Volatility Management for Cryptocurrency HODLers

Published in

Coinmonks

7 min readAug 25, 2021

--

I thought it would be interesting to look at a well-known technique known as volatility targeting or management and see if it can add value to a long-term allocation to cryptocurrencies.

What is Volatility Targeting?

Volatility Targeting or “Vol Targeting” is a technique used to attempt to keep the volatility of an investment or portfolio close to a pre-defined target. When the volatility of the asset is above the target, we reduce exposure and invest the proceeds in cash or a risk-free asset. When the volatility is below the target, we pull from the cash asset and invest more in the target asset (leverage is often used with these strategies, btw).

Example Time

As a simple example, we may want to limit the volatility of our investment in Bitcoin to 25% (annualized).

To do this, we need a few things:

Volatility Target — this is usually expressed as a daily or annual figure
Volatility Measurement — we need to estimate the realized volatility of the asset
Rebalancing Schedule — we need to decide how frequently we will update our estimate and rebalance to the target weight
“Cash” asset — this could be to hold cash in a brokerage account, or a fixed income ETF, or a stablecoin

Below we see a chart and stats for Bitcoin using daily returns from 1/2/2015 through 8/16/2021. The long-term estimate of volatility (using annualized standard deviation) is 77%. If we want to keep the volatility of our investment in BTC at around 25% annualized, that means we will have a less than full allocation to BTC, with the remainder in cash, for a majority of the time.

The trick to volatility targeting is to estimate what I call a “volatility multiple” or the fraction of total funds that should be invested in the asset. To do this we simply divide our target volatility by the estimate of volatility to get the volatility multiple.

If we want to use the full sample estimate of standard deviation, we would divide 0.25 by 0.76 to get 0.3289, suggesting we would have about 1/3 of our allocation in BTC and 2/3 in cash. A completely unrealistic backtest would be to just multiply .33 by BTC returns to yield an estimate of what our allocation would have done (again this is unrealistic but just an example) — The volatility of our new allocation is right about 25%, and the “ride” is much smoother.

Chart 2: BTC at 1/3 exposure (beginning 5/1/2015 to match vol target example)

Back to Life, Back to Reality

Apparently I’m in a mood to use song quotes today. To make this approach a little bit more realistic, we need to use a shorter-term, rolling estimate of volatility and rebalance our allocation between the target asset and the cash asset more frequently. Here we will use a rolling 90 day estimate of volatility and rebalance monthly. To rebalance we will use the last vol multiple of a month for the entire subsequent month.

Parameters for simple rolling volatility target test

This approach still adds value to a simple buy and hold approach, but the volatility is higher than the target — which is to be expected as we are using a simple rolling estimate of volatility. Interestingly, and encouragingly our simple model that rebalances monthly based on the rolling 90 day estimate of volatility does keep up rather well with the “perfect hindsight” approach of 1/3 BTC, 2/3 cash.

Chart 3: BTC with Rolling Vol Tgt based on 90 day Standard Deviation

“That Would Never Work In The Street”

We see that some kind of volatility management would have added some value to a buy and hold investment in Bitcoin, but perhaps that is a fluke? Many investors are going beyond Bitcoin and hold multiple cryptocurrencies in their portfolios. While we can’t go back in time, we can see how a volatility target strategy would have done if we control for coin selection — that is — run simulations!

This will be a simple experiment: we will choose 5 coins at random, apply a simple volatility target strategy to each coin on its own, and rebalance to an equal-weight portfolio of those 5 coins every quarter. While stablecoins may offer a bit of yield, we will not include that in our simulation results. We will look at the distribution of outcomes for 500 simulated portfolios using the volatility target method, and 500 simulated portfolios that buy and hold an equal-weight portfolio of 5 randomly selected coins.

Pitfalls: Please note that the sample is small in terms of assets and price data. That said, I do think the difference between the volatility managed portfolios and equal weight portfolios is interesting and potentially instructive. We are not accounting for transaction costs either, however monthly trades to adjust exposure to the vol target and quarterly trades to rebalance to equal weight should not be particularly onerous. In reality one would likely throttle rebalancing so that exposures must deviate beyond a reasonable threshold before making a trade.

Simulation Results

On average, our volatility managed portfolios have more than 3X the Sharpe ratio and 1/3 the drawdowns of simple equal weight portfolios. While we would be inclined to attribute this to the fact that the volatility managed portfolios have about 1/4 the volatility of the equal weight portfolios, that is confounded by the fact that both portfolios have nearly exactly the same average return of 30% annualized.

Table 4: Mean stats for simulated portfolios

In Table 5 we can see a full summary of the distribution of key statistics for the two types of portfolios and find that the volatility managed portfolios outperform quite materially by reducing risk dramatically without reducing returns. As cryptocurrencies are quite volatile these days, the choice of volatility target is largely arbitrary and unlikely to require leverage for most investors.

Table 5: Summary statistics for simulated portfolios

A visual analysis of the density of some key metrics is compelling, as well. The distribution of Sharpe ratios for volatility managed portfolios nearly dominates that of equal weight portfolios:

Chart 4: Sharpe Ratios of Equal Weight and Vol Adjusted Simulation Portfolios

And the terminal wealth of the volatility managed portfolios is in a very tight (and positive range) relative to the equal weight portfolios:

Chart 5: Terminal Wealth (Ending value of $1) for Equal Weight and Vol Adjusted Simulation Portfolios

Conclusion

Cryptocurrencies are highly volatile, and while that is part of the excitement, many investors may shy away because of it. One method for addressing this is to use a volatility target strategy which, at least in this admittedly limited sample, would have yielded a substantial reduction in volatility without necessarily sacrificing returns. I imagine many investors would prefer this approach to managing an allocation to cryptocurrencies. I certainly would.

Thank you for reading.

Additional Illustrations

Table 3: Stats for BTC with Rolling Vol Tgt

Chart 6: Maximum Drawdowns of Equal Weight and Vol Adjusted Simulation Portfolios

Notes

* Excluding stablecoins and a few with too little price history

References & Resources

Sample R Code: https://github.com/rufusrankin/WhatsVolGotToDoWithIt/blob/main/Code
Moreira, Alan and Muir, Tyler, Volatility-Managed Portfolios (October 25, 2016). Journal of Finance, Forthcoming, Available at SSRN: https://ssrn.com/abstract=2659431 or http://dx.doi.org/10.2139/ssrn.2659431

Data

Cryptocurrency price data from riingo/tiingo
Currencies selected based on a. being in Coindesk 20 Index as of May 2021 and b. Price data available beginning 12/30/2019

Disclaimer

This article is for information purposes only and does not constitute investment advice. Any opinions are those of the author and do not represent those of Ampersand, Drexel University or any of their affiliates.

Join Coinmonks Telegram Channel and learn about crypto trading and investing