Data Maps

data_map

The data_map is the name we give to a specific data structure that will be interacted with when retrieving requested datasource_topics and candle_topics data. It resides within the Strategy class that is being inherited and can be accessed by using the super() keyword. This data_map holds all the data that is requested via the datasource_topics and candle_topics. It is represented with the following type Dict[str, List[Dict[str, str]]].

Note that the returned values are typed as str. In order to use the values for calculation, please refer to documentation and coerce the str type to the appropriate types. Example:

self.data_map["candles-1m-BTC/USDT-bybit"][-1]["start_time"] # returns a string
float(self.data_map["candles-1m-BTC/USDT-bybit"][-1]["close"]) # returns a float, usable in calculations

This structure is given to the user thorugh self.data_map and/or super().data_map

split_data_map

The split_data_map, similar to the data_map is a specific data structure that can be interacted with to retrieve requested datasource_topics and candle_topics. Similar to the DataMap it resides within the Strategy class and is accessed in the same manner with self.split_data_map and/or super().split_data_map

It is represented with the following type Dict[str, Dict[str, collections.deque]].

The defining difference between the split_data_map and the data_map is that the data is stored by the topic's keys. For an example, a candles-1m-BTC/USDT-bybit topic would be stored by the 'open', 'high', 'low', 'close' values. This can be represented similar to the example below.

Example:

latest_close = self.split_data_map["candles-1m-BTC/USDT-bybit"]["close"][-1];
 
close_arr = np.array(self.split_data_map["candles-1m-BTC/USDT-bybit"]["close"])

With the new accessing method, we are able to create a np_array directly without having to use map() and also retrieve the latest data with much more clarity.

When to use which?

When writing your live strategy and/or backtests that are expected to be turned into live running strategies. The use of split_data_map is suggested. The more performant access as well as direct conversion to np.array will be more beneficial in such a case.
On the other hand, if your goal is to retrieve datasource and have the data be stored in a tangible JSON state or even converting ot CSV with ease, the use of data_map is suggested.