test-crypto-trade-bot

Crypto trading bot v2

Written on NodeJs.

Aim: try to find profitable algorithm to trade on cryptocurrency exchange.

Result: tests on real data shown that a lot of cases weren't taken into account. Bot is unprofitable and uncompleted.

Configuration

/config/.env

NODE_ENV=development

# Bitfinex
BITFINEX_API_KEY=
BITFINEX_API_SECRET=

# Bitmex
BITMEX_API_KEY_ID=
BITMEX_API_KEY_SECRET=

#Telegram (token to send messages using bot)
TELEGRAM_HTTP_API_TOKEN=

Exchanges details

(maker/taker)
1. Binance 0.1/0.1 LONG
2. Bitmex -0.0250/0.0750 LONG/SHORT
3. HitBTC 0 - 0.01/0.1 LONG
4. 1Broker - 0/0 but floating spread LONG/SHORT

Strategies list

1. (FAILED) Circle path
    Go throught n markets to get back to original state (currency e.g. USDT) 
    and get profit on order book disbalance.
    
    Pros:
        - Hight return in theory
    Cons:
        - Fee
        - Market always changes
    
    Failed because of exchange fees weren't taken into account when designing 
    algorithm. In v2 was introduced approach with using LINIT orders, but it 
    failed due to low market liquidity and slow order execution, so initial 
    profit conditions have changed already.
    
2. (PENDING) Volatility range
    Find range with flat volatility and buy on bottom/sell on top of it. 
    Same approach was used by me on WhaleCLub when hight volatility was detected, 
    so profit is made playing in range.
    
    Pros:
        - Checked personally with manual trading
        - Leads to medium risk/reward ratio
    Cons:
        - Tricky to determine a range
        - Range can be broken anytime
    
3. (PENDING) Indicators setup\
    Hire a few indicators and make decisions based on them. Here we have 
    some set of rules when satisfied allows us to open order in desired direction. 
    The main trick is to pick up set of indicators and calibrate theirs numereous 
    parameters. Every time set of indicators is changed a new strategie is created 
    in fact. So here we need something like 'Constructor' for that indicators. We 
    should have availability to add new module with logic for that set and just 
    plug in it easily.
    
    Pros:
        - Theoretically we can achive > 50% likelihood of profit trades
    Cons:
        - This approach is tight related to indicators setup. It can take 
        a long time until we found that 'correct' setup.
        - Setup can be broken anytime without any reasons