HI Team,
I have just completed both Decision Trees and Neural Networks Course. It gave very good insight into RNN and LSTM. It helped us to identify the next day trading signals or close price value based on the last few months of data with the day interval.
But, i would expect to build a model with the minute interval and then use that model during live trading to generate buy/sell signals. I hope this can be done using RNN / LSTM but not sure how to do it.
I expected this to be covered in this course because this is the very good usecase for Neural Network.
Can somebody help with some code snippets or any blog link which can help acheiving it?
Hi Arun,
You can use the OHLC information of the 1- minute data to train the DNN/LSTM model that we gave in the course. But there are a couple of things that you need to remember when trading on a smaller time frame:
- Close to Close returns may not be possible as taking a position just before the one-minute candle closes is not easy to execute.
- Have to take into account the slippage that will happen when executing trades at the close of the candle
- The transactions costs which were negligible compared to the daily price movements will have a significant impact when trading in a minute frequency. Hence, they need to be incorporated in the backtest.
- Best scenario would be to optimize your entry and exit along with the trend prediction as this will minimize the round turns and the transaction costs.