AI | Vega Insight

Find and capture trading opportunities

At the TTF and NBP our AI can profitably predict changes in the forward curves before they happen. We provide you with a suite of predictions, a testing universe and robust analysis.

We predict the future. You capture it.

Constant Learning

Commodity markets are volatile and influenced by data, with a high and variable update frequency. Vega Insight is structured to exploit a rapid "test-and-learn" cycle, which underpins our vision to introduce and develop “The Science of Trading” and our core value of “Constant Learning”.

Our products and services augment human trading decisions with Machine Learning (ML) capabilities. This allows our customers to systematically find and capture value from perishable opportunities.

Data is the starting point. We've cleaned, structured and merged historical exchange and broker data to create a history of the market timestamped to the millisecond. Against this we have mapped fundamental data to explore relationships.

Our virtual trading system allows complex strategies to be developed and tested. This is fed by our suite of indicators that give buy or sell signals that can be measured for value creation.

For one of our customers the identification of a 0.5% improvement in the price equated to a gain of USD 50 million per year.

Find Baseline

We use an “academic” approach to identify the benchmark assuming away execution practicalities and focus on change in value. Therefore, a reduction in loss from a changed input merits just as much examination as profit.

Test and Refine

There are two levels of testing and refinement in our Lab. Firstly, to deploy the best signals for your strategy. Secondly, we add “slippage” and real-world operational issues such as mandate constraints and execution costs.

Extract value

From your research and strategy testing you can deploy the live predictions either as a feed to an algorithm or as an input to your bullish/bearish sentiment, so that you can stay “one step ahead” of the market.

Solutions

We zoom in on relevant analysis underpinning our AI predictions and their use in trading strategies. Our insight is delivered in digestible reports grouped by trader type, strategies, approaches to fundamental data and market conditions.

While we also cover academic approaches and findings our focus is on creating a baseline for profitable strategies that takes into account market “slippage”. The trading strategies of interest are viewable in the Vega Lab, allowing you to explore results in detail or expand in new directions yourself.

Through these reports, expanding vocabulary and concepts we create a “knowledge tree” that systematically grows as the analysis expands as we identify what works and why. And equally as important, what might not work and why.

Available from 21.01.21

In the Predict area we deploy our AI developed algorithms in a live environment fed by data as it comes to market. The algorithms only require a few seconds to process and provide market predictions at different granularities.

You can download the signals directly to your system for algorithms or view them on our platform for complementary insights to inform your bullish/bearish sentiment as you approach support/resistance points in the market.

Available from 22.02.21

In the Lab, you can select different types of predictions from Vega Predict, fundamental data or your own datasets. This allows you to precisely examine a strategy or initiate a 'step-run' to explore strategy permutations along a particular dimension. Within the Lab, we explore two different types of trading:

“Base trading”, which is systematic sales by producer or systematic purchases by consumer. We assume the trader is either 100 000 therms per day long or short. The most profitable combination gives an annual value add of GBP 1 million.

"Proprietary trading”, which speculates on price developments for profit without a base position. This is more challenging as you must enter, hold and exit, which introduce more slippage.

Available now