Virtually every option trader will tell you they have risk controls in place. Whether or not they’re useful depends primarily on two factors:
- Do the risk control accurately capture the risks of option positions
- Are the risk controls faithfully followed or occasionally overridden?
A major contributor to poor risk management decisions lies with not understanding the distributional choices when dealing with financial instruments. Many assume that a security’s distribution is Gaussian, or normally distributed, like the graph below illustrates:
The problem with assuming this kind of distribution is many option-based strategies will not fit into this neat structure. Options are often described as having “asymmetric” payoffs, which do not match the above image at all. If the data does not fit, assumptions about “worst case” scenarios made off the above curve might be wildly inaccurate.
Even before a proper distribution can be estimated, a series of questions must be addressed when vetting the data
- Is it discrete or continuous? In other words, are the values the data can take limited/pre-defined or can the data take any value?
- Can a reasonable estimate of the outcomes based on probability be determined?
- How do outliers impact the results? Do they skew the distribution one way or another?
- When outliers do occur, how far do those tails extend? What happens “beyond the horizon”?
- What data set do we have available to analyze? If there are no “bad” occurrences contained within the data set, does that mean the model will assume bad outcomes are impossible?
This is just a small sample of the many questions that must be answered before any trade or position is placed, especially when using options which have asymmetric payoffs. Most blow-ups occur simply by an underestimation of the convexity, or non-linear behavior, that options bring into a portfolio.
These are points to bring up when considering an options strategy as it can give advisors a better handle on how the risk controls are developed.