Towards a fair valuation of bitcoin

Those who do not learn from history are doomed to repeat it.’ — George Santayana
Past performance is not indicative of future results.’ — Axiom of investment

Summary

This report makes the case that the 200 weekly moving average represents a good long-term forecast for determining a base level for a fair valuation for bitcoin. This metric offers the best compromise between filtering out the noise of bitcoin’s price movements without altogether destroying the signal. The report also argues that, like gold, bitcoin is not suited to fundamental analysis and that its limited price history is therefore the best and arguably only way to assess price in the context of its long-term development.

Disclaimer

This article is not intended to constitute trading or investment advice. The author is not a professional technical or fundamental analyst. Bitcoin is volatile and cryptocurrency is a high-risk asset. The material below is offered for interest and discussion only. For more information, visit www.Blockworm.net.

Introduction

Bitcoin’s meteoric price rise over 2017 attracted global attention, as did its almost equally spectacular crash through 2018. While this kind of bubble cycle is nothing new for bitcoin or its longer-term users, it has brought unprecedented interest in cryptocurrencies from all quarters. Bitcoin has polarised opinion, with experts (many self-proclaimed) on both sides of the fence predicting future values of anything from zero to one million dollars over the coming years. Few of these numbers have any rational basis, or else they are the result of a highly selective approach to data, or a poor understanding of cryptocurrencies, economics or other relevant disciplines.

Nevertheless, given that Bitcoin represents an entirely new asset class that has existed for only ten years, a ‘fair’ value for BTC is extremely difficult to establish. It does not pay interest rates or dividends, and cannot be valued using traditional metrics like P/E ratios. A far lower price and market cap could support the same real-world utility in terms of payments and remittance, and a large proportion of its price is speculative. Similarly, it is very difficult to judge a valid upper level for price given bitcoin’s potential utility, even as an order of magnitude.

Fundamental analysis is almost impossible; notwithstanding comparisons with gold and its $8 trillion market cap. Technical analysis, meanwhile — the study of price action and statistical probabilities — is extremely popular among cryptocurrency traders. In between these, a series of increasingly sophisticated models are being developed to help determine whether the network is under- or over-valued based on a snapshot of its properties at the time. These typically use on-chain and related metrics such as overall number of addresses, number of active addresses, number of users (assuming multiple addresses per user), exchange accounts (assuming multiple users per address), USD transacted daily, BTC transacted daily, and so on. Many of these approaches are rooted in variations of the Network Value to Transactions (NVT) ratio introduced and popularised by Willy Woo and others.

This article takes a far simpler and considerably less profound approach to long-term forecasting and pinpointing under- and over-valuation at a given point in time. It is interested in the most basic (and hopefully therefore reliable) long-term trends for bitcoin, bearing in mind the helpful distinction between forecasting and prediction. It is also intended very much to be a starting point for discussion, used in conjunction with other metrics, and not as a last word.

The gold analogy

Gold offers a useful parallel for bitcoin — which has, of course, become known in the community as ‘digital gold’. Like bitcoin, gold provides no revenues like bonds or stocks do. It has little intrinsic use — certainly nothing that would justify its $1,200+ valuation. While it is used in electronics, its real-world applications are few; it does not compare to commodities such as oil, wheat or coffee, which are traded based on (amongst other things) fundamental demand by global consumers.

Additionally, just like bitcoin, gold’s mining costs provide some medium-term context for price but do not reliably underpin it. (One difference is that the supply of new gold is determined by mining costs, while Bitcoin’s network hashrate is impacted by mining costs but the supply of new BTC itself is set algorithmically.) As 2018’s bear market played out, we saw Difficulty decrease significantly as some miners found the costs of running hardware outweighed their revenues; the free market for hashrate operated exactly as Satoshi Nakamoto intended it to.

Economic realities such as interest rates and GDP growth or otherwise may inform expectations of future gold prices, but this is the only kind of ‘fundamental’ analysis available. Study of mining infrastructure and investment is not a predictor of value.

It is therefore a category error to seek to value bitcoin on its fundamentals: it has none. Ultimately gold is worth so much simply because we collectively agree that it has value, and have used it as a store of value for millennia. The same, for a far shorter timespan, is true for bitcoin.

The technicals are the fundamentals

Thus when it comes to forecasting the future price or adoption of bitcoin, or seeking to determine some kind of ‘fair’ value for the present, we are presented either with an extremely complex situation, or a very simple one. Network effect, which may be measured by metrics such as active addresses and amounts transacted, correlates with value according to Metcalfe’s Law. In short, if the number of active bitcoin users doubles, we would expect price to quadruple, all other things being equal. They never are, even for well-established assets. Not only is bitcoin very new — meaning that few models have enough data to give a reliable picture — but it is extremely volatile, so price can deviate from expected value both dramatically and for a long period of time.

The ‘fundamentals’ are effectively the knowledge that people have used gold as a form of money and store of value for millennia, and are unlikely to decide, collectively and suddenly, to stop. In other words it is price, over the long term, that is the sole and fundamental reassurance that gold is a good investment — what might otherwise be considered a technical indicator.

Of course, in the case of gold we are talking about many, many thousands of years of history, across a vast range of locations, cultures and economic contexts. Bitcoin’s own history is just a decade long, confined to an unusual period in human history — the years since a global financial crisis, caused by factors that could not have existed in previous centuries — and we only have price data for eight or so of those years.

At the same time, that price data captures every factor within it. There is no better predictor for what the price should be at a certain point in bitcoin’s history — given number of active addresses, daily value transferred on-chain, exchange accounts, economic and geopolitical factors, and so on — than what it actually was. The impact of rate-of-change of growth in adoption is contained within that figure, albeit overlaid with a lot of noise on a day-to-day basis.

Here, we immediately run into the problem that price is necessarily a ‘lagging’ indicator: it shows what is and then immediately was, not implies what might be. Statistics like new exchange registrations and Google searches may predict short- to medium-term future price activity, but suffer from the same problem, one step removed: we only have those statistics once they have happened, and past performance is not an indicator of future performance. The trend of increasing numbers of new Coinbase accounts cannot continue indefinitely, and most likely has a two-way relationship with upward-moving price.

While bitcoin lacks gold’s longevity, its long-term price data may be the most reliable and possibly only means of determining fair value. To put it another way, its technicals are its fundamentals, provided that data is handled correctly. At the very least, we should compare insights from this data against other more sophisticated models to determine the degree of correlation.

Moving averages

Volatility is one of cryptocurrency’s signature properties, and it is a challenge to cut through the noise without killing the signal altogether. We are interested in direction of travel, speed of travel, and rate of change of speed. Bitcoin has rapidly risen in value since its inception; will it continue to rise in value? If so, is that rise accelerating or decelerating? At what point can we expect it to level out?

Our starting point for discussion is the 200-week moving average (200 WMA). Coincidentally, bitcoin recently touched the 200 WMA line for the first time in four years. It has never yet traded beneath it for any appreciable amount of time; in an exponentially-rising market, the 200 WMA reacts slowly enough that price has to drop dramatically to hit an average formed over four years. The last time BTC fell significantly below the 200 WMA was in the January 2015 capitulation from the last bear market, when it dropped to $152 (Bitstamp); the MA then sat at $212, where the weekly candle ultimately closed. After Q2 2015 BTC staged a strong recovery and remained well above the 200 WMA until December 2018.

The selection of the 200 WMA may seem somewhat arbitrary. After all, it is just a line on a chart and provides no guarantees of anything. In the short- to medium-term, there is no reason to expect bitcoin to respect that support level, and in the long term it must ultimately fluctuate around it. It could be argued that it would be better to use an MA that cuts through the price signal perfectly, spending exactly 50% of the time above it and 50% below. (The 30-week or 200-day averages would broadly achieve this, and we use the 30 WMA for context below.) However, we want a figure that cuts out even more noise than this, giving an overall impression of long-term direction and rate of change without fluctuating any more than necessary — it should be as smooth as possible, while still reflecting the trend. The 200 WMA does this, and the fact that BTC has come back into contact with it shows that its value is not completely out of touch with reality, as it might have seemed a few months ago. Like shorter-term simple moving averages, exponential moving averages (EMAs) are more responsive to recent prices, which in this instance is counter-productive.

Moreover, 200 weeks is roughly the length of the last complete cycle for BTC; from the last all-time high in November 2013 to the most recent in December 2017 was just over 200 weeks, and — on the basis of current information — the low-to-low for the bear markets that followed these will be around that number too (200 weeks is long enough to be a forgiving indicator if the bear market stretches on a while longer). It is therefore long enough to be representative of the highs and lows across a full market cycle.

Lastly, the 200 WMA is a well-established indicator for traders in the conventional markets, and often used as part of a ‘reversion to mean’ trading strategy.

Data for the 200 WMA is available only from mid-2014, since BTC only started trading in mid-2010, so this is the first point at which four years of data is available.

Chart 1: Bitcoin price (log), 200 WMA and 30 WMA

We start by plotting the 200 WMA, and then the increase in that MA one year ahead. Note that this is not the same as the one-year ROI, since the 200 WMA rarely coincides with price: we are interested in trajectory here, and the rate of change in the trajectory of the 200 WMA. (Price data courtesy of https://uk.investing.com/crypto/bitcoin/historical-data.)

Chart 2: 200 WMA and one-year increase

From this chart, we can see that:

  • The 200 WMA has always risen
  • Its lowest one-year increase was 136%
  • The average one-year increase is 228%

For comparison, for the 30-week (roughly 200-day) moving average:

  • The 30 WMA fluctuates around price, spending roughly half the time above and half below
  • The 30 WMA is more responsive to price than the 200 WMA, both rising and falling
  • Like the 200 WMA, the 30 WMA always rises year-on-year
  • The lowest one-year gain was 41%
  • The highest one-year gain was 3,516% (January 2013–2014)
  • The average one-year gain was 605%.

To reiterate, the gains in the moving averages do not necessarily translate to gains if bitcoin is bought and sold one year apart, because price only rarely coincides with each MA (especially the 200 WMA). We are interested here in the long-term average value of BTC, and the pace of change of this. As it happens, buying when price coincided with 200 WMA in 2015 would have been an extremely profitable move on any timeframe. It is not yet possible to know whether buying when price touched the 200 WMA in December 2018 will be profitable. Buying when price has coincided with 30 WMA has not always been profitable on a yearly timeframe — the last time this happened was in March 2018 when bitcoin was over $9,000.

Conclusion

Like gold, bitcoin and cryptocurrencies in general are notoriously difficult to value on fundamentals. Consequently, long-term price development provides the best forecast of future value. The 200 WMA screens out volatility while capturing overall trend. Price has rarely dipped below it, but did touch it in January 2015 and December 2018, demonstrating it is not a completely unrealistic measure. Moreover, the 200-week period takes in the cycle from peak of one bubble to the peak of the next, and likely the same for the low for each cycle too. Because the 200 WMA tends to track below the price of bitcoin, we would expect it to provide a lower end for a medium- to long-term forecast, assuming there is no ‘black swan’ event.

The 200 WMA currently stands at around $3,200. Conservatively, if bitcoin continues its recent annual growth (a premise that contains a very large number of assumptions) then we would expect minimum prices of $6,000, $12,000, $25,000 and $50,000 at the end of 2019, 2020, 2021 and 2022 respectively. These values do not take into account the effect of another bubble, which could temporarily lift prices to multiples above these levels — easily as much as 5–10 times, if past performance is an indicator. They merely represent the extrapolation of the long-term average trend and something close to a reasonable ‘floor’ price in the coming years.

In the immediate term, and perhaps more realistically, we can expect a somewhat different dynamic. The 200 WMA includes values from four years ago, when bitcoin was below $200. It did not trade above $1,000 again until January 2017. As these weeks fall out of the 200 period, and are replaced by recent ones in the $3,000 range, the MA will inevitably rise and price may well dip below it again.

In short, the 200 WMA doesn’t guarantee anything, but it is not a bad indicator on which to base a long-term forecast — and in any case, it may be the best we’ve got.

Like what you read? Feel free to get in touch, share this post, leave a comment or a tip: 1KWQBs7uNYa92HSRU9Rz1czdhEaME9gJic