Yes, but it depends on your label frequency. If daily frequency, then you have to downsample the intraday features - our Feature Engineering Toolbox can do that. If intraday frequency, then you just need to fill forward your daily features.

Yes, if you normalize the features and labels appropriately, then a single model can be built to predict all stocks. 

Yes. Mark-to-market returns can be used instead of realized Round Trip returns.

It is not mandatory, but any strings/non-integer values will be used for one-hot encoding. We recommend, at least for the prototype version, to use a date column for simplicity.

In a metalabel application, if a trade was not made for a certain sample, the target should be "NaN" or "Null" instead of "0".

We have a direct integration with QuantConnect, which you can you use to backtest your trading strategies.

We offer 616 global macroeconomic daily features with our Premium subscription. If you have any interest in testing these features for free, email info@predictnow.ai

Email info@predictnow.ai to get a free trial of the platform's API version.

Users can anonymize all their column titles before uploading their datasets to the PredictNow.ai platform. Since each user is attempting to predict the outcome of *their* unique strategy, there is very minimal arbitrage between the predictive accuracy and our 616 features. We also sign mutual NDAs with all of our Premium customers.

In the Dashboard click on “Account”. In Account, click the “Manage Subscriptions” button followed by the “Suspend subscription” button.