# How to choose the right altcoin for your crypto-folio?

## And Minimize Variance

Believe it or not, but the worst performing top-16 cryptocurrency for 2017 (if you were to invest on Jan 1, 2017 and reassess your situation on July 28, 2017) is everyone’s darling, Bitcoin. Over the last ~7 months, the return on Bitcoin stands at 3.3x against an average return of 26.8x for the remaining 15 coins put together* (which ranges between 81.5x for Stratis and 6.0x for Monero). Before the Bitcoin bulls charge at me for trying to dissuade people from investing in Bitcoin, let me make it amply clear that I am a firm believer in the long-term potential of the mother-coin; a fact that is reflected through my crypto-folio, more than 50% of which is made up of Bitcoin.

The toughest decision for me, however, has been to figure out the allocation of the remaining 50% of my capital I have earmarked for cryptocurrencies, particularly from the standpoint of reducing the volatility of my portfolio. I would like to determine the asset classes which move together and only invest in one/two assets from cluster of such assets. Fundamental analysis provides little to no insight into these alternative assets, and proponents of technical analysis are also at a loss when faced with the difficulty of analyzing new currencies with little historical data, irregular variations in volume and the debate around indicators lagging price and volume (there are several articles in favor of /against the two approaches — you can click here, here, here and here to read more).

This left me with the option of either picking a coin/ a few coins randomly or investing in a diversified class of coins, all supported by high average daily trading volumes. I decided to go with the top sixteen coins, by market cap, to avoid any pump and dump targets. I also wanted to minimize variance while maximizing returns and followed the FTSE Global Minimum Variance methodology, whereby I calculated the correlation coefficient between these 16 asset classes. The objective was to determine the correlation between two coins and accordingly cluster the ones that are highly correlated. The following table lists down my findings**.

There a few things I learned when I analyzed this table:

- To nobody’s surprise, the correlation between all coins is fairly close to 1, indicating that all coins move in the same general direction.
- However, there are a few that chose to find their own trajectory, even if headed in the same general direction. For example, Ripple and Litecoin seem to stand out with significantly less than the average correlation between these two coins and the rest.
- The correlation of most coins with Bitcoin is fairly high, indicating that Bitcoin does indeed drive the overall market. Only Ripple, Bitshares, Bytecoin and Lisk has a correlation coefficient of less than 0.9 with Bitcoin. Ether too seems to feature in almost all the coin clusters.
- Some clusters (cluster, as defined here, is a group of four coins with high correlation, not amongst each other necessarily, but with the cluster head) clearly outperform the others. For example the
*Stratis Cluster*average return stands at 43x, head and shoulders above the other clusters (particularly the*Bitcoin*(13x)*, Dash*(11x)*and Monero Cluster*(11x))

What does all of this mean for portfolio management? If I have to choose a certain number of altcoins that I will go with for my portfolio (say 4), I will avoid other coins in the cluster of that coin, to minimize variance. For example, if I choose to go with Ether, I will avoid ETC, though I might instead choose Siacoin from the Ether Cluster since it made 50x returns during this period (if I was to assume that historical performance is an indicator of future performance) against 24x of Ether.

To be clear, the idea of this exercise is not to forecast which one of the altcoins will outperform the other, but to minimize the variance of your portfolio through diversification, designed to avoid over concentration to achieve reductions in volatility.

I am in the process of testing this hypothesis and will be happy to report the returns and variance of my portfolio in a month from now (indexed at 1000 on July 28, 2017, with one-fourth of my total capital in Ripple, Stratis, Monero and Siacoin each).

**I chose to work with the following list based upon market cap, average daily trading volume and the time period for which the coin has been in circulation — Bitcoin, Ethereum, Ripple, Litecoin, NEM, Dash, ETC, Monero, Stratis, BitShares, Waves, Steem, Bytecoin, Siacoin, Lisk, Dogecoin*

***Correlation coefficient has been calculated as the average correlation coefficient of three time periods — July 29, 2016, to July 28, 2017, Jan 1, 2017, to July 28, 2017, and April 1, 2017, to July 28, 2017)*