Trade against the Machine

Jihuana Xu and Benjamin Livshits developed machine-learning models to predict upcoming pump and dumps.

The Bitcoin trader looked at coins that are listed at Cryptopia

“On average, we have 358 Bitcoin trader for a pump, according to onlinebetrug and one of which is the actually pumped crypto currency. The number of coins taken into account varies for each event due to the constant listing and delisting activities of the exchanges. The full sample contains 47,487 pump and coin observations, a total of 133 pump and dump cases were observed […]”.

In total, the researchers fed their models with data from 46 categories, such as the market capitalization of the coin, whether it had ever been pumped, the age of the currency, the transaction fees on Cryptopia and the volatility of the price. They used both the Random Forest classification method and general linear modelling approaches. The training was successful with various models. Between October 31 and November 4, Jiahua Xu and Benjamin Livshits were able to correctly predict five out of six pumps.

From this, they developed a trading strategy in which the weighting of a portfolio is based on the probability of a pump of the individual components:

“[…] a simple but effective trading strategy that can be used in combination with the forecasting models. Out-of-sample tests show that a return of up to 80 percent can be consistently achieved over three weeks, even under conservative assumptions.”

Those who would like to deal more intensively with the interface of machine learning and crypto-trading should warmly recommend reading the work of Jiahua Xu and Benjamin Livshits.

The Anatomy of Crypto trader currency

This is a fascinating read for those crypto trader who want to learn more about the predictions of crypto courses. Pump participants are generally aware that admins benefit most from a pump. Usually the admin is able to sell his previously hoarded coins to other group members at an inflated price during the pump.

The participants trust in the “theory of the greater fool”, that there is still a dumber one to whom one can sell one’s inflated coins profitably.