Market Pair Ranking & Confidence Indicator

Volume inflation has been a problem that CoinMarketCap has been actively addressing since 2018. We have noted that some of the exchanges listed on CoinMarketCap have reported inflated trading volumes to give the impression of legitimacy and a false sense of liquidity in markets. This has, in the past, contributed to eroding confidence in the cryptocurrency industry and has, at times, misled both investors and traders.

CoinMarketCap realizes the severity of the situation and has worked tirelessly on a comprehensive and algorithmic solution to the problem, as is our standard practice. After extensive research and crunching reams of data, we identified two main drivers of volumes — liquidity of markets, and the number of traders on the exchange. In general terms, within a retail-driven crypto trading market, it is difficult to have high volumes without (i) a large number of traders and (ii) liquid markets.

With modifications to our Market Pair Ranking algorithm and introduction of the Confidence indicator, we strive to achieve two main objectives:

• Rank market pairs of all cryptoassets using a combination of Reported Volume, Liquidity Score and Web Traffic Factor.
• Highlight the confidence level we have that the exchange’s Reported Volume is accurate.

Methodology - Market Pair Ranking

Market pair ranking will be done by taking the key variables of Reported Volume, Liquidity Score and Web Traffic into consideration. The algorithm takes all these factors as inputs into a machine learning model, and automatically ranks market pairs based on the dynamic weights attributed to each of these criteria. The top market pairs for each cryptocurrency will be those that have the highest relative Reported Volume, Liquidity Score and Web Traffic, instead of simply being weighted on Reported Volume alone. This triage of factors, fairly calculated by an automated machine learning algorithm, is expected to provide a more comprehensive picture of each market pair, ensuring that users make better decisions on where to trade.

Methodology - Confidence Indicator

We currently report the liquidity of all market pairs using our Liquidity Score, and estimate the number of traders on the exchange using our Web Traffic Factor. Taking these factors into account, together with time and sales, we then construct a machine learning model to estimate volumes of every single market pair that exchanges report. With estimated volumes, we can then detect outliers where exchanges report far higher volumes than the model predicts — based on the amalgamation of all these factors involved — allowing us to flag them accordingly via our Confidence indicator. 

The Confidence indicator will be categorized into 3 bands:

• High            >75%
• Moderate   50% - 75%
• Low             <50%

This Confidence indicator will provide our users with a quick and easy way to understand the level of confidence we have in the market’s reported volume. 

This model will gradually improve over time with more data being ingested and processed.