How Big Data Can Enhance Commodity Trading Strategies in 2020

Volatile Commodity Markets

For most commodities 2019 saw larger-than-expected price gains : crude oil rose by over 15%, gold by a similar amount, silver by over 12%, copper by around 5% and iron ore by over 20% - whilst natural gas, the oddball of the bunch, declined by over 25%.

Rising geopolitical concerns, trade wars, and climate change were all key factors driving the increase in commodity volatility, and they are unlikely to go away overnight leading many strategists to forecast continued gains - or at the least continued volatility - in the year ahead.

Natural gas has fallen by over 25% in 2019 due to global warming even though its energy cousin crude has rallied amidst increased tensions in the middle east and most recently the assassination of Iranian General Qassem Soleimani.

Sentiment Data for Commodity Trading Strategies

For traders, volatility means opportunity - assuming, that is, they can beat the market - a big ask in these days of split-second global communication networks and high-frequency trading. One piece of new technology which can help them develop commodity trading strategies is internet big data, more specifically, sentiment derived from news and social media.

Sentiment data is generated by sophisticated news-reading robots that can scan the web quicker than the blink of an eye, and then generate sentiment statistics which traders can leverage to benefit their decision-making.

Due to recent developments in natural language processing, it has now also become possible to distinguish between whether the subject of an article is being discussed in a positive or negative light.

The research suggests that trading signals based on news sentiment can help predict commodity market direction, especially over the next 24 hours.

The usual method for monetizing this data is to gauge the news sentiment around specific events known to impact on an asset’s price and then use the results as the basis for generating trading signals whether as part of a discretionary process or systematic algorithm.

In the case of oil, for example, the news event might be the publication of the latest inventory report, such as the weekly Energy Information Administration's (EIA) Crude Oil Inventories, the news from the latest OPEC meeting, or GDP data, because of the proven link between economic growth and oil demand.

In a 2018 study conducted by data scientists at RavenPack, a key finding was that commodity trading strategies were twice as successful during periods of high volatility , making them ideal for current market conditions.

“By incorporating regimes which only trade during high-volatility regimes, we can improve on the out-of-sample return irrespective of whether we look at annualized, risk-adjusted, or per trade return,” says Peter Hafez, Chief Data Scientist at RavenPack. “IR climbs from 0.65 to 1.27 and annualized returns from 9.8% to 21.3% for the high-volatility strategy, despite the reduction in the number of trades.”

The data also has a wider application to other commodities. In a separate study in 2017, the RavenPack team used news sentiment to predict price moves for a basket of precious metals using a similar method.

Here too it was found that by focusing on the news around specific events of heightened significance to each commodity they could predict market direction with a reasonable probability of success.

When used as the basis of a trading strategy, the signals generated a 35.7% annualized return, using an equal-weighted ensemble of 10 different machine learning algorithms.

Commodities Event and Sentiment Data

Easily implement the above research or similar using the RavenPack Analytics Platform, which includes sentiment data on over 250,000 individual entities and events, available on visual dashboards or via web APIs. Request a trial today.