The introduction of cryptocurrency to the marketplace in 2009 raised many questions for financial analysts, ranging from “What is it?” to “How will it affect the marketplace?” When Bitcoin first became available to buy, sell, and trade on online exchanges the following year, nobody really knew what it would do to asset prices, like housing and stocks, or how to project what it might do in the future. Such projections help inform major decisions in both the micro and macro economies, from whether to buy a home or when to take a company public on the stock market.
While crypto has been studied in the years since, analysts relied on observable factors, such as market value and past returns. IU researchers are taking it a step further, using data from the actual blockchain, which is essentially a secure and decentralized ledger of cryptocurrency transactions.
People trade Bitcoin using various exchanges, like Coinbase. Most people using exchanges are regular people and not large institutional investors, Kantak said. The ultimate holder of the Bitcoin on the blockchain is the exchange, and those exchanges have accounts, or wallets, on the blockchain that are holding coins for all those using the exchange to trade cryptocurrency.
Kantak said that preliminary findings from other studies using less precise data and observations seem to indicate that off-chain, or exchange, activity experiences a great deal of mean reversion, meaning price volatility and historical returns eventually revert to their long-term average levels. This is usually indicative of the exchanges managing their inventory risk, he said. By contrast, on-chain activity shows momentum, meaning Bitcoin prices typically go up.
Kantak and Xioran Cheng, a former Kelley student who first brought up the project idea as an independent study proposal, needed to be able to segment the activity on the blockchain into its various players to determine who is actually holding these coins. These entities could include an exchange, a mutual fund, or a large holder. By tracking on the blockchain how and when Bitcoin is traded, Cheng and Kantak can then see what these factors might mean for future asset prices. To do this, they needed to use Carbonate, IU’s large-memory computer cluster that was designed to support data-intensive computing.
“This is every transaction of Bitcoin going back to when Bitcoin started, so it’s a huge amount of data, and it’s all compressed in these files called BLK files,” Kantak said. The BLK files contain numerous complications when it comes to extracting clean data, he said. For example, spoofing, which is when cybersecurity criminals pretend to be someone else for nefarious purposes, muddies the data. Also, a single user or organization can have multiple wallets, and they must look at how transactions are taking place between wallets and amalgamate them to attribute them to a single user. This will allow them to study the transactions and how such transactions affect asset prices going forward.
The process of downloading and uncompressing these massive files, then opening them, cleaning them up, and segmenting the data is a massive undertaking. The only way to do it is with a supercomputer, Cheng said.
Apart from the work of verifying or contradicting the outcomes of earlier studies, which they hope to finish by this summer, Kantak and Cheng are excited to see what else the exchange and blockchain data can tell them.
“There’s a lot more we’d like to do between on chain and off chain,” Kantak said. “But first we need to go through this segmentation process so that we know who is initiating what trading activity. Without Carbonate, we couldn’t do this.”
