trade equities and bonds on a proof of concept, which will ideally lead to these managers – and potentially others – adopting the technology into their models.
“Our technology and intelligence helps these other managers analyse data faster and more efficiently. Also, it helps them to remove the emotional biases from human decision making. It’s in the early stages but could turn into another revenue stream for us”, Guez said.
“As our AI continues to learn from all of the trading that we’re doing, we expect that our hit ratio will increase. That’s the beauty of machine learning. We really think it is the missing piece of the puzzle for systematic investment programs”
The rest of 2020 will see AI For Capital continue to grow its AUM both by generating alpha and adding new investor clients and continue to tweak the model by adding new strategies and refining the existing ones. Guez says that she expects the machine learning overlay to continue to be an alpha generator for the program.
“As our AI continues to learn from all of the trading that we’re doing, we expect that our hit ratio will increase. That’s the beauty of machine learning. We really think it is the missing piece of the puzzle for systematic investment programs.”