There are many talented, mathematical minds in the Risk function of banks and other financial institutions, and they are experts in devising models to manage risk by predicting possible market behaviour. It appears that a PhD in Astrophysics is much more useful in a Risk department than experience of how markets actually behave in the real world. The problem has always been that the models fail regularly, with spectacular consequences. No model can allow for the huge vacuum that existed after the SNB pulled their bid at 1.2000.
The most widely used risk measure is Value at Risk (VaR) and the amount of “value” (read potential loss) allocated to EUR/CHF positions (for example), when combined with some measure of historical volatility of that currency pair, will determine the maximum allowable position size that the trader can hold.
The problem with this approach is that as historical volatility falls (as it did all through 2014 in EUR/CHF) position sizes can be increased whilst still staying within the allocated VaR limit. In layman’s terms, the trader says to himself “this currency pair is not moving much, so I can take a bigger and bigger position and STILL stay within my VaR limits”.
Let’s hope that this mythical trader was long of Swiss francs last week……