Loading...

Pros & Cons of the Varying Length of the VAR Lookback Window

The first question you may ask is: What are we even talking about? "the VAR Lookback period" --> When we estimate Value at Risk and talk about using the non-parametric, historical simulation technique, we need the "shifts" from the timeseries over a period of time as inputs. This duration of time is called the "lookback window" or "lookback period". Right away we can see that if one lookback period is "longer" than another, there will be more data points considered when estimating VAR. Is that good? Let us discuss the Pros & Cons and explain why CryptoDataDownload chooses a particular approach.

Pros
A longer lookback window means that we include a longer timeseries and so therefore, we implicitly get a better representation of how an asset has traded over the timeframe. When a lookback window is too short, it may "not remember" certain important events of the recent history. Even right now, most Value at Risk models would not include the shifts seen during the COVID Crisis, as they are now almost 2years+ removed. Having a longer window would include these important risk events. This idea raises the question: Should we just have a lookback window of all data points? Back in 2017, that is exactly how we were calculating it! Cryptocurrency was so new and there were not enough volumes or data points to feel OK without the full timeseries. But that is not the case today. We have seen almost a decade of strong volumes and cryptocurrency infrastructure development that we feel no longer demands using the entire lookback window.

Cons
While we would include more historical events with a longer timeseries window, we are invariably less responsive to new data points as time passes. Because the window is longer, it will require more time to pass before data falls out of the window and so shifts are "stickier." This is the trade off: responsive Value at Risk to a changing risk environment (shorter window) or a more "conservative" estimation as you include events from the past (longer window).

How We Do It
Our approach is to use a shorter, one year lookback window for 1-Day and 10-Day Value at Risk. For cryptocurrencies, this means a full 365 days. This allows our model to be fairly responsive to changing market conditions. However, this does not mean that we forget about historical events. We separately have created a risk taxonomy of specific stress events. These historical cryptocurrency events, like the Bitfinex Hack of 2016, and also global macro events like COVID or the China Commodities Slowdown of 2015 or the Reflation Trade in the late summer of 2021, play an important part in our risk analysis. Markets are intercorrelated across asset classes and so utilizing stress scenarios helps estimate cryptocurrency performance under different market regimes.



Notice: Information contained herein is not and should not be construed as an offer, solicitation, or recommendation to buy or sell securities. The information has been obtained from sources we believe to be reliable; however no guarantee is made or implied with respect to its accuracy, timeliness, or completeness. Author does not own the any crypto currency discussed. The information and content are subject to change without notice. CryptoDataDownload and its affiliates do not provide investment, tax, legal or accounting advice.

This material has been prepared for informational purposes only and is the opinion of the author, and is not intended to provide, and should not be relied on for, investment, tax, legal, accounting advice. You should consult your own investment, tax, legal and accounting advisors before engaging in any transaction. All content published by CryptoDataDownload is not an endorsement whatsoever. CryptoDataDownload was not compensated to submit this article. Please also visit our Privacy policy; disclaimer; and terms and conditions page for further information.