Bitcoin: Analyzing Historical Whale Activity and the Implications
The Bitcoin market is a young market that has been largely controlled by a small minority. This minority, the market “whales”, drive the direction of the market, whether that be up or down. This often results in highly volatile movements, with Bitcoin rising or falling hundreds of dollars in a couple minutes, and even thousands of dollars in a couple days. Considering the importance of this minority in determining trends, I chose to investigate their historical transactions with relation to the rest of the market. Ideally this activity will give us insight into the future direction of the price trend. I sourced my dataset from kaggle. This dataset was acquired and put together with data from blockchain.com.
After cleaning the data I started exploring with a straightforward chart comparing the daily number of transactions by the top 100 addresses with Bitcoin’s price on a logarithmic scale. I chose to use a logarithmic scale for price because Bitcoin’s growth over the last 7 years has been exponential and as a result any intermediate market trends would be unclear. This chart is intended to show us if there is any relationship between Bitcoin’s price movements and the number of transactions.
We see some sharp spikes in the number of transactions, especially between 2012 and 2014, as well as some near the end of 2017. Some of these spikes also seem to precede strong upward price movements, which could suggest that increased whale transaction activity foreshadows a strong uptrend. This would make sense as whales are likely to be buying and moving their Bitcoin more often if they expect renewed interest in the market. As a side note, the period of noticeably lower transaction variance in early 2017 is a result of smoothing the data with a SMA as it seems to have been inconsistently recorded during this time period.
My next step was to look at these changes in whale transaction activity in comparison to the rest of the market. The whales are naturally incentivized to extract as much wealth out of the market as they can, so it is in their best interest to be fooling the rest of the market with unexpected changes in price trend. This allows them to buy low and sell high in large quantities.
It is apparent that there is a significant historical difference between these two groups, though that difference seems to be decreasing over time, likely signifying more widely distributed ownership. Between 2012 and 2014 the top 100 most popular addresses were at their most active, while those outside the top 100 are at their least active. This makes sense considering Bitcoin was only known in niche communities at the time. There are also a few instances where a spike in the top 100’s transactions is followed by a spike in those outside the top 100. This is also a good sign for our theory, since it shows the whales tend to lead the pack.
Taking these last two charts into consideration it seems that we may be able to use this data for gauging Bitcoin’s future trend direction. After trying a few different time periods I decided to plot a 90 day rolling correlation between the popular and the non-popular addresses, since 90 days is short enough to identify the trend early but also long enough to filter out any short-term noise. I then imposed that on Bitcoin’s log-scaled price and marked any periods where the correlation was especially high or especially low. This is intended to show how the whales’ tactics and Bitcoin’s price changes are related.
Here we can see that the correlation(or inverse correlation) between the two groups has a relationship with the future price movement. The green zones(>0.4) represents when the transaction activity of the two groups agree, which means the trend is likely to continue. The yellow zones(<-0.4) signifies the two groups disagree on things and the trend is changing, leaving the non-whales with their hands in their pockets. From this analysis we can conclude that the correlation between whales and non-whales transaction activity is of significance and can be helpful in determining long term market movements. Bitcoin’s price is apart of a financial market so naturally this is not a perfect metric, but it can be used in conjunction with other indicators to gain a probabilistically-based idea of the future trend direction for the Bitcoin market.