• | Data Science | Web 3.0

Use Machine Learning to Find the Best Price for Your Virtual Land in Decentraland

Users have spent millions of dollars buying and selling virtual land parcels in a blockchain-based virtual world called Decentraland. We designed a machine learning app to help users find the best price to sell their land for, or assess if a land parcel they wish to purchase is priced fairly.

This app can reduce the entry barriers for new users as they no longer have to study the market deeply to make sure they are purchasing a reasonably-priced land parcel. Experienced users can combine their knowledge with the predictions of our algorithm to make better buying and selling decisions.

The figure above shows the interface for our Decentraland LAND pricing app which generates a demand curve - predictions of the probability of a successful sale as a function of price - for a given land parcel. Users select their desired output currency (MANA or USD) which will affect the displayed units on the horizontal axis. They then enter the token ID of the land parcel they wish to price. The token ID is the string of numbers after the last "/" in the web address of the land parcel on the Decentraland Marketplace or on OpenSea. Users can also generate the graph for a random land parcel by clicking on "Random Token ID". Finally, the "Token Description" link takes users to the OpenSea webpage of the land parcel which contains a description of its characteristics and any historical pricing data, if available.

The generated graph shows how the probability of a successful sale declines as the price increases, as one would expect. Most LANDs sell within one weeks if they do sell, so if you see that your LAND is not selling within this time frame it may be necessary to lower its price.

Users who want to sell their LAND quickly should offer prices towards the left side of the graph, where the sales probability ranges from 70-100% depending on the market conditions. Prices which imply a low sale probability towards the right side of the chart are usually too high. Intermediate price ranges with a sale probability around 40-70% tend to reflect the "true" value of asset. Sellers can choose the price they wish to set given their willingness to risk the possibility that the sale does not succeed. Buyers can check the chart to confirm that the LAND they're considering is not priced too high, or is even being sold at a discount and may fly off the market very quickly.

Try the app here: Decentraland Pricing App

Note on estates: Users who wish to price estates, which are a combination of individual land parcels can use the app to obtain a demand curve for a single land parcel in the estate, and multiply the numbers on the horizontal axis by the number of land parcels in the estate. As land parcels close to each other tend to generate similar demand curves, this procedure is an approximate way to find the demand curve for an estate consisting of multiple adjacent LANDs.

Note on updates: Remember to check the "Last Update" date in the app notes. If it's too far in the past (>1-2 months) and the market's behavior changed significantly in the meantime then the estimates may not be reliable.

43 views0 comments