Cloud Hosting

The contagion effect from coronavirus epidemic impairs valuations for all asset classes. Q.ai tries ... [+] to build an optimal multi-asset portfolio in this environment.

The drop in global markets over the past few weeks is forcing Wall Street and leading fund managers to reconsider what truly is an appropriate diversified multi-asset allocation in the current market environment.In our attempt to answer this question, we look beyond basic correlations and consider other non-linear forms of price relationships to help us better understand how to build a multi-asset portfolio with Bitcoin.

We are looking at a portfolio made up of the following assets in order to figure out how they have been moving during current sell-off as the Coronavirus epidemic worsens:

What we discover is that it is easier to build a diversified portfolio simply by focusing the majority of the portfolio allocation on Bitcoin, SPY, USO and TLT.Since all these assets are inherently US Dollar exposed, adding UUP to the mix would simply increase exposure to the US Dollar.The strength of SPYs correlation with IWM and EFA suggests an increase in either ETF would simply increase our potential exposure to stocks (which appear to highly correlated globally!) while TLT has a natural obvious price relationship with HYG.

Conventional approaches to portfolio construction is built around the concept of Modern Portfolio Theory (MPT) which assumes that asset allocation should be built around the volatility of assets and their related correlations.Volatility simply measures how much the price change of an asset changes from its average expected change while correlation measures how different assets in a portfolio move together.As an example, if you own Facebook (FB), Apple (AAPL) and Citigroup (C) in a 3 stock portfolio with FB (FB) and Apple (AAPL) being highlight correlated, MPT would say you should own more of Citigroup (C) to mitigate the high correlation between Facebook (FB) and Apple (AAPL)assuming the volatilities of each stock were similar.The biggest issue with this approach is that it is backwards looking, meaning it does not consider future returns but rather assumes past performance will be indicative of future performance. On top of that, a simple MPT approach favors lower volatility assets; since it is backwards looking, the approach assumes a low volatility asset in the past will continue to be a low volatility asset in the future. That is a pretty big assumption! In addition, simple linear relationships do not properly explain how assets might cluster during shocks, like the Coronavirus epidemic we are experiencing at the moment, or just simple clustering in general.

A simple correlation analysis points to a rapid rise in positive correlations in between Bitcoin, IWM and HYG.Notably, the correlation between IWM and Bitcoin seems the strongest though the relative jump between Bitcoin and HYG is the strongest.A positive correlation simply means that Bitcoin is moving in the same relative direction as both IWM and HYG (when Bitcoin is up, IWM is up, and vice versa).While it is worth noting that correlations between Bitcoin and USO (and great proxy for oil and commodities in general) and EFA (international stocks proxy), there hasnt been a material uptick in the cross-sectional correlations of the other assets in our target multi-asset portfolio.In other words, while the sensitivity of other assets in our target multi-asset portfolio remain more or less the same, Bitcoin appears to have a heightened sensitivity.There is probably a US Dollar effect in play here though correlations between UUP and Bitcoin have not markedly increase.The US Dollar effect on Bitcoin is probably worthy of its own standalone article!

Correlation analysis of assets in Q.ai target multi-asset portfolio that includes Bitcoin. ... [+] Correlation analysis is based on price moves from February 20, 2020 through March 20, 2020

While correlation and volatility analysis is useful, we do not think alone it helps to build truly robustly diversified portfolios.The limitations of MPT is that its analysis is purely linear and backwards looking in general but, in reality, markets are anything but linear...and past performance is never a useful predictor of future returns.As a result, we like to use clustering techniques based on artificial intelligence (AI) methodologies to see if we can discover more interesting and non-linear relationships that a basic correlation-volatility matrix might miss.

For our analysis, we use a basic Affinity Propagation algorithm.The ease of use with Affinity propagation is that while it is a centroid algorithm, it is does not require us to estimate the number of clusters like other similar algorithms such as K-Means. What we are looking to see if how the different assets in our target multi-asset portfolio cluster.

Outside of tight clustering between SPY, EFA and IWM, we not observe examples of clustering in our target multi-asset portfoli over the past month. As expected, we see the strength in relationship between BTC and HYG increasing over the past month.Previously, the relationship between both BTC and HYG would be described as nominal at best.The relationships between SPY, IWM and EFA are relatively strong but what is more interesting is that there are no other significant relationships between the other assets in our target portfolio and Bitcoin this is a major deviation from what our correlation analysis implied.

Analysis of clustering of assets in Q.ai target multi-asset portfolio that includes Bitcoin. The ... [+] methodology that is used is Affinity Propagation. Price data is for the period February 20, 2020 through March 20, 2020.

As a result, based on our analysis of both correlations and how different assets in our target portfolio cluster, we think it makes the most sense to focus the majority of the portfolio allocations on Bitcoin, SPY, USO and TLT.Since IWM and EFA both have high correlations with SPY at this time, allocating between the 3 becomes a nominal exercise though SPY will offer better diversification vis--vis USO and Bitcoin.

Read the original post:
Bitcoin Highly Correlated With HYG And IWM ETFs As Contagion Effect From Coronavirus Epidemic Spreads Across All Asset Classes - Forbes