Are Cryptocurrencies Predictable Using Fundamental Valuation Measures?
Fundamental value investing is popularised by Warren Buffet (PDF) and backed with considerable academic and market tests at least since Fama and French (1992) (PDF). Value investing is interesting because it can predict the future return of assets compared to the general market. But how can value investing be applied to cryptocurrencies?
The idea behind value investing is to buy an asset at a low price and then wait for other investors to realise the higher value of the asset. The problem is to find those assets.
One way to search for them is to calculate the price-to-earnings (PE) ratio for a lot of assets and compare them. The PE ratio measures how many years of earnings you would need to earn back your investment. As a greedy investor, you want to have your money back as fast as possible and therefore the price-to-earnings ratio should be small. In other words, a value investor would buy assets with a small PE ratio and potentially even sell assets with a high PE ratio.
Fundamental value investing has been applied to equity markets, but also to Bonds, commodities and fiat currencies (see Asness, Ilmanen, Israel and Moskowitz, 2015).
What value measures are appropriate for cryptocurrency markets?
The cryptocurrency space is thought to lack the basic data sources and history to make value investing, or investing on fundamental factors, an effective strategy. So we wanted to see if there is a way to use value investing strategies and if so, if a value investor would have been rewarded by the market over the last 12 months until 7/31/17.
Crypto markets are not the same as equities and therefore we need to look for relevant measures of value. As noted above, one standard measure of value, the price-to-earnings ratio or in short PE ratio, cannot be applied to the crypto assets due to lack of data.
Instead, Willy Woo, a bitcoin and alt-coin investment analyst, suggested to use Market Capitalization to Volume as an estimate for the price-to-earnings ratio.
The idea of valuing a company based on volume has president. Software-as-a-Service (SaaS) companies are also traditionally not measured with their earnings, but their volume of transactions.
The argument for SaaS companies is that net income takes a long time to materialize and that even for the largest SaaS companies, there is little relationship between earnings and valuation (see How to Value a SaaS Company, Whitepaper, Saas Capital, 2016).
At the 2017 Token Summit, Chris Burniske suggested we call Market Capitalization, Network or Utility Value so as to differentiate from equity markets. Accordingly, we will be using the term Network Value to Volume as the cryptocurrency PE ratio.
Value investing is an investment paradigm which involves buying securities with a low PE ratio.
The flipside to value investments are called growth investments. These investments have a larger than average PE ratio and would therefore be considered overvalued to a value investor. However, these investments might exhibit an above-average growth to justify the larger PE ratio.
Bitcoin provides an interesting example. Woobull and Chris Burniske suggested that the NVV ratio for Bitcoin is mean-reverting around 50. Mean-reverting means that, if the NVV ratio is above 50, one should sell Bitcoin and if it is below 50 then buy Bitcoin. This is exactly what value investing predicts. Buy assets when they are cheap, i.e. have a low NVV and sell them when they are expensive, i.e. have a high NVV.
It’s difficult to measure the performance of such a strategy from one asset alone or over short time periods. Generally, a large universe of investments is sorted according to these measures and then the lowest (highest) group of investments bought (sold) (see Asness, Ilmanen, Israel and Moskowitz, 2015).
Putting the Network Value-to-Volume (NVV) Ratio to Work
With that in mind, we applied a NVV ratio to all 481 cryptocurrencies that were available on 7/31/16 at Coinmarketcap and calculated the performance over one year. For our NVV ratio here we used the trading volume on exchanges and not the transaction volume on the blockchain as Willy Woo and Chris Burniske suggested. Therefore our NVV ratio here is a liquidity ratio. Also, we considered low trading volume as zero and if the cryptocurrency was not listed anymore at the end of the period, we considered the price as zero.
We calculated the performance of the cryptocurrencies over this period and ranked them according to their NVV. With this ranking, we grouped the cryptocurrencies in five buckets with an equal number of cryptocurrencies. The average return multiple and the average NVV is shown in the figure below.
Value investing is one of the cornerstones of finance. However, last year it completely underperformed in cryptocurrencies with the value measure above.
Cryptocurrencies with a high NVV had a better performance than cryptocurrencies with a low NVV. Hence, value investing was a very bad idea last year compared to buying high NVVs.
For example, the group of cryptocurrencies in the 4th quintile had a return multiple of 25.93. On the other hand, the group of cryptocurrencies in the 1st quintile had a return multiple of only 9.02.
This is exactly the opposite of what value investing would suggest. Low NVV should have a larger return in the future than a high NVV. We also calculated the performance of these groups over 1 month and 3 years with almost the same result.
The question is how long does it take for investors to realize that they are overpaying.
For example, Bitcoins were trading with a higher multiple than 50 on multiple occasions, but so far has always come back to that level after some time.
In fact, the NVV multiple of bitcoin was 166 at the beginning of the period and therefore well above the multiple of 50. As an anecdotal example, Bitcoin also underperformed the market and returned a mere 3.2 over last year.
So, maybe this value measure works for large cryptocurrencies like Bitcoin, but not for small cryptocurrencies. Fama and French (1992) also suggested a second factor as well which they called the size factor. This factor is measured by the market capitalisation of the company.
Do large and small cryptocurrencies behave differently through a value lens?
We used the same dataset and calculated the NVV for large and small cryptocurrencies separately. We assigned the cryptocurrencies to five groups of network value (size) and 5 groups of Network Value to Volume (NVV). The table below shows the return multiple over one year on each of the 25 different buckets. Each bucket holds about the same number of cryptocurrencies (besides the extreme buckets of small size and high NVV).
Interestingly, large cryptocurrencies seem to show a value effect, whereas small cryptocurrencies seem to follow the growth model. Cryptocurrencies with high network value (blue row) had lower return multiples with higher NVV. In other words, over the past year one should have bought cryptocurrencies with a moderate NVV rather than with high NVV. This only applies to the largest cryptocurrencies though.
On the other hand, small cryptocurrencies (red row) show an increasing return multiple for higher NVVs. Therefore the best strategy for them would have been to buy cryptocurrencies that are hyped and have rather high NVVs.
Our research shows that only the largest and most common cryptocurrencies show a value behavior. Only for these, a value investor would have been rewarded for buying low Network Value to Volume, i.e. either at a low price or that the cryptocurrency already shows a high volume.
In this article we wanted to explore if a value investor is rewarded in the cryptocurrency market. Value investing is one of the most famous strategies in equity markets and is promoted by investors such as Warren Buffet.
We applied the measure of Network Value to Volume (NVV) to measure value in cryptocurrency markets. We found that investing in value over the last year made this investor a return multiple of 9.02. However, this is very far from the return multiple of 25.93, if he would have invested in a cryptocurrencies with a high NVV. Therefore, growth investing actually was very successful for the last year compared to the market, but not value investing.
However, buying low NVV paid off, when looking only at the largest cryptocurrencies measured by the network value. For young cryptocurrencies with a low network value, the best strategy was to focus on their growth.
In the future, other factors should be tested in cryptocurrency markets as well and be analysed over time. For stock investments, the list of investment factors has been well extended. For example, Jegadeesh and Titman (1993) (PDF) show that buying momentum stocks outperform the broad index. Momentum stocks are selected if they did well in the past.
Ang et al. (2006) (PDF)found that stocks with low volatility have high average returns. Kallerhoff (2016) applied machine learning to these factors and improved the efficiency of investing even further.
Future research might also consider the transaction volume of each cryptocurrency instead of the trading volume that is applied here. The trading volume is on the exchange. In general though, the volume on-chain should be considered to calculate the cryptocurrency PE ratio as suggested by Willy Woo and Chris Burniske. Our measure of value is therefore a liquidity measure, which seems to be important in cryptocurrency markets as they are exchanged frequently. We have also not analysed the relationship between trading and transaction volume.
Some Exciting News!
I’ve co-founded a Crypto Hedge Fund with an incredibly talented founding team.
Protos Cryptocurrency Asset Management invests in pre-ICO tokens and trades established tokens like bitcoin using advanced quantitative strategies. It’s founders have 13 years combined experience investing in crypto, have managed funds/proprietary investments of over $1B, have founded and sold 3 tech startups for proceeds of 400M and 1 investment bank at a valuation of $250 million +.
Learn more at www.protos.tokenhub.com