In September 1997, the Fund had an annualized standard deviation of approximately 10.7%, which was significantly lower than the Fund’s long-run goal of 20% annualized standard deviation.
There are several reasons to explain why LTCM had difficulties in raising its level of risk.LTCM structured the majority of its trades in a way that required minimal initial outlay of capital. Furthermore, most of the Fund’s contractual trade agreements were self-financing. This 100% financing combined with the long-short structure of the Fund affected the methods of risk assessment and risk management employed by the firm. One important aspect of the resulting risk assessment methods of LTCM was the fact that trades based on widening price discrepancies would generally attract more capital from investors and arbitrageurs. This influx of further capital usually resulted in lower downside risk on trades as valuations became more extreme.
However, in our opinion the most important factor affecting LTCM’s ability to raise its level of risk is the correlation analysis methods used by the firm. Over the long-term one-year horizon, the firm based its correlation analysis on the fundamental factors that affected trades. When looking at the fundamental risk factors affecting these positions, the majority of the long-term trades were largely uncorrelated and therefore resulted in very low levels of risk. When calculating correlation for short-term trades, LTCM included price movements that resulted from the short-term liquidity needs of traders.
The firm adopted a conservative approach and assumed that many of these positions were positively correlated, however this still did not lead to an increase in the overall risk levels of the Fund.Value at Risk is a method used by many financial institutions to measure levels of risk in terms of a probability distribution of potential profits and losses. For LTCM, this method involved calculating what the approximate change in portfolio value would be if there were a one basis point change on a particular trade. One can then apply this value (from a one basis point change) to the expected total one-year basis point change for a particular trade and calculate the standard deviation of the trade in monetary terms. When calculated for all of the Fund’s positions, one can calculate the overall standard deviation of the portfolio in monetary terms.The adoption of the Value at Risk method played a role in the failure of LTCM because this method does not take into account liquidity when measuring risk.
While Value at Risk is an appropriate approach to analyzing portfolio correlation and calculating risk based on expected profits and losses, it only assumes liquid positions. Therefore, when the Fund started to face heavy losses and was forced to unwind many of its illiquid positions, the previously uncorrelated positions suddenly became highly correlated, and this risk was not taken into account.