API Case Study: Clustering Digital Assets with SynergyCrowds

May 16, 2019

API Case Study: Clustering Digital Assets with SynergyCrowds

Now more than ever, market participants require advanced and trustworthy tools to make informed decisions. However, it is difficult and costly to access reliable information for a range of market activities, including risk assessment, price discovery, portfolio management, and correlation studies.

Our partner SynergyCrowds is attempting to improve decision making in the cryptoeconomy by building a decentralized knowledge platform, accessible to even the smallest market participant. Various applications will run on top of this platform providing market information to end users. One such application is an AI powered clustering analysis tool. This clustering tool uses Kaiko cryptocurrency market data sourced directly from our API.

What is clustering?

Clustering is a type of analysis used to split a set of elements into groups of resembling individuals. The aim is to group individual elements such that objects in the same cluster are more similar to each other than in different clusters.

Clustering can be compared with classification, which is an analysis used to split elements into groups based on a pre-defined set of categories (such as good vs. bad, profit vs. loss). The major difference between clustering and classification, though, is that before a clustering analysis is run, the predefined groups are unknown. The only thing known is that a group of individual elements need to be split in some way. Ultimately, the goal is to build groups as different as possible that contain very similar elements within the groups.

For example, imagine clustering a crowd of people at a football stadium. You could cluster them based on the color of shirt they are wearing, or by gender, or by age. There are a number of different clustering strategies that can be applied to the same group of individual entities.

Clustering Digital Assets with Kaiko Data

SynergyCrowds applies the concept of clustering to cryptocurrency markets by examining the price of digital assets over time. Because there are thousands of different assets, it is difficult to perform large pricing analyses across all pairs or specific assets trading against a reference. Clustering is a powerful tool to independently group assets based on their price evolutions.

Synergy Crowds’ AI clustering tool is powered by Kaiko’s VWAP [Volume Weighted Average Price] endpoint from our Market Data API. They use Kaiko daily VWAP data on:

  • 3169 assets
  • 51 exchanges
  • 3531 instrument pairs

For each instrument, they aggregate the volumes and prices across exchanges. They check a total of 6090 exchange-instrument combinations every day using our API.

Each clustering analysis is performed on a 50-day timeframe using historical daily VWAP data across all pairs.


SynergyCrowds discovered clusters of similar crypto assets. Their similarity is computed not only by comparing prices at certain times, but also revealing possible dependencies in their behavior over that time frame. They identified three different clusters:

  1. Climb — cryptos with an ascending trajectory
  2. Immobile — cryptos with a flat trajectory
  3. Fallback — cryptos with a downwards trajectory

Each clustering analysis is performed against a reference asset such as BTC or ETH. This allows users to investigate clusters of pairs trading against a specific reference asset. For example, if a user selects USDT as the reference asset, they can easily compare all discovered clusters for assets trading against USDT.


This case study sought to show how one of our partners, SynergyCrowds, uses our Market Data API to build data-driven applications to provide value to the wider cryptoeconomy. Clustering digital assets can serve a variety of use cases, specifically in terms of portfolio optimization. This application allows market participants of all sizes to better manage and understand the dynamics of their portfolios.

To read a full explanation of SynergyCrowds clustering, check out their medium post here.