Strategy for extreme and unusual events

It is a good idea, and typically these events (commonly termed shocks in asset pricing literature) have a persistent effect on market prices afterwards as well.

However the tricky part would be figuring out whether market prices pre-event (expectations of the shock) accurately depict the magnitude/direction of the actual shock. For instance, investors may be overly pessimistic/optimistic about interest rates before the Fed's quarterly FOMC meetings, so that's an idea you could potentially explore and what sort of factors may be beneficial to predicting the correction that happens post-event. It can be a bit difficult for systematic strategies to capture this though consistently without imposing some rather strong assumptions on market behavior. I'd recommend you initially focus on a particular industry/event class as it's important that your model is more predictive than the overall market's expectations for the strategy to work.

Thanks for your feedback,
Because of the difficulties you describe, I thought about these two points:
1. We will only refer to events that are considered extreme in relation to any expectation [I know they are rare events but so the probability that will affect the markets in the short term is higher]
2. We will only address the short-term, so we don't have to find a systematic response to the long-term [which is much harder to find]

Focusing on extreme events according to set parameters can be easier and safer than finding systematic long-term effects

Because these events are rare, we will have to look at a wide range of assets and this will increase the chance of the events triggering the algorithm

Any feedback is welcome

I'll start:" Some researchers have studied the effect of aviation disasters on the capital market. Bosch, Eckard and Singal (1998:515) examined whether there is a spillover effect or a switching effect on rival companies. They found a switching effect when the non-crash airline overlapped with the crash airline and a spillover effect when there was no such overlapping. Kaplanski and Levy (2010:178, 196), for example, used data from January 1950 to December 2007 to examine the New York Stock Exchange (NYSE) rate of return. The researchers assumed that aviation disasters affect people’s mood and anxiety, which in turn have a negative effect on the capital market. The model they used included the stock index return as an independent variable and a dependent dummy variable representing the day of the week, the day after a holiday, the first days in the tax year and the size of the effect. The event window began on the day before the event and continued three days after the event. The researchers found a significant effect, with an average market loss of more than $60bn per disaster, although the actual loss was less than $1bn. On the third day after the disaster, the stock prices started to increase. They also found that the effect was larger when the stock was smaller and riskier and when the stock belonged to a less stable industry"[www.ncbi.nlm.nih.gov/pmc/articles/PMC6407467/].

Hello Miriam, 

I think it is a good idea and probably central to gaining an edge in asset management. What I looked into but never got to verify were "seismic event modelling" according to this paper: 

www.cfainstitute.org/en/research/cfa-digest/2015/12/interpreting-financial-market-crashes-as-earthquakes-a-new-early-warning-system-for-medium-term

it deals with rare events that form pre-shocks and after-shocks that could be used as a risk management tool. Several asset managers I know are looking for such systems that will just tell them when a shock is coming. 

Thank you so much for the idea, I will definitely test it and see what can be done

Hi Miriam,
may I suggest the use of options strategies as these can be designed to capture alpha from extreme moves outside of the expected movement of a market. The difficulty is in designing a strategy that doesn't lead to a death by a thousand cuts.
Risk events can be categorised as Black, Grey and White swans.
Black swan events cannot be predicted.
One could argue that Grey and White swan events may be predictible, but the warning signs are typically very specific to the event context, so this isn't easy.
In general, I'd suggest that trying to predict and time a major risk event is near impossible, and so this requires a strategy that is actively in the market, positioned in anticipation of a risk event.
Also real-world market factors quickly come into play during a major market event such as lack of liquidity, where asset prices quickly collapse due to the buy-side withholding liquidity to maximise their advantage over the sell-side.
Best,
ES.

@Ernest-Thank you so much for your feedback, your comments make me think from the beginning,
I thought that even if an event cannot be predicted before it arrives, it can still be responded to quickly based on similar phenomena in the past, am I wrong?
Thanks so much for the incredible collaboration