k-kit / liquitrader Goto Github PK

Schedule upkeep tasks

• check last time balances were updated, if greater than 1 min (time is subject to change, but sounds like a good interval) then fetch balances

apply default sorting and stuff from v1

1. initialize config
2. apply bull / bear override if enabled / conditions met
3. parse candle time-frames from conditions
4. generate condition objects from each strategy in config
5. Initialize exchange with quote currency, time frames, api etc..
6. exchange.update_balances()
7. fetch trades and calculate averages for non zero values
8. start exchange upkeep thread
9. pass pairs to run technical analysis on all pairs and store indicator statistics in pairs['symbol']['indicators'] as dict
10. pass pairs and config to master strategy analyzer
11. process all buy strategies for each pair and return a list of possible_buys
    * evaluate each buy strategy
    * assign evaluation results to conditions dict in pair
    * each buy strategy returns amount or none, if multiple amounts are returned, take the largest and add to possible buys
12. ^^ dca buys: append to dca_possible_buys
13. process all sell strategies append to possible_sells
14. DCA sells can just be appended to possible_sells
15. temp balance = exchange.balances[quote_currency] (use temp variable to decrease need for api calls to update balance) (stores quote currency balance):
16. for possible sells:
17. check orderbook depth
    1. try to find highest price can sell
18. make trades
19. add trades to 'profit tracker'
20. increment temp balance if trades are made
21. for possible buys:
22. check min buy balance, max pairs, 24h market change, quote_currency change (1h/24h)
    1. check order_book depth, only buy within max_spread % of price
    1. if checks pass return trade
    • update pair trade history
    • add trade to profit tracker
    • decrements temp balance
23. use temp balance in checks for dca_buys (only run on pairs where amount > dust value)
24. mostly the same as possible buys but with some dca specific checks
25. run exchange.update_balances()

• for coin balances where old coin_balance != coin_new balance
  • update pair[trades] exchange.client.fetchTrades(symbol)
  • calc averages

1. ...profit?

dict: pairs
dict: indicators
** dict: stored_averages {average price, trades, last_trade_id, quantity}
dict: strategy_evals
** = requires data persistence

need to decide where to store DCA levels, possibly with stored averages?

also where to store pair trade history (pair trade list not the same as profit tracker, used for average calcs)

add field for cross candles underneath "operator" in editor if operator= cross up or cross down

automate testing to make sure sockets are being kept open without downtime
this can maybe be done by starting sockets, then killing internet, sleep 2-5min then reconnect to internet and test all sockets successfully reconnected

Tests:

• various strategy combinations for each strategy type (buy/sell/dca)

  • ex. we have 3 buy strategies, which return the amount we want to buy, or none.
    1. evaluates to : 10
    2. evaluates to None
    3. evaluates to 100

• the analyzer should return 100 in this instance, test for all orders and combinations.

1. load user/pass/keys from text file into db
2. database stuff
3. find way to route to wizard on first run AND prevent navigating away from wizard
4. do not start the liquitrader or exchange thread until wizard is complete

• can work on this together if you'd like

• ability to pause all bot function on error and handle accordingly
• if rate limit error pause 60 sec first error, 5min subsequently
• auto restart after non-rate-limit critical errors

use dropdown instead of input

• this is mildly tricky because not all exchanges have the same timeframe, yet in the wizard we wont have access to which exchange is picked and what timeframes it has on the backend since it hasnt been set yet
  potential solution: ship gui with dict of exchange timeframes and pull exchange selected from wizard state

So this worked when I removed the wizard page (it looked right at least) but once I added the wizard code back it seems to be getting set to hidden even though visible is set to true

WIP

• Holding table:
  • symbol
  • ask price
  • bought_price
  • % change
  • volume
  • current value
  • dca level
  • trailing sell
  • dca trailing
  • 24h change
    • details for strategy evaluation
    • indicator values
    • and more will add later
• Market
  • symbol
  • ask price
  • volume
  • trailing_buy
  • 24h change
    • ... details
• buy history
  • tbd
• sell history
  • tbd

replace GAIN with
{A} ABOVE/BELOW {B} BY: {x}
x can be a number or a percent (will require changes in gui and condition_tools

binance only allows you to fetch 500 candles at a time so we will need to create a historical database of candles for all pairs for backtesting

• might only need for binance but tbd

1. if change is selected show "change over x"
2. if right is number or price dont show timeframe

• Pairs: DCA level(times DCA'd),
• Average bought price + last buy trade ID + amount
• Pair trade history
• Bot Trade history/profit tracker (not yet implemented) (buy/sell log)
• net balance at beginning of each day for days in buy/sell log

currently getting:

C:\Users\kylek\AppData\Local\Programs\Python\Python36\lib\asyncio\events.py:145: RuntimeWarning: coroutine 'GenericExchange._candle_upkeep' was never awaited
self._callback(*self._args)

• What conditions triggered + values for each trade

metrics

• day profit
• exponential
• growth
• daily profit
• aggregate pair profit data (numtrades, total gain, avg % gain etc.)

change this function to fit CCXT fetchOHLCV form in order to make the candle_tic_to_df function work for both this and generic

Keep in mind the version of Bearpuncher GCA will use will require an API to add users, and set the users 'balance' in relation to the net balance of the account, which will effect what that users profit-tracker displays for returns. We will also need the users join date, and any subsequent deposits / balance changes that will effect the data required for the user

note this will not be in retail bearpuncher just something to keep in mind during the design process

Currently if you edit a condition and save it it creates a new condition, either needs to update old condition or delete.

@luketimothyjones if you want to take some time and i'll go over the structure of the react code with you just let me know when