FAQ

Frequently Asked Questions

Do you use your own predictions for trading?

No, we do not engage in trading as a business. While we do enjoy occasionally trading on an individual basis, we do not possess the expertise or resources to operate as professional traders (PTs).

PTs employ specialized infrastructure, proprietary trading strategies, and market intelligence, along with skilled individuals, to generate consistent returns. PTs also enjoy low fees that are not available to the public and negotiated with the exchanges on a bilateral basis.

Our role is to provide an additional layer of valuable information that can enhance PTs' overall performance and mitigate their risk exposure.

How do you test the quality of your predictions?

We use the latest techniques in machine learning (ML) to train our models on very large sets of data. The quality of model predictions are validated with data that has not been used in training the model. Further more, we perform forward testing (testing with the most current data that has not been used in training or validation phase) for at least two weeks before a model is published or updated.

The quality of predictions are measured with standard metrics such as accuracy, precision, recall, F1 score and a number proprietry domain-specific measures.

How can we verify the published quality metrics?

We believe in transparency. We publish historical and real-time quality metrics for each model. You can independently verify a subset of these figures by collecting the real-time perdictions for a period of time and comparing your results with our quality metrics.

How do I know that your predictions will enhance my trading strategy?

We provide standardized quality metrics along with the prediction quality matrix (PQM) for each of our predictions. The PQM allows you to define your own relevant quality measures based on your unique trading strategies. These metrics offer an initial assessment of the predictions' suitability for your specific scenario.

Additionally, we publish the returns and Sharpe ratio of simulated trading strategies under ideal conditions. This information gives you an indication of the potential performance of the predictions.

However, the most effective way to evaluate the predictions is to integrate them with your own trading strategy in practical use. We suggest starting with forward testing without actual trading, measuring hypothetical returns. Once satisfied with the results, progress to low-volume trading. Finally, if the outcomes meet your expectations, you can confidently engage in full-scale trading.