Unique Addresses And Bitcoin Market Cap - A Powerful Correlation

Includes: BTC-USD
by: Hans Hauge


Unique addresses in the Bitcoin network might be a proxy for the number of current users, but there are limitations.

We look at linear regression and standardized outputs to get a better understanding of the implications.

Using the resulting regression equation, we evaluate times in the past when there have been large disparities between the actual and predicted values and what happened afterwards.

What are unique addresses and what is their relationship to Bitcoin's market cap?

Unique addresses in the Bitcoin ecosystem are payment addresses that have a non-zero balance. This metric is one way of understanding the daily use of the Bitcoin network.

Upward or downward modifiers

At times, the number of unique addresses tends to increase, and at other times it tends to decrease. This is affected by the following factors.

Things that increase unique addresses.

  • When fees are high, people leave their UTXOs in multiple addresses, because it costs too much to consolidate them.
  • When the network is popular, more unique addresses are in use.
  • As the Bitcoin network grows, in general there will be more unique addresses over time.

Things that decrease unique addresses.

  • When fees are low, people (or their wallets) will consolidate their funds into a single address.
  • Batching reduces the number of unique addresses in use. More batching will decrease unique addresses and less batching will do the opposite.
  • When the market tanks, less unique addresses are in use because people just aren't transacting as much (not as fun to speculate).

Armed with this knowledge, I was suspicious that there might be a relationship between the number of unique addresses and Bitcoin's market cap. This is based on the assumption that Metcalfe's Law might apply. To test this we would need to map the user count (or even a proxy of this figure) the total value of the network. The number of unique addresses is not a perfect stand-in for the number of users, but it is true that in general the more users we have, the more unique addresses in use and vice versa.

The data set

I downloaded the number of unique addresses from Blockchain.info as well as the total market cap for all time. Then I took the average values from each for every month going back to August 2010, when the market cap was first established.

This resulted in 94 data points going back eight years. Then, knowing that these values are each increasing exponentially, I decided to take the log of each and run linear regression to check for a correlation.

Unique addresses regression output Data Source: Blockchain.info

What I found would seem to confirm my suspicions. There is a correlation between the number of unique addresses and the Bitcoin market cap, and it's a very strong one. I believe this is because the number of unique addresses is one (but not the only or necessarily the best) proxy for the number of people actually using Bitcoin on a given day.

Recall that the Bitcoin network issues new Bitcoin every ten minutes. Right now the amount it issues is 12.5 Bitcoin, which comes to about 1,800 new Bitcoin entering the system each day.

As more users pile in, the scarcity factor drives the price up. If something spooks the speculators, then a price correction could also sideline activity, which would mean the number of people in the ecosystem would drop, and the price would correspondingly be pushed down, basic supply and demand.

Predictive values based off our regression analysis

Using the regression equation, we can feed in the number of unique addresses and chart the predicted Bitcoin market cap against reality.

bitcoin market cap versus predicted Data Source: Blockchain.info

I also took Bitcoin's market cap and divided it by the regression predicted value. Then I standardized this ratio and created this chart below.

Unique Addresses T-Score Data Source: Blockchain.info

You can see that based on this metric, there have been three great peaks, when the market cap of Bitcoin was massively higher than the model would have predicted. Also there have been two long periods of time in which the values were below average and stayed that way for some time.

Another observation is that the spikes are getting more dramatic over time but the dips do not seem to be increasing in magnitude. Using this metric, the last peak in Bitcoin's market cap dwarfed all other previous bubbles.

Keep in mind, this is just one metric, just one way of looking at it.

So, what does the number of unique addresses suggest that the Bitcoin market cap should be right now?

We've already discussed the ways in which the unique addresses could be temporarily shifted up or down and we've talked about what some of the long term trends are. But it may surprise you that using this measurement of the Bitcoin network as a yardstick, the predicted market cap of Bitcoin on June 1st of 2018 is $21,993,572,103.22, compared with the actual Bitcoin market cap of $131,769,788,441, six and a half times higher than expected.

These figures might seem incorrect from looking at the chart, but remember that the axis on the left is in log scale which can be visually deceptive if you're used to looking at linear scale charts. Below is the same chart but with tick marks added to the Y-axis to illustrate this point.

unique addresses with tick marks Data Source: Blockchain.info

The way I look at this metric is just as one piece of the puzzle. I don't know if the value that it expects is appropriate, so I combine this with other formulas, other frameworks and look at them as a group before making a judgement call.

There might be ten, twenty, or even hundreds of analysts setting price targets for a large company. In the same way that it has become popular to look at average analyst rating, I consider this input as not the whole story, but just as one possibility of the boundaries. Some of the metrics we are going to discuss are as predictive as this one and yet they produce wildly different estimations of the price or market cap.

Notice also that there were two long periods of time when the Bitcoin market cap was significantly lower than expected (the green arrows). Those times were around June of 2012 and January of 2016. See Below.

buying and selling opportunities Data Source: Blockchain.info

These indicators would seem to suggest the area surrounding the green arrows are where the most upward pressure was being placed on the Bitcoin market cap. Conversely, the red arrows put downward pressure on the market cap (kind of like two object wrapped by a rubber band that move apart and then snap back in the opposite direction towards each other). Remember a large gap between the predicted and actual values could indicate that:

  1. The market cap of Bitcoin is much higher than we would expect for the number of current unique addresses.
  2. The market cap of Bitcoin is much lower than we would expect for the number of current unique addresses.

Even though the model rarely predicts the exact value of the Bitcoin market cap, when there is a large disparity, it always seems to predict an eminent change of direction.

The two largest deviations from this formula to the high side come from January 2014, and February 2018. In the period following January 2014, Bitcoin's market cap fell from 10 billion to 3.3 billion over the next 15 months. The period following February 2018 is where we currently are now, so we don't know exactly how far down the market cap will go yet. It's worth noting that the high point in the Bitcoin market cap came in December, not February, whereas the signal from January 2014 did seem to coincide with the change of market direction more closely.

The two largest deviations from the regression formula to the low side were in June of 2012 and January of 2016. In the period following June of 2012, the Bitcoin market cap would surge from 48 million to 10 billion in 2013. In the period following January 2016, the Bitcoin market cap would go from a measly 6 billion up to 297 billion in December of 2017.

This is not investment advice

Have I mentioned that this is not investment advice? Use this information with great care.

What are your thoughts? Let me know in the comment box below.

A final comment on the abuse of regression analysis

Some of you with a PhD in statistics might be about to point out that linear regression is not to be used on data that is not normal, and the Bitcoin market cap is hardly normal. Therefore I should be burned at the stake.

This is a valid concern. My use of regression in this unconventional way was to demonstrate that a relationship of some kind exists between unique addresses on the Bitcoin network and the Bitcoin market cap. Using this model might actually help someone in the real world, so you'll have to forgive me of this sin (and in actuality this practice is very common because every time you draw a trend line in excel on time series data you are doing the same thing).

Disclosure: I am/we are long BTC-USD.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.