Travelling in the BlockChain Ecosystem with Python

Travelling in the BlockChain Ecosystem with PythonZecca LehnBlockedUnblockFollowFollowingJan 30Aggregate Market Caps of all CoinsWith over 2500+ active blockchain projects around the globe, each with it’s own unique statistical characteristic, we rarely see a top level analysis of the overall crypto market, because cleaning and collecting the time series is too time consuming.

And on the retail side, we don’t have a clear set of functions to collect, clean, and explore the critical data needed to customize portfolios from the blockchain ecosystem.

The following blog post will start as a strong foundation for more in-depth quantitative approaches related to things like volatility, clustering, forecasting, and log return based portfolios using data science and quant strategies.

We now have access to aggregate and exchange specific trading data through CrytpoCompare with their excellent API — it’s free for retail investors.

This post will walk through parallelized downloads of long dated time series, with speedups of 8x (depending on the # of cores on your machine) , and I’ll explore other practical stats from each available symbol (ticker).

The post will explore the nature of tokens which utilize protocols (i.


, platforms) such as Ethereum, with tokens based on ERC20 among others.

We’ll see that the current blockchain market currently stands at $150,404,143,218 ($150 B) — still a fraction of other alternative markets.

First things first, you’ll want to download Anaconda on your local machine, and set up a conda with Python 3.

5+ in an environment, then launch a Jupyter Notebook to run the code below chunks.

Better yet, if you haven’t already tried, run the following code in Google Collab for free.

Next, we’ll find the number of token projects per protocol.

Counts of Tokens per ProtocolNote here, there are 1700+ ERC20 tokens under the Ethereum protocol.

WAVES and NEO are next with 53 and 25 tokens respectively.

We’ll then create a TreeMap to visualize the proportions across different protocols groupings.

Token Ratios among Platforms with >1 CoinWe can clearly see a large disproportion of Ethereum ERC20 based projects over all other blockchain protocols in the market.

These projects helped fuel the ICO market in late 2017 through present.

When first building the code for this post, I then tried downloading all of the listed 3500+ assets, I’ll include the code here for those who have trouble with concatenating all of the crytpo aggregate prices.

Then we’ll move on to downloading all of the historic assets in one parallelized routine.

Wall time: 5min 17s!!.for 150 series!!Now we’ll plot showing the top 150 assets where the color coded bins represent 1 or more tokens within the respective project.

Note, the high market cap projects on the left generally have more than 1 associated token from their root protocol.

We’ll then drill into all of the available tokens / protocols in a parallelized routine — this will speed up our processing time dramatically.

Create a binary plot to show when projects started from 2014 through 2018, sorted by market caps.

Here I’ve plotted all 2600 assets with available price data on CryptoCompare.

The other 1000 plus assets are either pre-ICOs or they are currently not collecting aggregated exchange data per symbol.

It may be worth noting, there are plateaus in May 2017 and Oct 2017/2018 where many new projects came online.

Some higher market cap projects grew tremendously fast based on their release dates, where others with similar release dates have not grown.

Next we’ll look at where the large market cap ERC20 tokens are in this distribution.

The plot here shows all 29 protocols color coded.

Because ERC20s rule the current ecosystem, we see them in yellow to the right.

They are lower marketcap, and were largely released into the market from the peak of the crypto market in late 2017.

This explosion of choice could be a large reason why the crypto market declined so rapidly, as so much capital started to spread across many token offerings.

In a recent article in TechCrunch, we’ve seen “1,000 projects have failed so far in 2018”.

However, as of 1/30/2019, there are still 2633 active projects.

Just to put things into perspective, Forbes reported 800 ICOs were offered in 2017 alone.

Here is a random sample of 50 projects and their respective prices since 2014.

These assets are all high in positive correlation.

Prove it to yourself with the above parallel code.

We’ll pull some code from a gist I adapted from StackOverflow using Bokeh module in Python.

It’s great for exploring multiple time series using D3— with zooming functionality and hover based info.

And again, we’ll review the top assets by marketcap.

What about volatility??.We’ll explore that in a future post.

Now we want to pull in the Circulating Supply in parallel from CryptoCompare, and merge it with our original stats, which only includes Max Supply.

The former is used currently as the widely accepted method of measuring market cap of a digital asset.

For a more in depth discussion of future and prospective measurements which may be more suitable in blockchain valuation — see this excellent post by Nomics.

Another popular site Coinmarketcap.

com uses Circulating Supply for their marketcap calculations.

Now merge the stats for exploring aggregated marketcaps by protocolLet’s now plot the market caps by each individual coin marketcap.

We see that Bitcoin (BTC), Ripple (XRP), and Ethereum (ETH) still dominate on a market share basis.

And, after we’ll show what the market caps of all tokens within a protocol ecosystem add to the mix.

Market Cap Shares of Aggregate per CoinNext we show marketcap on a protocol grouped basis by marketcap.

Now ERC20 based tokens start to crowd out the space.

CryptoCompare appears to have some tokens with anomalous outliers within marketcap calculations.

I hope they soon fix these errors, which were filtered out in the code below for symbols such as:“BITCNY”, “BTF”, “NPC”, “MTN”, “DOT”, “I0C”,”AMIS”, “WBTC*”, “HIVE”, “OCC”,”SHND” (they appear to be inaccurately reflected in marketcap calcs)ETH is now a close second to BTC on a grouped by protocol marketcap basis.

Source: https://giphy.

com/explore/bravocadoThanks for taking the time to explore the blockchain ecosystem together.

Looking forward to answer any questions you may have.

The above code should be fully reproducible for some time.

If you have any suggestions for posts using the above data, please share your ideas.

.. More details

Leave a Reply