Last Updated on
The Crypto-ML team has continued to stay busy developing new avenues to generate exceptional trade alerts through machine learning. Here, you’ll find an update on the items in development.
Cryptocurrency Day Trading
As initially announced here and updated here, the Crypto-ML team’s top priority has been the development of a day-trading platform for users. Day trading offers the ability to capture short profit spikes, even during larger downtrends.
While Crypto-ML currently provides users the ability to check in once per day and swing trade, the vision for the addition of day trading was as follows:
If users have a couple of hours free throughout the day, they should be able to log into the system, make a few trades, and earn real income.
The major difference between the current Crypto-ML signals and the new day trading signals is the data capture. The current platform pulls data once daily. The day trading version pulls data continuously, minute-by-minute.
As we have continued to test results against this new data set, we have found the limiting factor is the human. This is because the regularity of quality signals is inconsistent. If we try to force a certain number of trades into a small time period, profitability drops off significantly. However, if you remove frequency as a limiting factor, profitability goes through the roof.
Imagine a platform that delivered quality “buy” or “sell” signals every 15 minutes. In this case, human interaction would make sense. But what if some trades were open for minutes and others open for hours? As a user, you wouldn’t know what to expect or how long you’d be trading. What would happen if you saw a “buy” signal, waited for 3 hours, and then had to leave your computer to go eat dinner? You may miss the corresponding “sell” and forfeit gains.
The reality is we need to provide this freedom to the machine-learning model, but that makes the original vision impractical.
Therefore, to have a great solution, the answer is to remove the human and introduce automated trading. This is a new engineering skillset that is not currently on the team. The right person has been identified and development should be underway shortly.
We initially targeted having this feature live by the end of the month (June), but that will not be plausible. We expect a quick turnaround and hope to have a concrete update by mid-July.
It should be noted that the trading logic and machine-learning aspect of the day trading platform are complete and ready. It’s the final linkage to execute trades that remains.
Bitcoin Machine-Learning Algorithm
As you may have noticed, the Bitcoin trading signals have been a bit “trigger happy” recently, resulting in losses. It’s as though the machine learning models are struggling with the current market circumstances, indecisively moving in and out of trades rapidly.
Our users understand these triggers are generated by a machine and not a human. Sometimes these signals defy expectation, but upon reviewing historical data, you will find that rapid trades (often for losses) are followed by large spikes up and gains captured by Crypto-ML. While the future doesn’t always follow the past, patterns absolutely repeat and we could find ourselves on the brink of another run.
Regardless of this expectation, the Crypto-ML team has spent considerable time evaluating the Bitcoin model from a governance perspective, ensuring it is optimized for today’s market conditions. In fact, whether focusing on near-term data or looking at bigger-picture data, the Crypto-ML models for BTC remain consistent and exceptional. Where we have made major changes to machine-learning models in the past, no such change is currently warranted. The platform is optimized and we eagerly await the next major win.
Futures and Forex Machine-Learning Trading
In an effort to continue to add value and grow the capabilities of Crypto-ML, research has expanded into other instruments, including Futures and Forex. This means regardless of the cryptocurrency market, you may have access to signals on the stock market, futures contracts, and foreign exchange pairs.
Our research includes evaluation of:
- EUR/USD foreign exchange pair
- GBP/USD foreign exchange pair
- USD/JPY foreign exchange pair
- /WTI Crude Oil futures
- /NQ Nasdaq futures
- /ES S&P 500 futures
Look forward to updates on this topic and expansion into new financial instruments soon.