Tonight’s business school homework: what is the value of bitcoin in 2023?

My first assignment for a “FinTech Innovations” course is to make a prediction: What do you think the price will be five years from now, in…

Feb 14, 2018

Background

The Ross School of Business at the University of Michigan is promoting the education of financial technology (FinTech), specifically blockchain technology, through new initiatives in the past six months.

Ross School of Business’ 3-Prong Approach

(1) Classroom lectures: FinTech Innovations a 3-credit elective cross-listed in finance and technology co-taught by Professors Andy Wu and Bob Dittmar (class name TO 428)

(2) Global experience: FinTech Fieldwork with Alphabet (Google) in London, United Kingdom

(3) Action-based learning: Wolverine Crypto Trading where 20 students learn weekly through discussion and present monthly to the UMich and Ann Arbor community. Each member invests $1,000 per semester from the fund, making their own investment decisions.

I am appreciative to be involved in all three initiatives!

Turning in my homework

So, back to the FinTech Innovations course assignment to write a 2-page analysis responding to: What do you think the price will be five years from now, in 2023? (Assertions backed up by concrete data and analysis will receive more points.)

In short: Bitcoin has value because people think it has value. I’m going to walk you through my three favorite ways to value Bitcoin:

(1) Metcalfe’s Law as a Model for Bitcoin’s Value

by Timothy F. Peterson, published October 9, 2017

Bitcoin’s price is best modeled as a network.
To measure the size of the network, we should look at the number of wallets created. Wallets to not represent users, as each person/organization can have more than one wallet. However, the number of wallets is a signal of growth.

We have data on wallets from November 29, 2011, when two wallets were created.

Bitcoin’s value is based on the community that uses it. The more transaction pairs you have (people you can transact with), the more value the digital asset has.

There are five ways to acquire bitcoins: mining, accepting them as payment, purchasing them in the open market, accepting them as a gift, or stealing them. In every case, one must first have a wallet.

As the number of wallets grows, the transaction pairs increases exponentially. See the graphic below where (from row 2 to row 7) transaction pairs in millions grows 1,300-fold when the number of wallets increases 34-fold.

12 months of wallet growth compared to the (faster growing) transaction pairs from December 2011 to December 2012

The value of a Bitcoin should align with the transaction pairs. This is (roughly) n² where n represents users.
You might have heard about the network effect before. That’s because of a guy named Robert Metcalfe who published his ideas as… Metcalfe’s law!

Regression summary over time.

TL;DR: The value of Bitcoin will grow as users grow. User growth is determined by the number of wallets created (a signal for number of users). This is 84% accurate when looking at the historical price of bitcoin… and that’s pretty good!

(2) Predicting Bitcoin Price Fluctuation using Twitter Sentiment Analysis

by Evita Stenqvist and Jacob Lönnö, published June 14, 2017

Swedish Masters of Computer Science students analyze the sentiment of opinions on Twitter regarding Bitcoin, compared with Bitcoin’s price.

Problem Statement
• Is there a correlation between Twitter sentiment and BTC price fluctuation?
• Can a naive prediction model based on sentiment changes yield better than random accuracy?

What is sentiment analysis?

In a nutshell, sentiment analysis is about finding the underlining opinions, sentiment, and subjectivity in texts, which all are important factors in influencing behavior.

Examples of how Tweets were classified using a model called “Vader”

Research Design

The researchers tried 5 minute, 15 minute, 30 minute, 1 hour, and 2 hour timeframes. In the 5 minute time frame, the accuracy hovered around 50% as the social media data did not have significant predictability.

5 minute intervals show small accuracy variance

In 1 hour we see wide swings in the accuracy of the algorithm to predict the price of Bitcoin

Reconnecting with the Problem Statement

Is there a correlation between Twitter sentiment and BTC price fluctuation? It depends. They were able to get 83% accuracy (in very specific conditions), but cannot identify magnitude. The algorithm only predicts if the price will go up or down.
Can a naive prediction model based on sentiment changes yield better than random accuracy? Yes. Random correlation is 50% and they were able to achieve above 70%

On the left, the daily number of Tweets mentioning Bitcoin. | On the right, the daily number of transactions on the Bitcoin network.

TL;DR: Sentiment analysis is indicative of price as they move together more (positive thoughts -> higher price) often than not. It is not inspiring in the short-term prediction accuracy and is useless for long-term trends.

(3) Predicting Bitcoin Price Through the Futures Market

Cboe bitcoin (XBT) futures experienced 30% high-to-low range during last week’s trading. There’s limited data, but in recent weeks, XBT futures have had a positive correlation with the S&P 500 (SPX). The current sell-off in bitcoin is the most significant, since the drawdown came to a conclusion on January 14, 2015. The current peak-to-trough move in spot bitcoin is 70%. Over the same time frame, XBT February futures have fallen from 21,250 to 5,880. That’s a decline of 15,370 points, or 72.3%, slightly more than the spot market.

Bitcoin commonly has 30%+ sell-offs. The last 8 weeks do not shock long-term bit-lievers.

“As you can see in the chart below, Cboe Futures Exchange (CFE) has seen a steady uptick in trading volume since XBT futures were introduced two months ago. ADV in the month of December for all XBT futures contracts was 4,007. ADV from inception through January grew to 5,899. If we incorporate the February volume to date, ADV in XBT futures is running 6,449 (through February 7).” wrote Kevin Davitt — Cboe Senior Instructor

Average Daily Volume (ADV) is increasing for Bitcoin futures. The market is increasingly interested in trading futures meaning that investors feel that they have a better prediction than their peers when trading futures contracts.

TL;DR: The market of institutional investors clearly does not agree what will happen with the price of Bitcoin. Futures contracts rise and fall in predicted value as the price changes. It seems like everyone is a chicken with their head chopped off, throwing darts at “carefully analyzed” price targets.

Conclusion

In 5 years the price of bitcoin will best be predicted through our first approach, Metcalfe’s Law. I believe that the number of users (represented by wallets) will grow exponentially. This will be spurred by existing companies that we do business with enabling everyday users to start a wallet. The best examples are Square Cash and Apple Pay.

To quantify the growth of users, I rely on Blockchain.info to gather the data on the blockchain wallet users. Below is a graph of the wallet growth from 2012 to present.

In order to extend this line for another 5 years, I used the Growth function within Excel. This takes 3 arguments: range of wallets, range of dates, date to predict wallet number at.

Rows 1496 to 2831 represent the last 12 months of data on the number of wallets. 2832 represents a prediction based on exponential growth as of March 1, 2018.

Predicting based on the last 12 months of wallet growth, there will be 588,500,018 wallets on February 1, 2023. (Check out how I got there with this link.)

I predict Bitcoin wallets will grow from 23 million Feb 2018 to 589 million in Feb 2023.

I project 24.5x wallet growth in 5 years. I attempted to fit this into Metcalfe’s Law, however found myself needing to fill in the log of the bitcoin closing price in order to complete the calculation. My (humorous) attempt at solving this can be accessed via Google Drive.

If we square this to represent bitcoin value growth, we get 600x growth. Given the price of one bitcoin about $10,000 today that would mean $6 million per bitcoin or a total market cap of $126 trillion. Obviously, that’s insane. (It would be amazing if I eat my words in 5 years and this isn’t insane.)

I am thus going to default to a simple linear relationship (after all that!) and say 24.5x wallet growth will lead to 24.5x price growth and this:

In 5 years the price of bitcoin will be $245,000!

I welcome feedback and criticism to this piece in the comments section below. Definitely a blast to write! Thanks to Professors Andrew Wu and Robert Dittmar at the University of Michigan for the prompt.

Fact of the Day 1

Discussion about this post