The task at hand involves a set of CSV files stored in the dataset folder. Each file has a specific structure, as shown below:

time,open,high,low,close,volume,rsi
1609464600000,0.0047319,0.00475,0.0047189,0.00475,8315920.0,63.4633309356425
1609465500000,0.0047467,0.0047639,0.0047199,0.0047365,25398760.0,59.419527863771584
1609466400000,0.0047312,0.0047383,0.0047204,0.0047263,2457619.0,56.49069147000932

The objective is to train a model that can recognize a specific pattern referred to as the "climbing pattern":

The model's goal is to identify the safest place to predict an upward trend. By recognizing more accurate patterns, the model should improve its ability to predict upward trends.

The dataset folder contains one file named DOGEUSDT.csv, which serves as a sample training dataset. If necessary, the model should be trained on the entire dataset within the folder. There is also a list of symbols that are currently commented out, potentially for future use.

The dataset includes the Relative Strength Index (RSI), which can be a helpful feature. Additionally, other technical indicators can be added if deemed necessary.

To generate more data, the create_dataset.py script can be executed.

The model should be saved to disk after training.

The prediction specifications are outlined in predict.py:

• Retrieve the latest data for a given symbol.
• Predict the next 10 candles (if possible) and detect if an upward trend is present.

def predict(dataset):
    # load the model from disk


    # calculate the result
    next10Candles = []
    trendIsUp = False
    return next10Candles, trendIsUp