Behavioural Finance

New sources of alpha


The search for alpha is about to change direction. New sources of public information offer an edge for the successful stockpicker. Surprisingly, the most useful perspective on information is not that of traditional, standard finance – its concept of market efficiency is challenged by the real world evidence. Although CAPM is a useful framework to have, behavioural finance can provide greater insight for the practice of day-to-day investment.

There is much information that is not captured by traditional analysis – most focuses on what is provided by company management rather than searching more broadly, and analysts tend to focus on numbers not text. Analysts need to think more about what other qualities information has – such as independence or whether incentives lie behind it. Prices may convey more information about a need for liquidity, than value. And, capturing what prices convey about risk may be better done by technical, rather than fundamental, analysis. This is because prices may be capturing more genuine fundamental information than what we normally think of as disciplined Street analysis.

A broad range of metrics may at times be needed instead of stockmarket ratios. Much conventional sell-side analysis focuses on p/e and yield – and on P&L rather than balance sheet. It seems brokers phone up managers almost every day to report that the p/e of Royal Bank is 7x, as if it mattered. This year has seen a breakdown in metrics, and in quant systems. Consensus forecast p/e is one of the less useful analytical tools.

One conclusion is even more challenging for investment professionals. Often public information capturing private insight can be superior to professional opinion. Indeed, there is growing evidence that intuition is an overlooked processing ability, which can be compromised by gathering too much unstructured information. More is not better, however easy it may be to market the concept of a huge research team to clients or consultants.

Information & context

Before looking at the right process for generating alpha, it is useful first to reflect on the nature of information. Investors require that information is public, and assume it is largely independent of context. Indeed, much of financial economics depends on this absolute attribution of information. It is assumed that all will interpret information in the same way, so that security prices tend to an equilibrium, at which they are in some sense ‘right’. But there is too much focus on assuming that prices primarily convey facts on value and risk, reflecting all information. As soon as you recognise that a price may instead be telling us more about liquidity or might not be at an equilibrium level, you lose the CAPM risk framework. Investment professionals cannot have it both ways – believing that prices are wrong enough to allow alpha, but right enough for risk metrics.

CAPM is the prevailing financial markets theory, but there is much of it that is questionable, and not particularly useful in looking for performance. Indeed, why should a theory that tells managers that they cannot outperform be useful in the search for alpha? So, the starting point in this search for alpha is to recognise how much of prevailing wisdom, and many stock prices, might actually be wrong.

This is as serious a failure for risk management and performance analysis as it is for stock research. Much of risk analysis is built round the concept that prices provide information on value and risk, and are typically ‘correct’, reflecting consensus. Yet, while that equilibrium makes for neat maths and helps provide Sharpe Ratios, Value at Risk, multi-factor analysis etc, equilibrium is often inherently unstable. There is good reason to think that equilibrium is rare, and to start from the premise that many prices might instead reflect biases, misunderstandings and misinformation. Some interesting recent academic research is focusing on agency incentives, showing that for funds such as hedge funds, there can be skewed incentives which seem likely to become embedded in asymmetrical market returns. That may explain the skew evident in some stock returns.

That is, the fact that some managers can optimise a fund to collect regular small gains, with the potential for the occasional big win, but possibly also with an even bigger though rarer loss, itself then distorts the demand for securities with particular price patterns or optionality. Until recently, a significant premium was paid for smoothness because it allowed leverage and a greater fee. Yet, market-based risk measures may be fooled by the illusion of consistency. One Bear Stearns fund ran without a single losing month in its 40 month history, before almost wiping out this year. Agency incentives mean people can game against market preferences, benchmarks, or any flawed generally-accepted risk measures.

Also, an equilibrium provides little incentive for stability – it would typically be a point of low turnover and therefore failing to reward traders or those providing capital. If the information on value was correct, it would mean that much investment research becomes a free good, and there would be little alpha opportunity for skilled investment managers. Equilibrium is an interesting concept, but likely to be transient. That is good news for managers searching for alpha.

There is alpha opportunity, but the implications for risk should not be overlooked. Prices may provide information about liquidity needs rather than value, and so the risk metrics such as multi-factor analysis can be wrong. In addition to appraisal of fundamentals, managers should look at what a price or pattern of prices might reveal about the underlying investors, and those who are currently primarily driving the share price.

Perspectives on public information – crowd sourcing

Interesting research is being done in the area of crowd wisdom and crowd sourcing. This is the potential for the wider public to capture and relay information which is otherwise submerged. It is useful to think about the wisdom that might lie within a crowd, and its power or ability to relay that. That is, gather information by assuming that the facts are somewhere public, and then wondering how that public expression might be captured.

The key issues for the value of this information are what knowledge the crowd has, and whether this is aggregated in a way that is not distorted by gaming or social considerations. The public can perform better in areas where time is involved in capturing or evaluating information. Lay people will often perform better in estimating real physical factors rather than the immaterial world. Another key factor is whether the data gathering is genuinely independent. Given a financial or social incentive, even experts can compromise information. For example, forward prices in short Sterling have often been a better indicator of the future course of base rates than surveys of leading economists.

One good, well documented, example of this was with Canadian miner, Goldcorp Inc. Today this business is capitalised at US$25 billion. However, seven years’ ago it was in trouble; besieged by strikes, debts and very high production costs. Analysts assumed that its 150 year old mine in Red Lake, Ontario, was dying. Capitalised then at under $100 million, and without evidence of substantial new gold deposits, Goldcorp seemed unlikely to survive. However, its Chief Executive, frustrated that his in-house geologist couldn’t reliably estimate the value or location of gold on their property, did something unusual. He published his geological data on the web for all to see, offering a total of $575,000 in prize money to those who submitted the best methods and estimates to find 6 million ounces of gold. Every bitof information on the 55,000 acre property was put on the website and more than 1400 virtual prospectors from 50 countries got busy crunching the data. Entries were from geologists, but also remarkably from many other disciplines ranging from intelligence systems to computer graphics. The contestants identified 110 targets on the property, half of which had not previously been identified by the company. More than 80% of the targets yielded substantial quantities of gold. The winning entry was from a collaboration from two groups in Australia, which together developed a powerful 3-D graphical depiction of the mine. Indeed since then, a total of 8 million ounces of gold have been found – worth well over $3 billion.

Northern Rock provides a recent and interesting example of crowd sourcing, and it is interesting to contrast the expert view with the private information that was being made public and finding expression in share price patterns, although not in City research. How many financial analysts looked through all of Northern Rock’s website or its entire filings? In late July, the then Chairman of Northern Rock plc and several directors bought shares at more than four times the price they ended September at. Yet, blogs show what information was actually out there. On the BBC site, one investor wrote “we were thinking of opening a Northern Rock account earlier this year and looked at their website and decided against it. They seemed to be lending any sums to any Tom, Dick or Harry with bad credit ratings. Northern Rock seems to be a chancy business, and we decided not to put our money in it.” The share price pattern was arguably capturing the wisdom out there, and that was more informative than the price itself. There was clearly wisdom lying outside the board that was more useful than Stock Exchange announcements.

Indeed, Northern Rock is a good example of how to assess crowd behaviour. Only a few fundamental analysts on the sell-side got it right. Just one major house described Northern Rock in early September as “the most leveraged bank in Europe, with an assets-to-equity of 58.2 times in 1H07.” Of course Northern Rock was labelled and regulated as a bank, rather than a hedge fund. Although directors bought stock shortly before this, this was possibly not in material amounts relative to their wealth or earnings from the company. Good reason to disregard that. The blog information and website did reveal more, but generally this does not appear in investment research. What credence should we give to the FSA announcement of 14 September that it judged that Northern Rock “has a good quality loan book,”subsequently embellished by the FSA chairman describing Northern Rock’s business model as being based on “high-quality residential mortgages?” And, the technical picture captured in the first quarter a change in share behaviour and a clear correlation with what was going on in the US credit market.

Another interesting piece of information is that Individual Voluntary Arrangement (IVA) practitioners had throughout this year been complaining about the actions of one particular bank in apparently blocking a far higher proportion of IVAs than others, seemingly with a view to forcing debtors into a long term repayment programme, in the belief that the borrower would not risk bankruptcy. Of course that was helpful in not having insolvency recorded against the bank. This even got as far as practitioners making complaints to an MP about this behaviour, believing it to be in breach of two parts of the Banking Code. The bank with this unusually high rejection rate was Northern Rock, and we should consider why this might be, whether any part of this knowledge found its way into the share price and what it might mean for future bad debt experience. Investors should gather information as widely as possible using all the tools available and evaluate that relative to independenceversus incentives, and likely knowledge captured, before rating the value of each piece of information.

Organisations should think of their employees, customers or committees as crowds. Wise crowds need diversity of opinion, independence of members from each other, decentralisation and a good method for aggregating opinions. It is important to create diverse teams and to try to collect information independently, ahead of meetings, if possible. One of the successes of wise crowds and failings of structured groups, is often that private information is withheld in the latter. It is important to find ways to bring this out.

What this means for risk & technical analysis

A misunderstanding of risk is at the root of many of the alpha opportunities available. The dominance of CAPM in risk, via Value at Risk and Sharpe Ratio, pays too much attention to statistical analysis of price history and not enough to base data. Where stocks have dramatically and rapidly changed their behaviour, beta and prior price volatility have been poor indicators of risk. Surprisingly, the best indicator of a shift in risks has been a change in price momentum. This would have neatly caught Northern Rock as well as a number of other sub-prime related stocks. For extended periods, some stocks go up more than they typically go down, and others have sharper falls than previous price volatility would typically indicate. Generating alpha does mean stepping aside from the standard financial model.

Analysing text, not just numbers

The language of company reporting can also provide surprising insight into the nature of current trading and the board’s real view of prospects. Currently, little analysis appears to be conducted on this. Yet, it is possible to be systematic about analysis of text and – taking company reports and other announcements in soft form – search for the incidences of particular words. A key sentence on trading has a different meaning buried within a penultimate paragraph than it does at the beginning of the first. And the company announcing a trading update conveys something different depending on the timing of that in relation to the period end. Confidence that results will beat expectations means something different given after the reporting period, rather than weeks before. Timing and language of statements matter, and they must also be placed in the context of what has gone before.

This can be illustrated with Vanco Plc’s last annual accounts. Most analysts work on the numbers, but the context for the numbers is the work of the audit and remunerations committees. Investors would expect to see compliance with the Combined Code, and regular committee meetings – with any remuneration discussions at least matching up with the number of audit meetings. Bonuses and other incentives usually depend on establishing the numbers first. What investors often miss in company reports is the useful table of committee and board attendance.

The Vanco governance statement in fact mentions that the Code requires at least one member of the audit committee to have recent and relevant financial experience. “The board considers bearing in mind its small size and the overall transparency of its financial statements that the appointment of an additional non-executive director with financial experience … is currently not necessary.” These tables can show whether the executives attended audit meetings and in Vanco’s case we also discover that the Chairman of the Audit and Remuneration Committees was changed after the year end. After the year end the Chairman of the two committees was replaced “due to the standard guidelines of independence of non-executive directors.” The new Chairman then presented the remuneration report in the accounts despite never having attended an audit committee meeting during the year. Although there were seven audit meetings in the year, there was none in the last four months – but one on remuneration. And, the one remarkable event of the year was that all five executive directors sold atotal of almost £20 million worth of shares near the all-time high – after the shares rose tenfold in four years and very shortly before losing almost all of that again. A lot of text provides context for numbers and is capable of systematic and rigorous analysis.

Processing information

Many of the best stock opportunities over the past five years have been in businesses recovering with the global upturn. Typically new management is involved, but there may be challenges of debt, poor margin and other execution risks. Yet, price/earnings ratios are widely used for this type of business, despite failing to reflect the opportunity for margin improvement. Some of the best performing stocks of the past two years, such as Sainsbury, Compass, Invensys and Morrison, did so from prospective p/e multiples in excess of 20 times. In only a few areas is less attention paid by analysts to p/e, one being in pharmaceuticals where pipeline appears the dominant metric. Indeed, so much so, that analysts missed the potential to extract value in other ways, as AstraZeneca showed in 2005 and early 2006.

For some of these recovering companies, there has also been a misunderstanding of execution risk, evidenced in a preponderance of negative research. It is important to remember that analysts can signal views, recommendations, price targets or earnings forecasts. Only change in the last of these is any real guide to underlying confidence, and this can capture a lot of additional information from analysts – not just the earnings outlook. Typically, research pays little attention to balance sheet issues, which is particularly important in the current credit environment.

Looking for companies with bad long term price momentum, execution plans that are given little credence, and which are typically valued on market metrics such as p/e, or where research is predominantly negative, is fruitful for generating alpha. However, this alpha may not emerge evenly, as there is some evidence with these stocks that performance comes in jumps – as favourable news, or simply lack of any further problems, forces short sellers to capitulate or encourages former investors to return. Many of these shares also exhibit under-ownership by mainstream UK institutional investors. The unusual way in which alpha can emerge can lead it to be dismissed as price momentum or a misunderstanding of beta.


I have focused on information, but I do not believe that more is necessarily better. There is growing evidence that intuition can play a useful role in processing information, and should usually not be disregarded. Often, where repeated decision-making is needed, fast and frugal heuristics can perform better. Six to eight pieces of information may be all that is needed for this. More data than that is usually geared to explaining the past, rather than predicting the future. An expert delivering ten or more reasons for something, as with many economic predictions, citing too many pieces of information, can make for an interesting interview or fill up a research document, but the predictions will probably be anchored in the recent past. While it seems to cut against professional rigour, the work that is being done on unconscious processing will prove in time of great value in investment, and find its rightful place in daily practice.

In summary, I believe investors should gather information more widely, and use that to test whether they or the consensus might be wrong. A broader range of metrics should be used. The key is to recognise that the City consensus is not necessarily the same as collective public wisdom, and that it is fruitful to search beyond the Street and set every piece of information into context. With this approach – and questioning what is brought to our attention, rather than the information we search for – investors can deliver consistent alpha.

This article is based on a Hedge Royale Conference presentation, London, 31 October 2007



Colin McLean is Founder and Managing Director of SVM Asset Management, and has been an investor for over 25 years. SVM has run equity long/short hedge portfolios since 1992 and currently managse two hedge funds; SVM Highlander Fund and SVM Saltire Fund. At SVM, he makes use of a number of the findings of behavioural finance in managing investments.