We use AiLA-V001, which is a volatility index constructed from liquid options across the most liquid commodities, to measure the markets view on volatility
Can we observe any relationships between the AiLA alpha product (AiLA S series) performance and the prevailing volatility conditions reflected by the AiLA-V001?
Can we observe any relationships when periods of high volatility are removed from the equation?
Given the lack of data sufficiency with one single AiLA index, all AiLA products were pooled together for the analysis.
The historical expected returns were investigated for AiLA products, conditional on different volatility circumstances.
We use implied volatility and vol index interchangeably as vol index constitutes a basket of implied volatility.
This analysis was done with high volatility periods like the 2020 COVID-19 crash being excluded.
Performance vs IV
We split data into three bins based on the prevailing IV index values, i.e., levels of volatility.
The filtered (removing periods of high volatility) and unfiltered cases shows a slight upward trend that suggest a weak indication of expected returns to be larger during periods with higher implied volatility.
This indication disappears when normalizing the returns to the sample std.dev. of their corresponding bin.
Interestingly, the filtered data shows an inverse relationship between normalized performance and implied volatility, which suggests that high volatility of returns coincide with periods of high implied volatility.
Therefore, it would primarily be explained by the different size of the returns rather than the ability to predict their sign.
There is an indication of a uniform product performance when traded on a risk adjusted basis.
Performance vs Change in IV
We split the data into three bins based on the prevailing change of IV index values, i.e., decreasing, flat or increasing IV.
This indicates a different pattern for long-short and long-only AiLA products.
The AiLA long-short products indicates a tendency to perform well in both a decreasing as well as increasing IV environment.
It is interesting to point out that AiLA long-short products perform well in periods of high increases in IV, which may suggest that the long-short strategy takes advantage of the changing market conditions at a faster velocity.
The AiLA long-only products indicate a strong tendency of performing best in a decreasing IV environment and to slightly underperform during increasing IV periods, which is also consistent in both the filtered and unfiltered data set.
The long-only pattern appears intuitively consistent with the typical IV development during a market crash period, e.g., such as the first half of 2020.
Performance vs Change in IV
The performance pattern w.r.t increasing and decreasing IV is generally reflected also when broken down to an individual AiLA product level.
Here, historical monthly returns are compared to increasing and decreasing IV periods. We can see that for 24 out of 30 AiLA products, monthly returns are higher in increasing IV periods than the monthly returns in decreasing IV periods.
The plot for historical daily returns is inconclusive in determining the above, likely due to a daily period being too short to assess the effect of changing IV.
It should, however, be noted that the limited data set and corresponding statistical errors generally makes it difficult to draw generalized conclusions.
An implied volatility index (AiLA-V001) was used to measure the markets view on volatility and was used in a study of a potential relationship with the performance of the AiLA alpha products (AiLA S series).
An indication of a slight performance difference w.r.t the level of IV was obtained, however, the potential effect was found consistent with the difference in size of the returns.
A different performance pattern was, however, observed between the AiLA long-short vs long-only products.
For the long-short products high expected returns were obtained both during increasing as well as decreasing IV, whereas for the long-only products decreasing IV typically was associated with significantly higher expected returns than increasing IV.
The usage of a filtered data set where periods of high implied volatility were removed proved to be useful as it showed similar patterns to the unfiltered data set, which show that we can assume the same patterns of returns in most implied volatility situations.
These result are intended as an initial overview regarding relationships between the AiLA alpha (S Series) and AiLA Vol (V Series) products, with further research work expected to follow on the topic of volatility.