Measure the liquidity (market depth) of all cryptocurrencies using the measure "how much can I buy/sell of a given cryptocurrency while only moving the price x%?" (where x is currently set to 5). The more liquid a cryptocurrency is, the larger the quantity one can buy/sell while moving its price by some fixed percentage.
Build a weighted directed graph, where the nodes are (crypto)currencies (e.g. "USD", "BTC", "ETH", etc.) and the edge from e.g. "USD" to "BTC" constitutes the sell orders/asks of a BTC/USD orderbook (because the sell orders of a BTC/USD orderbook are consumed in order to move from "USD" to "BTC"). Conversely, the edge from "BTC" to "USD" constitutes the buy orders/bids of a BTC/USD orderbook (again, because the buy orders in a BTC/USD orderbook are consumed in order to move from "BTC" to "USD"). The edge weight is the inverse (1/n) of the amount n that you can sell/buy into the given orderbook while only moving the price at most x% (currently x=5).
By
- Saving the result from applying Dijkstra's algorithm with source vertex "USD" and target vertex "crypto_currency" (to find the most liquid path from "USD" to the given cryptocurrency)
- Removing the edge(s) returned by 1
- Repeating
we discover, in descending order of liquidity, all the paths from a given currency to "USD". We repeat this for all cryptocurrencies, in order to find how much one can sell/buy into a given cryptocurrency while only moving the price the specified percentage (using all available exchanges and markets).