Few still believe that equity markets are efficient. But no-one thinks they are that inefficient, either. Martin Steward speaks to four asset managers who believe they have a proprietary information edge over the run-of-the-mill research process. Two exploit the wisdom of individuals; and two the wisdom of crowds

Sandy Rattray
Head of Man Systematic Strategies

Analysing the analysts
Between September 2009 and May 2011, the team that would become Man Systematic Strategies when Man Group acquired GLG Partners in 2010 published three papers on broker analyst recommendations. ‘Do European Brokers add any value through recommendations?' (Sept 2009) found that, indeed, they do, but also that buy recommendations outperform more consistently than sells, and that GLG's selected universe of European broker salespeople outperformed the average European analyst. ‘When European analysts disagree, who should you pay attention to?' (Feb 2010) found that large brokers' non-consensus recommendations have a higher impact on stock prices than those from smaller houses. And ‘Do Analyst Experience, Location and Gender Affect the Performance of Broker Recommendations in Europe?' (May 2011) found that experienced analysts produce better buys and juniors produce better sells and (in contrast to other recent research) that there was no performance difference between male and female analysts.

This barely scratches the surface of the insights the team has developed over six years of cataloguing analyst recommendations and exploiting them via systematic strategies. But it covers the big ones, and shows how counter-intuitive the facts can be. Do most portfolio managers think big brokers exert more influence? Probably. But do they think buys consistently outperform sells? That's not commonly-accepted wisdom. Do they even think analysts' recommendations work at all? No.

The interesting thing is that GLG itself had to go into these strategies with an open mind to pick up much of this. How did it find out about the buys-versus-sells anomaly? Its first fund to exploit its database was a long/short strategy. A year later an investor who was restricted from buying hedge funds asked for a long-only version - and that long-only strategy started outperforming the hedge fund. After discovering this skewed performance, GLG allowed the hedge fund to wind down, but also started reporting back to its broker salespeople a breakdown of buy and sell recommendation performance, rather than a combined number.

"That was a simple technological development, but it had quite significant behavioural guidance," says Sandy Rattray, head of MSS. "It not only signalled that we wanted more buy than sell ideas, but suggested that the contributors would do better if they focused on buy ideas for us."

Like many other portfolio managers, GLG's team included skeptics who doubted that its preferred group of salespeople delivered a better selection of research than the average. The academic evidence, after all, suggested that outperformance was a mirage.

"There was an internal debate: ‘Why are we bothering? Why don't we run this strategy globally just using IBES, say?'" Rattray recalls. "But by the end of the debate the importance of having your own data was clear."

The anomaly in most of the academic papers on this subject was the vast universes of stocks in the analysts databases being used for the studies. As a result, these studies had a huge skew towards micro and small-cap recommendations that are useless to large institutional investors and consequently never figure in the recommendations proffered by the salespeople that GLG uses.

When GLG narrowed the universe of recommendations down to STOXX Europe 600 names, and also corrected for size bias in the stock price returns, the outperformance of GLG's preferred salespeople against that universe was obvious. Later research into the difference that experience makes - "Our contributors were the ones sent to talk with Pierre Lagrange when he was here," says Rattray, "and he wasn't talking to people with one year under their belts" - suggested one of the reasons.

Why doesn't everybody do this? It wouldn't necessarily take seven years for a competitor to build a comparable database, Rattray concedes, but it does take a considerable commission pool and a full-time effort to keep brokers generating their ideas, entering them into Man GLG's customised web portal - and submitting themselves to rigorous measurement. As the research has revealed, you really do get what you pay for.

"Ultimately, a good salesperson doesn't talk to everybody - that's the difference between him and an analyst," says Rattray. "If you are a big player in the market, as we are, you can get good coverage: good coverage helps, and not all salespeople are created equal. It might be an unpopular thing to say, but it's a fact."

Athanasios Ladopoulos
Partner and senior fund manager
Swiss Investment Managers

Following the insiders
During a career ranging from the Abu Dhabi Investment Authority (ADIA) to the event-driven hedge fund Kinetic Special Situations and Xanthos Funds (where he led a management buyout), Athanasios Ladopoulos developed a deep understanding of industry spin-offs.

"When a successful new venture gets spun-out of its mother company you might not think that directors buying a significant amount of shares in the mother company would be an indication that the spun-out entity would do well," says Ladopoulos, a student of the father of directors' dealing analysis, George Muzea. "However, it almost always is. Why? When the spin-out is announced they get shares in the spun-out entity proportionate to their holding in the mother company."

It's just one example of the insights that directors dealing habits can offer into future company performance, which Ladopoulos now exploits in the Directors Dealings fund, the first offering from Swiss Investment Managers and the first hedge fund dedicated to the strategy.

"There is no stronger indicator than someone at the helm of a company putting their money where their mouth is," he says. "If a director buys $2m of stock, that might be 10 years' gross salary on the table. Now imagine that two directors out of five do the same thing."

But beyond strong performance, Ladopoulos also makes big claims for the strategy's diversification benefits. After all, it has nothing to do with the traditional bottom-up or valuation analysis applied by what Muzea called "the trivial many". "We don't care about traditional valuation," says Ladopoulos. "We only care about the interpretation, by a senior director - one of Muzea's ‘vital few' - of some future event."

But hold on. Aren't lists of directors' deals regularly published for the benefit of amateur stockpickers? Sure, but these lists are crude.

"Blind trading of directors' deals has zero value, in any sector, in any company," says Ladopoulos. "Of the 1,000 trades per week across Europe and the US, we consider 990 to be noise. From the other 10 we execute perhaps three. Over our first 11 months we have taken about 52 positions, of which 44 remain live."

This summer illustrated the danger of buying director's deal ‘beta': July saw frenzied activity, even in the banking sector, but following that would have led to significant pain in August and September. "You need to understand the difference between ‘value' directors, who buy as prices decline, and ‘catalytic' directors, who buy in anticipation of a surprise," says Ladopoulos. "The universe is 70% value directors, but activity from the catalytic directors can indicate more reliability in the value directors' signals."

The only way to know the difference - and, indeed, whether or not directors make good investment decisions, period - is to analyse their track record. This is Swiss Investment Managers' edge: a huge database of directors' deals, reaching back as far as 10 years and two previous companies for some individuals, overlaid by a proprietary ranking system. "No two directors are the same," says Ladopoulos.

The fund takes long and short positions, all of which are conviction trades, rather than pairs or hedges. Over the summer the fund was about 65% net-long, but still managed a 2.4% return in August. Of course, not every position is a success. Even if the director is ultimately right, he might have chosen the wrong entry point: if the shares don't move substantially after six or seven months, the fund will generally ease out. At the other end of the spectrum, a stock might outperform expectations. Either way, profit-taking and stop-losses are determined according to the change in the risk of each position, rather than any valuation metric or price target.

The strategy targets 15% annualised returns with 10% volatility and, even including a stormy August, it has so far achieved about 25% with 6% volatility. The aim is to grow the portfolio to about 200-300 positions per year as the firm takes its model beyond Europe into the US. So far just one fund is offered, with zero leverage, but another, geared version is in the pipeline.

Johannes Jacobi
Grassroots Research Analyst

Rooting out returns
If you want to know how the new Reebok shoes will sell, where do you turn? To the head of sales at Adidas? To your retail analyst? Those insights will help. But if you want raw data, as opposed to hopes or educated guesses, and you want them ahead of the market, why not put an army of clipboard-carriers outside sports shops, ticking a box every time a pair of ‘EasyTones' goes out the door?

One asset manager does precisely this, systematically. RCM saw the light nearly 30 years ago, after the market lost millions on one of the entertainment industry's most infamous flops. Based on the hit movie and riding the crest of the video-gaming wave, Atari's ‘ET: the Extra-Terrestrial' was set to take Christmas 1982 by storm - until it didn't. Sales were so dire they inspired an urban legend (apparently millions of copies are buried somewhere in New Mexico).

"The whole fund management industry was to blame for not asking consumers if they were interested," says RCM's Johannes Jacobi. "That simple question could have picked up this gap between expectations and actual sales."

Jacobi is an analyst in RCM's 11-strong Grassroots Research division, which commands the resources of 72 freelance journalists, around 250 ‘Field Force investigators' and a database of more than 50,000 industry experts, turning out hundreds of company or industry studies each year for its portfolio managers. Its latest quarterly newsletter carries insights into topics ranging from the outlook for car sales in China and hospital equipment spending in Europe to Allergan's Botox treatment for chronic migraine.
The introduction above suggests that Grassroots exploits the wisdom of crowds rather than individuals. But while an important part of its work is done through crowds - of consumers, and of the students, professional market researchers and retirees in the ‘Field Force' teams who survey them - individual experts are equally valued.

These include the broad range of industry experts in the Grassroots database. "We evaluate every person we have spoken to," says Jacobi. "How helpful have they been, how successful was the report that resulted after we spoke with them, where are their specific areas of expertise?"

Then there are the experts - from doctors to shop assistants - who are selling or using products at the coalface. He singles out a report on Smiths Medical, part of the Smiths Group conglomerate that provides low-tech hospital supplies like hypodermic needles. Reporters were sent out to ask customers a handful of simple questions, the key ones being about quality and pricing and barriers to changing suppliers. Most said that pricing and quality were fine, but some added that these were not the key issues: the most problematic thing for hospitals was the 4-6 weeks wait for delivery.

"Great feedback," recalls Jacobi. "The reporters rephrased the entire worksheet and subsequently discovered that Smiths had a big distribution problem all over Europe. Portfolio managers adjusted for that new information and a few weeks later Smiths put out a statement on th problem which resulted in a negative price reaction. Sometimes it's just half a sentence from a medical practitioner or the like, but it can be significant for understanding potential market dynamics. You can't get at these things in meetings with a pharma company's IR rep or CFO."

At the other extreme were those Reeboks. When the ‘EasyTone' range came out, Grassroots interviewed staff at leading sportswear retailers in Germany and the US (in June 2010) and the UK (in July). "The feedback was overwhelming - shoes were selling out within days of arriving," Jacobi recalls. "Reebok, bought by Adidas six years ago, had gone through a restructuring and was looking for a success story - and the market had no idea about these sales figures." In August 2010 RCM's analyst changed his vote to a strong buy off the back of the Grassroots reports, and Adidas shares rose by about 30% over the next four months.

Grassroots has its strengths and weaknesses. It works well for the consumer, technology and healthcare sectors, but much less well in the more tight-lipped world of financials. And there is a perennial debate within RCM around whether to focus research on stocks or sectors about which portfolio managers do not have a clear opinion (but which are therefore smaller active risks), or those about which they have the strongest opinions (and are therefore larger active risks).

"We have to use our resources wisely," says Jacob. "But putting it into the context of overall asset management costs, there's a long way to go before we find ourselves spending too much on producing too many reports."


Simon Hawtin
Derwent Capital Markets

Micro blogs, macro profits
On 10 October, 2011 @lalyydrew tweeted @justinbieber: "♥♥♥♥♥♥♥♥ TE AMOOOOOOO 5 DAY TO CHILE♥♥ AND #7DAYS TO MISTLETOE". On the same day, @DowTimingAl tweeted, "Dow Soars 250 Amid Euro Zone Hopes: DJIA soared, led by BofA"; and @MiszEclipse advised us that "The best way to eat lucky charms is eat the marshmellows [sic] last".

Which of these carries relevant information if you are trying to decide whether to go long or short the FTSE100? That's right - all three!

That, at least, is the contention of brothers Paul and Simon Hawtin, co-founders of Derwent Capital Markets, who, until autumn 2010, had been running managed accounts for family and friends in a fairly orthodox-looking quantitative index-trading strategy developed by Paul in his Saxo Bank days.

Since then, they have been working with a data-mining system that takes a completely random 10% cut of each day's 200m-plus Twitter posts, analyses key words and phrases, and comes up with a reading of global sentiment expressed as six different ‘moods' - ‘calm', ‘sure', ‘happy', ‘kind', ‘alert' and ‘anxious'. Shifts in this sentiment are used to trigger long or short CFD positions in FTSE100 futures, S&P500 futures, Dow Jones futures and some large-cap equities.

"It's completely random," confirms Simon Hawtin. "We've looked at focusing on financial words, and even taking data exclusively from people who are working in finance, but we find that skews the results to the point of having zero predictive power."

It sounds crazy, but the idea is backed up by research, published in the Journal of Computational Science in October 2010 by Johan Bollen, Huina Mao and Xiao-Jun Zeng: ‘Twitter mood predicts the stock market'. The article created a stir on financial news outlets and caught the eye of the Hawtin brothers, along with many other hedge funds who wanted to get in touch with the authors.

"To be honest, we thought that ours was a bit of a long shot," says Hawtin. But it is a small world - even before social media made it smaller. Xiao-Jun had once worked for Knowledge Support Systems, which also employs the Hawtins' half-brother. That contact got them on the phone to Bollen at the University of Indiana, and the brothers soon discovered that their Christmas break in Barbados coincided with Bollen's own holiday in St Lucia. "Over a few beers, looking out over the ocean, we agreed to collaborate," Hawtin recalls.

Exclusively licensing Bollen's research and system, Derwent Capital's strategy awaits signals strong enough to make its 15-25 trades per month, with 1-7-day time horizons, based on trends in sentiment. "A one-day uptick in mood of 1.5 standard deviations would make us go long, with stop losses and limit orders informed by some technical analysis," says Hawtin. "We're still getting to grips with how stuff about Justin Bieber and what-I-had-for-breakfast impacts financial markets - but the results speak for themselves."

He recalls the Tohoku earthquake of 11 March. Unsurprisingly, the mood on Twitter darkened. More surprising was the magnitude of the shift - five standard deviations on Saturday, 12 March. "The typical lag of the market response is 1-7 days, but the sharper the shift, the shorter the lag," says Hawtin. "The signal to short came on the Saturday, we executed on Monday, and the market really came off."

More typically, Hawtin claims that the system predicts up or down movement in the DJIA, usually three days ahead, with 87% accuracy. It is less effective at predicting magnitude, which is the team's current area of research.

It is early days, but there are clearly big question marks. Even if we accept that a random selection of tweets reflect the global ‘mood', and there is some causal link with buying or selling in equity markets, it seems difficult to argue that ‘fear and greed' are always pre-eminent in market returns. However, the period during which Twitter has become a meaningful phenomenon has coincided with a three or four-year period when markets have, indeed, been driven by sentiment and the macro picture.

Derwent Capital is a long way from establishing the operational infrastructure to accept institutional mandates. Perhaps by the time it has, it will show a persuasive track record. In the meantime, your decision about whether to invest may depend on your mood. Feeling ‘kind'? Maybe.