Skip to main content

Last night, seemingly without cause, Bitcoin’s price plunged by 3%, a $6.3 billion dip in its total market cap. Slowly, over the course of the ensuing hours, the news trickled in: cryptocurrency exchange OKEx suspended withdrawals amidst an investigation by Chinese authorities.

Four questions naturally arise:

  • What happens before a Bitcoin price jump or drop?
  • What are the main drivers?
  • Is there any way we can predict it before it happens?
  • Who are the main players responsible for that price change?
Fig 1: An 11% spike in Bitcoin price on July 27th 2020, associated with a $1.63B trading volume

You’ve probably asked yourself one of the above questions at least once. These are the golden questions investors are trying to answer everyday. Retail investors simply hope that the price movement will be positive so that they won’t lose the savings that you decided to invest in cryptocurrencies. But most probably you don’t really have time to follow the market everyday and your investment is mostly a bet on Bitcoin as a whole.

In this article we’ll try to provide a brief overview and visualizations of some of the most important areas to watch out for when making an investment in cryptocurrencies.

The problem can be roughly broken down into 4 main macroareas:

  • Financial/market data
  • Onchain data
  • Order book data
  • Social interaction data

Financial/Market Data

For financial/market data we include all those fundamental variables typical also of more traditional assets like stocks. In these buckets fall, for example, the transaction volume, the number of active addresses, transaction fees, price of other cryptos, and the derivative market.

Fig 2: Candlestick chart with Bitcoin price data of the past 3 months, together with transaction volume (bottom bars) and moving averages (lines)

Onchain Data

Onchain data refers to all the data stored on the public ledger (the Blockchain itself). Analysis of onchain data is very important to understand how the blockchain network is behaving and what different entities are doing on it.

Are whales liquidating funds?

Are exchanges increasing their reserves of cryptos?

This data is usually very hard to read as you don’t necessarily know if an address belongs to a big exchange, Elon Musk, or your uncle Bill. Graphs like the one below are VERY hard to obtain, as you need to catalogue millions and millions of addresses on the Blockchain and label each of them according to the category they pertain to.

Fig 3: Exchange balance of Bitcoin over the past year

At AnChain.AI we have been labeling this huge amount of addresses for the past 2 years.

How can you do that if you’re a retail investor sitting in front of your computer?

Well, historically, you don’t. Unless you want to waste a lot of your time.

The sheer enormity of this task required us to develop advanced machine learning algorithms to classify each address and conduct manual checks to assess their correctness.

Order Book Data

Order book data refers to all the data related to exchanges’ order books. Specifically, an order book contains the list of all the outstanding orders for a specific currency pair. For example, if you want to buy Bitcoin using your US dollars, you usually go on an exchange website such as Coinbase or Binance and place your order to buy (bid) Bitcoin with USD. The exchange connects you with someone who is placing your exact opposite order and executes the transaction.

This data is important to check how liquid a particular market is, meaning how easy it is to convert your digital asset into cash without affecting the market price. Also, by checking existing open orders, you can get an idea of where people are placing their orders and see what the biggest levels of resistance or support are.

Fig 4: Market depth obtained by connecting to the exchange

Social Interaction Data

Today, a huge amount of information about the crypto market is exchanged on social networks like Twitter, Telegram, or Reddit. Macro sentiment trends could be inferred by monitoring people’s level of engagement with particular hashtags and keywords, certain profiles, or websites.

What are the main topics of discussion?

Are today’s comments from people particularly more positive compared to other days?

Is there a big call to sell on a particular platform?

Fig 5: Number of positive comments on the Bitcoin’s subreddit over the past 6 months

Could information from social network provide signals on where the price is about to move?

That’s definitely something that needs to be monitored to have a complete overview of the market.

For example, check in the graph below how a spike in positive sentiment on Bitcoin’s subreddit preceded by a day an 11% increase in Bitcoin price on July 27th 2020! To generate this graph, we used Natural Language Processing models to analyze every comment made on Reddit and give them a score between -1 and 1, where 1 means very positive and -1 means very negative. The results are quite interesting as you can see, but very computationally costly.

Fig 6: BTC price vs Bitcoin’s subreddit Polarity over the past 3 months


For the amateur cryptocurrency investor, gathering this data seems like a fool’s errand. Accumulating this much info could easily take hours.

But, what if it didn’t?

What if you could access all of these datapoints in one place, one platform specifically designed to empower a data-driven trading effort?

Sign up to our newsletter to be the first to find out!