Take the two previous Bearbull columns, I have been discussing two companies - coal supplier Hargreaves Services (HSP) and oil services group Petrofac (PFC) - with a view to deciding which shares will make the better investment for my income portfolio. Yet, arguably, what I really did was construct readable little pieces about these companies that hung together logically, but which considered just a fraction of the relevant factors and did not reach a specific conclusion.
How could they, since the terms of the forecast were not framed tightly enough? They should have said something like this: given that the Bearbull Income Portfolio targets a minimum total return from its investments of 50 per cent within five years, which shares are more likely to achieve that, Hargreaves or Petrofac? That could have provided the framework on to which a decision map of binomial probabilities could be pinned, the answers to which could have led to a plausible conclusion.
The difficulty is that's a formidable and even tedious task. To make a reasonable judgment - that won't necessarily turn out correct anyway - do we really have to list every knowable factor that could affect a company's shares, weight its importance and assign a probability factor to its outcome?
It may not come to that, but it is useful to be aware of what it takes to make better forecasts - and to be a better forecaster - since investors are primarily in the forecasting game. By way of a quick-and-easy guide, readers might turn to an unlikely source - the latest issue of the ‘Applied’ section of the Journal of Experimental Psychology, specifically an article by academics primarily from the University of Pennsylvania about a 'geopolitical forecasting tournament'. This is the link: http://www.apa.org/pubs/journals/releases/xap-0000040.pdf
This was a two-year study from 2011 to 2013 involving almost 750 competitors - mostly male, mostly from the US - which addressed questions such as, 'Will Bashar al-Assad still be president of Syria by the end of January 2012?' The questions - most of which required 'yes' or 'no' answers - followed a generic pattern. Their investment equivalent might be: 'will the share price of The Weir Group (WEIR) top 2,400p by the end of 2015?' More complex questions were of the form: 'which is the most likely, that, for its 2015 financial year, GlaxoSmithKline (GSK) will (a) cut its dividend by 20 per cent; (b) pay a dividend unchanged from 2014; (c) raise its dividend by 10 per cent?'
The academics found that five factors made forecasters better at their task:
■ Forecasting intelligence. Within this label, three features stick out. Able forecasters were good at inductive reasoning, meaning they could draw sensible conclusions from diverse facts. In investment terms - and put simply - that might mean poor share-price performance would be forecast for a company that had heavy borrowings, inadequate cash flow and lost a key contract. Second, they did not jump to conclusions - in the jargon of behavioural psychology, they had 'cognitive control' - and, third, they were numerate.
■ Open-minded. Good forecasters applied the famous quote of JM Keynes: "When my information changes, I alter my conclusions. What do you do, sir?" In other words, they were not overly influenced by their own prior beliefs and were open-minded about changing forecasts that went wrong.
■ Knowledgeable. Participants in the political-forecasting contest with better background knowledge produced more accurate forecasts. It's fair to assume the same would apply to investing. True, knowledge can lead to overconfidence but, mostly, it's better to have it than not. After all, it seems obvious that those who know about the inverse relationship between interest rates and the price of fixed-interest securities would be better judges of the effect of rising interest rates. Or those who know about the volatility of emerging markets will have a better grasp of the likelihood that the BSE Sensex index will trade within a band of plus or minus 20 per cent of its current level for the next six months.
■ Training and practice help. Would-be forecasters can improve their performance if they are simply warned about the dangers of serial errors of judgment, such as wishful thinking, belief persistence or hindsight bias. Similarly, basic instruction in probability theory - dealing with chance outcomes - helps. Someone who understands the slim chances of rolling a dice four times and making a 'score' of at least 20 would intuitively grasp the likelihood of success for an equity investment that depended on just four outcomes even if, individually, each one seemed quite likely. Something else that helps is making more forecasts rather than fewer. It implies more interest in the subject and offers more opportunities for constructive feedback.
■ Teams work. In the forecasting contest, teams did better than individuals. The benefits of pooled information, mutual motivation and the pressures of meeting demanding norms outweighed disadvantages, such as pressures to conform and poor co-ordination.
Sounds like an advertisement for investment clubs. Maybe Bearbull should join one. I forecast I won't.
