History (convention and behaviour)

[rengel]

Hello there,

I wonder about these things. Maybe it is technically not feasable at this point .. I don't know.

1. History strorage convention
Right now I have separate *db-files (i.e. poloniex_0.1.db). The importer and the mimic to join segments is working for me, but sometimes very slow and unstable. I saw a thread offering history data, that will speed things up and not everyone would need to do the same time consuming thing.

Wouldn't it be nice, if the naming convention would identify a db file by trading plattform (exchange), currency and asset?
poloniex_USDT-BTC.db for example. This is only a question up to discussion.


2. Something like "Hot-Start"
History data is mandatory for technical indicators, no way around this.
But, is it already possible, to start (Paper- or Live-Trading) with any given size of history, which is not aggregated continuesly from reading new coming data, but by retrieving all the needed history size from the past before the start of the strategy? So basically an importing - process.
When all missing candles were collected/imported to join a complete history of the demanded size to the actual point in time, the strategy would then be allowed to make trades. 

Regards,
Rainer

[susitronix]

hoi Rainer

the history is in one packet e.g. binance
but you can use Sqlite software to sort things out or merge together


i made a tutorial about Sqlite:
TUT

hope this helps

[susitronix]

for life gekko you can also load the necessary data first manually and then start the live gekko.
Gekko will looking for data in the /history folder first, before trying download the data (slow slug)

on binance with the --ui gekko, it does not work, but with CLI gekko it works..!

its not gekkos fault but binance´s i guess. lol

[rengel]

Hello susitronix,
and thank you for your replies. The database thing is fine, I guess.

As of the behaviour of creating the history for paper/live-trading, I only noticed, that the history defined by youreself is only downloaded live/fresh, after bot-activation.

Behaviour I know:
10min Candle Size + History size of 36 means the bot collects candle data for 6 hours, before the first trade signals can be executed.

Behaviour I though of:
10min Candle Size + History size of 36. New candle data is downloaded. The importer searches for the demanded history size among the local data I already have. If not available, these candles are downloaded and, as fast as this data can seamless be joined with the new candles retrieved after bot activation, trade signals can be executed.
So by this one could save vast delays caused by waiting time during history collecting period before first trade. According to your statement, this could be working in CLI mode. I will have a look at that then, some time soon.

Regards,
Rainer