Hedge fund inside SEB has model that called the Trump bond rout

This hedge fund has “the model” that foresaw last year's bond rout.

Obeying their computers, the five-person global “quant team” at SEB AB in Stockholm anticipated last year's bond sell-off and went to “short” from “long” in U.S. bonds even before Donald Trump's election to U.S. president accelerated a surge in yields.

“Our bond models reacted rather strongly to the election of Trump,” Hans-Olov Bornemann, head of the global quant team at SEB Investment Management, said in a phone interview Monday. “The re-positioning accelerated when the election results were published.”

Mr. Trump's victory stunned global debt markets and sent the Bloomberg Barclays Global Aggregate Total Return Index down 4% in November, the deepest slump since the gauge's inception in 1990. The yield on the 10-year benchmark U.S. Treasury has jumped about 100 basis points in the past six months, in part driven by speculation that Mr. Trump will unleash fiscal stimulus and cut taxes.

Mr. Bornemann, who manages €1.7 billion ($1.8 billion) in the SEB Asset Selection hedge funds, is still short bonds. SEB Asset Selection had an annualized return of 5.4% since the start in 2006.

The fund's purpose is to increase the risk-adjusted return of customers as the “ultimate complement” to a portfolio. The quant team, which includes a former nuclear engineer, relies completely on a model by adopting a trend-following strategy.

“We just call it 'the model,'” Mr. Bornemann said. “We don't want to deal with subjective things. Human beings -- including ourselves -- shouldn't be making qualitative and emotional investment decisions.”

The model calculates daily the probability for equities, bonds and currencies to rise or fall on average on a three-month horizon. The calculation for each asset is based on technical, fundamental and behavioral variables such as price, volume, valuation, growth and investor positions. The optimal portfolio is then composed and implemented.

Since the fund started, the team has only intervened manually once -- in December 2008.

“It was an extreme situation during the financial crisis,” he said. “We thought it was prudent to cut the risk by 50% in a proportionate manner across all positions.”

A couple of weeks later the model had cut the risk itself to half so the team removed the restriction. All the other days the team has followed the buy and sell orders from the model, Mr. Bornemann said.

In addition to being short bonds, the fund is shorting interbank rates. It's long the U.S. dollar, except against the Australian and New Zealand currencies, and long stock markets, except India and South Africa, due to “increasing optimism and expectations of higher earnings growth.”

“One can discuss whether it's cheap or not,” Mr. Bornemann said. “If an investor thinks that bonds look expensive and are going in the wrong direction while stocks may not be that cheap, but still a better choice. There are hopes for rising prices. In the relative play between asset classes, stocks are more popular.”

Mr. Bornemann, a graduate in finance from the Stockholm School of Economics, has headed the five-man quant team since 2003. While Mikael Daeckfors was responsible for quantitative risk analysis of the proprietary book of the Wallenberg family's Investor AB, Matthias Eriksson worked as a nuclear power engineer at ABB Atom, before joining SEB.

Even though the model took years to develop completely, Mr. Bornemann now only has to sit back and let the computers run for about 30 minutes every day. The fund makes money 54.4% of the days and loses money 45.6% of the days, he said.

“That's pretty good,” Mr. Bornemann said. “If you do that over time you make good money over the long term.”