Username: Password:

Author Topic: Feature hypothesis  (Read 10365 times)

Offline korsehok

  • Rookie
  • *
  • Posts: 2
    • View Profile
Feature hypothesis
« on: June 11, 2017, 03:52:46 PM »
DISCLAIMER: I am borderline retarded when it comes to developing any software so these suggestions can be taken with that kept in mind.

Now with that said, if these are any inspiration for better ideas or something to develop towards I would be happy to have given them.

Config classes for pairs: CLASS A, B, C, D etc. You could assign class category adjustments in allpairs-params and in pair-specific config files you could tag the pair as a class. This would make it easy to adjust multiple pair settings simultaneously. Could make functions to change between different sets of config automatically based on certain criteria.


The Dumpwatch function: Buylevel adjusted relative to short term positive value swing. If a coin has pumped a lot, for example 150% in the past 48 hours (I don’t know what would be good metrics?) Have an added strategy option to automatically change config class to another that is more careful and does not buy so quickly when the price starts to fall. The price can continue to go up, but the more it goes up the harder it is going to come down(when it comes down) and thus has relatively bigger chance to dump lower for the increased volatility.

The Correction Index: The bot could listen on to BTC price change and other technical stats, BTC lending rates, total exchange volume, average total price changes in the markets in the last X-amount of time specific to a coin or exchange, USDT:USD ratio, other factors to make educated guesses on future market reactions. Could make a “Market correction likelihood index” and give different factors unique weightings. When index grows over certain limit, change all pair config class to another that is more careful.

I feel these functions would decrease the amount of negative trades and the amount of time your coins are locked "in bags" if they are possible to create.