Crypto-ML is Radically Changing
Last Updated on December 9, 2021 by admin
Effective immediately, Crypto-ML will no longer provide trade signal and Auto Trade services. This is a major change that comes with mixed emotions. But with this change, Crypto-ML will be able to provide much higher value services to our customers, focusing on the rapidly-evolving areas of our decentralized future.
Over the last 5+ years, we’ve invested our minds, hearts, and souls into our trade signal and machine learning technology. We’ve continued to chip away at edge cases. We’ve consulted with industry experts. We’ve uncovered novel methods for handling tough scenarios.
Through all of this, we’ve learned an exceptional amount about how markets work and how to best engage with cryptocurrency. This type of experience can only be gained by being ruthlessly transparent, posting each and every trade publicly, and having a clear feedback loop.
This post will cover some of our next steps, the reasons behind this change, and what you need to do.
Deeply Engaging in a Decentralized World
As we move into this coming year, Crypto-ML will transition from providing trade signals to providing insights on how to positively interact with cryptocurrency while also maximizing the potential returns on your investments.
This means diving into:
- Identifying the best approaches for long-term investing
- Learning how to analyze crypto project fundamentals
- Maximizing returns with DeFi, including yield farming, staking, liquity mining, and more
- Participating in Web3
- Optimizing your safety and security
With Web3, we are at the precipice of a major shift that stands to radically change our technological landscape. This presents a once-in-a-lifetime opportunity for many of us.
The challenge is the landscape is evolving rapidly. Without dedicated focus, it’s difficult to keep up and ensure you are optimally positioned. We seek to provide human-based guidance while leveraging AI as a tool to cut through the clutter and deliver data-driven insights.
Our hope is to provide clarity and confidence to those who believe in a decentralized future.
Current Auto Trade Customers
If you are an existing Auto Trade customer, please take action.
- Turn off Auto Trade from our Auto Trade Dashboard.
- Remove your exchange from our Exchange Setup.
- Delete the API key from your exchange (Binance or Coinbase Pro).
Finally, please watch your email for additional instructions on handling your subscription.
Crypto Trading Challenges
For context on this change, we want to provide some insight into the challenges trading systems face in the crypto markets. There is considerable nuance to each of the points below, but this will at least provide a high-level view.
Our models have been successfully predicting upwards of 98% of real-life price movement. That is an amazing accomplishment in the realm of machine learning. However, it turns out the small exceptions–even that 2%–really matter. Those exception events have the ability to cause havoc and erase months of progress.
In this business, the edge cases can be devastating.
Large Intra-Day Movement
Risk management is the key to any successful trading system. Cryptocurrency, however, has intra-day ranges that exceed the tolerances of most risk-management systems. This means if you use traditional risk management techniques, you are likely to get stopped out at an unacceptable rate.
Our team has tested many types of risk management alternatives. Some of these systems are able to hold through drops that rebound; however, those can cause situations with unfavorable risk-reward ratios and expose open positions to escalating losses.
The crypto markets are also notorious for lightning-fast, double-digit drops. Crypto-ML has regularly built solid gains over a month or two, only to then have those gains erased in a flash 15-minute drop.
To find comparable losses in the stock market, you have to look over the course of 100 years at events such as The Great Depression, Black Monday, the Dot-Com Bubble, and the Financial Crisis of 2008. By contrast, you will find many examples of Bitcoin’s fast drops within just the last year.
Utilizing risk management tools does little to help and can even harm your account in these situations. For example, if you have a stop loss set to exit at -3%, your order will likely still be filled at double-digit negatives because of the speed.
This particular problem has garnered most of our attention over the last couple of years. We have worked with experts, technical indicators, and off-exchange data to create models that attempt to predict these drops. Our finding is the patterns are insufficiently consistent to predict these events in a reliable manner.
As we look at these fast drops, we need to acknowledge the fact that the cryptocurrency markets are still largely unregulated. Additionally, certain market participants have asymmetrical power in terms of information, fund, and technical access.
Whereas traditional markets are highly regulated and have mechanisms such as circuit breakers and plunge protection, crypto markets have:
- Limited reporting and transparency requirements for exchanges and participants.
- Whales that move large quantities of funds, triggering flash crashes, arbitrage, and panic.
- Regular pump-and-dump schemes.
- Native high-frequency trading bots.
- Other collusion events such as whale wall spoofing, wash trading, and stop hunting.
As an example, Bitcoin dropped 22% in a couple of hours this weekend. Around the same time, Tether minted $1,000,000,000 USDT (tronscan.org transaction details) with only a few lines of code. That amount of USDT was enough to buy 23,809 Bitcoin. Only a handful of people on this planet can afford to buy $1B of Bitcoin. Yet this financial power was generated instantly without any transparency.
Were the minting and flash crash events coordinated or coincidental? With manipulation, it’s never quite certain.
But what we can say for certain is the cryptocurrency market affords some people the ability to generate huge amounts of money through exploitation.
Practically speaking, most traders will face numerous adverse events that do not typically occur in traditional markets. These events are hard to predict and can be disastrous.
Can Machine Learning Beat The Market?
While this post may seem very about the ability to use machine learning to deliver consistently good results, that is not our intention. The challenges we highlighted likely can be overcome.
In fact, cryptocurrency trading offers numerous advantages for machine learning-based systems:
- Data availability: many data sources are available to build deep pictures of the market. These include readily available exchange, on-chain, off-chain, sentiment, search, social, and macro data points. This gives dedicated machine learning engineers ample data to work with.
- 24*7 operation: since crypto exchanges are “always open,” data comes in a consisent stream. Traditional markets introduce complexity with open times, close times, weekends, and holidays.
- Ample volatility: while both a blessing and a curse, crypto volatility presents exceptional opportunity.
If you’re a trader, watch for other projects to crack the code on the challenging edge cases we’ve described. In the upcoming months, we will provide deeper dives to arm you with sophisticated methods for analyzing these systems.
Last, our platform targeted a variety of time frames but primarily focused on swing trading. Successful machines will likely operate best in the high-frequency trading space, scalping tiny irregularities in exchanges. These types of machines are not practical for retail traders.
Please stay tuned for updates as we progress and evolve.
And regardless of this change, our slogan still stands. We want to level the playing field and ensure you have the best possible opportunities to participate in our decentralized future.
Let us know your thoughts in the comments below.
Crypto-ML provides machine learning for crypto traders and investors. Gain crystal-clear signals and deep market insights. Add predictive capabilities to your toolbox. Learn more and join for free.