Trading with a machine learning approach is a way to be protected in crisis market

Today, many people ask us what we think about the high volatility on world markets, about forecasts and the most important question: how to avoid financial losses. In this situation good advice will be to diversify the portfolio correctly. For example, to reduce part of the investment in S&P 500 and US stocks, buy additional currencies such as Swiss Franc, buy commodities (such as gold and oil). But all these decisions will be partially correct, because there is no guaranteed safe investment tool. Anyone who creates such portfolio should know and consider a number of potential problems.

1. Create the wrong diversified portfolio. There is the risk of creating an improperly balanced portfolio. In this case, losses on one of the assets may exceed the profit received from other assets.

Recommendation: Keep in mind possible volatility and asset characteristics. As a simple and efficient solution, we recommend using Markowitz portfolio theory to check your portfolio allocation. Validation on historical data may prove your hypothesis or highlight possible issues.

2. Risk to buy on the overbought or sell on the oversold market.

Global markets are constantly exposed to short-term speculation and high volatility movements, which as a result have no global consequences. For example, a similar situation was on the S&P 500 index in December 2018. A short-term drop in prices led to panic sales. Investors lost money due to the sale of S&P 500 at low prices.

Recommendation: Most transactions on the world market are performed by automated trading algorithms, it is difficult to compete with such algorithm by using manual trading. We recommend using automated solutions to validate your hypothesis, use statistical indications or pattern detectors. Only a deep understanding of the market or sophisticated statistical analyzers can help to avoid such problems.

3. Risk to lose money due to market conditions. During various economic events, as well as during volatile movements, trading conditions on trading instruments can significantly become worse. For example, after European Central Bank interest rate decision in September 2019, on the EUR/USD pair spread was increased in 4-6 times than usual.

Recommendation: Try to skip trading on the high volatility market and during different economical or fundamental events. More often, it is better to wait for a calm market and execute your transactions on it. Always check the size of the spread if you use MARKET orders, but rather switch to the LIMIT orders if possible.

4. Risk to lose money due to the wrong STOP LOSS. Also, it is important to pay attention to another feature STOP LOSS, which implements primitive protection functionality. Areas of stop loss of participants are often beaten out by the market maker, which leads to losses for traders.

Example of trader’s LONG positions liquidations by market maker on BTC/USDT, Binance exchange, January 2020.

Recommendation: The simplest solution is to use further STOP LOSS to avoid closing positions on volatile price movements, but sometimes this may lead to bigger losses. Ideal solution is to use adaptive algorithms that will adjust the size of the maximum loss depending on volatility and other market conditions.

How machine learning can help with this?

In other words, it’s not easy to consider all possible issues and to avoid them. It is necessary to keep in mind a large number of factors, to confirm hypotheses statistically, to validate them on historical data, that’s why automatic analysis methods and machine learning are well suited for portfolio creation task.

Machine learning approach can validate your manual trading decision, but better to use fully automated solutions for trading. For example, use machine learning to optimize automated strategies, to fit strategies to current market conditions, to dynamically change trading behavior on different volatility.

Without even analyzing the fundamental information, the algorithm will be able to recognize and quickly adapt to current market realities.

Here is an example of adapting a simple strategy with a machine learning approach for BTC/USDT pair. The strategy does not predict future price changes, but it adapts to current trends characteristics and current volatility.

About our company and our product

In fact, creating anti-crisis portfolios is just one of the tasks of trading and investing that our company is working on. Trading is quite complex and consists of a number of aspects. Our machine learning project is a solution we build during over 10 years working with global financial markets.

AlphaCube is a completely new technology for using machine learning in trading. Our technology allows to generate ten million of the trading strategies each minute, validate them, select best diversified combination and apply them immediately for a live trading.

What are the benefits of AlphaCube? 1. Protect from loses in a crisis period.

2. Automatic diversification by different trading instruments, markets and by using different trading strategies.

3. Flexible adaptation to the market situation, market conditions and volatility.

4. Real-time detection of price movements patterns and participants behavior.

5. Proven methodology on historical data, on walk forward validation and on live trading.

6. All strategies components build based on professional trader’s expertise. Discover more on our web-site: Bogdan Ivaniuk. CTO, London Analytics.


 London Analytics Ltd.


London, UK



+44 (0) 2071 014035

©2020 by London Analytics Ltd.