Beach Horizon soars on pattern-spotting machine

David Beach's 25 years' experience with pattern-spotting techniques has been replicated by a computer in Beach Horizon's office

David Beach, veteran pattern-spotter and chief investment officer of hedge fund Beach Horizon since 2005 and founder of Beach Capital Management in 1998, started his first investment fund at Sabre Fund Management in 1989.

He has been profiting from his ability to spot patterns in financial markets for more than 25 years. While many portfolio managers trade on fair value or price momentum, Beach remembers patterns. A commodity that goes up, down, up and up may be more likely to skyrocket. A commodity that goes up, down, up a little bit and down a notch may be more likely to collapse.

Beach Horizon holds positions in precious metals, base metals, energy, meats, grains, softs such as orange juice and cotton, equities, rates, and developed and emerging markets currencies.

How many patterns are there? "I can tell you that but I might have to kill you afterwards," Beach laughs. He then admits there are more than 10 but less than 20 basic patterns, with many more subcategories. "There aren't that many that you need to remember them," he says.

He began using his technique to manage money at Sabre. Later on he founded his own discretionary trading programme at Beach Capital Management, a managed futures firm. He retired from managing the programme in 2006, having generated average annual returns of more than 20% over the period. He launched Beach Horizon's systematic trading programme with Sanjeev Lakhanpal and Paul Netherwood in 2005.

Beach has now been replaced by a large black machine, which sits in a glass cage in the hedge fund's London offices (pictured). Beach Horizon automated a version of his pattern recognition techniques in 2013, a process that has been no mean feat.

Netherwood, a partner at Beach Horizon, explains the problem sitting in his office. He draws a line that zigzags up and down with similar magnitudes, and a line that zigzags up and down with diminishing magnitudes. Is the first line a different pattern to the second? He draws a line with zigzags midway between the two. What class is the third pattern and how does a machine differentiate between them?

The "fuzzy" nature of this problem – what philosophers call the Sorites paradox: when does a heap become a non-heap? – has made it hard to automate Beach's pattern-spotting techniques for over two decades. Beach says he had little time to do so in his early trading career and the automation process is not yet over.

But it has already started paying dividends. Last year, Beach Horizon had one of its best performances since 2008, returning 41.7% to investors. This year, it was shortlisted in Hedge Funds Review's European Single Manager Awards for hedge fund of the year. This was in part due to accretive conditions for commodity trading advisers (CTAs) as a group but also to research innovation, Beach Horizon's team says.

Societe Generale's Newedge CTA Index, an equal-weighted benchmark drawn from the largest open managers, rose 15.7% in 2014.

Performance has since levelled off, however. Beach Horizon returned 0.15% in the year to July 2015 as CTA strategies in general entered a losing streak.

But what lead to the algorithmic breakthrough that allowed man to be replaced with machine? First, Beach Horizon's researchers succeeded in eliminating noise, which tended to confuse any attempt to detect signals.

"There are a certain number of patterns that reoccur," says Beach. "Ultimately you're trying to apply them to the medium term to capture a significant trend lasting several months. It's not a simple process. Some of the patterns have got nine, 10, 11 turning points over a series of many weeks."

He describes the pattern mechanisation as "aimed at reducing the impact of failing trends and the end of trends". Beach Horizon's success in pinpointing the signals for turning points has reduced its downside volatility. As the trend reverses, the CTA is better placed to retain profits, the thinking goes.

Lakhanpal, also a partner at Beach Horizon, points out other CTAs have been trying to diversify their portfolios since the financial crisis - in terms of the time horizons of trading portfolios and improving risk management - but "this doesn't tackle the problem of abrupt trend reversal at its core".

A further benefit to the proprietary technology is to better detect signals in sideways markets, when there is no overpowering upward or downward trend. "An indicator to predict sideways markets is the Holy Grail in the trend-following world," says Lakhanpal. "However, we've been able to make significant inroads into the problem through this technology."