Comparing Risks, By Manager, By Day
November 22, 2010
You're asking managers of your funds to react to changing market conditions every day. Scores of managers, scores of funds, nationally and globally.
Yet you haven't yet started to compare their assessments of risks-or reactions-in anything close to same-day or next-day reviews.
The era of real-time action and undetermined-time reaction is drawing to an end. New software tools, constantly updated databases and advanced math that "normalizes" results as if all markets closed at one time are seeing to that.
This is the case, for instance, with global risk modeling tools provided to mutual fund, hedge fund and pension fund managers by Axioma, a New York firm which helps asset managers assess risks and find the underlying reasons for above-average results.
Axioma maintains a database of price movements for its Portfolio Optimizer product that is updated daily and allows portfolio managers to see the degree to which their holdings are subject to above- or below-normal swings.
The norms? The swings can be gauged against the overall performance of markets or the more narrow performance of industry sectors or other factors.
"Essentially, when people build portfolios, they have intended and unintended bets, that are implicit in that portfolio," said Chief Executive Officer Sebastian Ceria. "When you build a portfolio bottoms up, the question is: do you know what are all the bets you are taking with the portfolio you just built?"
Axioma's software helps fund managers figure out not just the risks of the "intended bets" they have made on stocks or industries that they are familiar with, but the risks of "unintended bets" that populate their portfolios by default.
The software does this by two forms of reviewing risks of particular holdings: fundamental analysis and statistical analysis.
Fundamental analysis is the more straightforward to understand. Stocks are assigned numerical values for such characteristics as "value,'' "leverage," "growth" and "momentum."
In effect, if the stock has a lower ratio of stock price to earnings than similar stocks, it gets rated highly on ''value." If it has less debt to earnings than similar stocks, it rates well on "leverage." And if its stock has moved upward or downwards at a great rate in recent times, then it rates highly on "momentum."
But such analysis is not necessarily the best means of predicting performance when there is volatility. For that, the company provides statistical analysis.
The software combs the historical data for patterns. Recurrent patterns become "statistical factors" used to predict future movements.
The factors have no names. They are not linked to definable reasons for stock movement, like lower interest rates. They are simply numbers which can be considered to have distinct correlations with different outcomes.
It's as if you were trying to predict a person's particular age, but all you have in hand are a long series of data on other factors about human beings. In your database are lots of samples of the height, weight and width of human beings. By having enough samples, your program can determine with fair reliability what the result is likely to be for what Giuseppe Paleologo, director of Axioma's professional services, says is a "hidden variable," such as age.
The factors can be used to compare the performance of stocks-and fund managers-on a daily basis. If a given statistical factor has a rating of 1.00 for a given day and a particular stock is rated -0.0019 on that factor, and if the factor gains 3 percentage points for the day, that particular stock will be expected to decline by -0.57% in its price, due to that factor.
The point is to help managers see where their "bets" are lining up with expectations and where they are falling short. If the fundamental analysis shows that a given stock could gain either 18.9% in the next year or lose 18.9% and economic conditions in Spain account for nearly 25% of that swing, that may be a factor to address.
"If you have too much of a bet in Spain, and you are not really talented at deciding whether Spain is going to be good or bad for making investments, then essentially you should get out of your Spanish exposures, because that is not an area where you have skill,'' Ceria said.
In May, Axioma introduced a series of country-by-country risk models that allow fund managers to see how their portfolios are affected by trading activity and results in different time zones and different market conditions.
Those models can be customized by fund managers. As Izabella Goldenberg, senior manager of professional services, points out that it's no longer good enough to just have a risk model that shows how stocks in developed countries will perform in different economic conditions - it's important to factor in how all developed countries other than the United States might react to conditions. Similarly, there can and should be an emerging markets risk model with China in it and one without it.
The real benefit, though, is not just being able to get a clear handle on risks that you can hedge away and those you can't.
This approach "will allow you to compare managers in a common language," on both performance and exposures to risk, Ceria said.
And help you separate luck from skill.