JPMorgan AI Model Sees Stocks Down With Yields Over Next Month

(Bloomberg) -- The S&P 500 Index and yields on 10-year Treasuries will be lower in one month's time, according to a JPMorgan Chase & Co. model that uses artificial intelligence.

The firm applies an algorithm with an aim to predict the direction of asset prices using five inputs that trace data back to 2006, including: fund flows, economic momentum indicators and investor positioning. JPMorgan's strategy team, who built the model with the firm's AI group, says it has a success rate of between 75% and 89% over one, three and six-month periods. After July, it expects stocks and yields to rise.

“Our model currently implies a bearish outlook for equities over the next 1-3 months with a ‘down' signal for horizons of one month, two months and three months,” the strategists led by Nikolaos Panigirtzoglou wrote in a note June 28. “The S&P 500 signal for the six-month horizon is bullish however, suggesting that any near term correction in equities would revert before year end.”

Here's how the team assesses the success of the so-called Random Forest algorithm at predicting the direction of the S&P 500:

Separately, the bank has a year-end target of 3,000 for the S&P 500, which would be a gain of about 1.2% from the index's level of 2,965.10 at 10:35 a.m. on Monday in New York. The median forecast among strategists tracked by Bloomberg is for a level of 2,950 by the end of 2019.

The yield on benchmark Treasuries will decline over the next month, the model shows, before rising over two-to-six-month time frames, according to the report.

--With assistance from Rita Nazareth.

To contact the reporter on this story: Joanna Ossinger in Singapore at jossinger@bloomberg.net

To contact the editors responsible for this story: Christopher Anstey at canstey@bloomberg.net, Adam Haigh, Ravil Shirodkar

©2019 Bloomberg L.P.