The goal of this project is to play with crypto currencies and make profit from triangular arbitrage. We want to create a Python arbitrage bot.
This script is listening to crypto market and is waiting for arbitrage opportunity. If it founds one, it automatically does the transaction. It also sends a message to a Telegram chat when it completes a transaction.
Triangular arbitrage is a way of making money on a market without predicting the future.
To understand what's traditional arbitrage is, imagine Alice selling Bitcoin 1000 dollars while Bob is buying Bitcoin 1100 dollars.
If you buy a Bitcoin to Alice and you sell it instantly to Bob, you have an instant profit of 100 dollars. You can use this method to make profit from different crypto exchanges, but it has severe constraints: every exchanges need to be running at same time, you need to pay fees on each transaction, you need to sync your wallets...
An other method to do arbitrage is called "Triangular arbitrage". This method can be used on a single platform.
To understang triangular arbitrage, imagine:
- Alice is selling a Bitcoin 1000 dollars
- Bob is exchanging 1 Bitcoin for 100 Litecoin
- Carlos is exchanging 1 Litecoin for 11 dollars.
We can buy 1 BTC for 1000 USD, then exchange our 1 BTC for 100 LTC, then our 100 LTC for 1100 USD. We won 100 USD during the transaction.
This bot uses ETH as base wallet, and BTC as exchange currency. So we have two types of arbitrage:
Forward: ETH -> ALT -> BTC -> ETH
and
Backward: ETH -> BTC -> ALT -> ETH
- ccxt
- python-telegram-bot
- matplotlib if you want to use graph_balance.py
- Binance
- Bittrex
- Bitfinex
The first step is to check if the trade would be profitable or not, using the estimate_arbitrage_forward/backward functions. We estimate that it's proftiable if this functions returns a probability greater than our THRESHOLD.
Once we have found our opportunity, we will try to get the wanted asset at the best price possible:
- We try buying/selling at the best price in orderbook by creating a limit order.
- While the order is not completed after WAIT_LIMIT_ORDER, we decrease the price in the orderbook.
- Once we have completed the order, we pass to the next step.
So, for example, we want to buy ADA/ETH. Here is the bid orderbook:
| Order index | Price |
|---|---|
| 1 | 0.00032668 |
| 2 | 0.00032669 |
| 3 | 0.00032675 |
Here we try to buy 1 @0.00032668, we wait WAIT_LIMIT_ORDER seconds to see if the order has been fulfilled, if not we try to buy 2 @0.00032669, and so on until we buy the asset.
Once we completed the first step of our arbitrage, we go onto the next one until the arbitrage has been completed.
First, install all dependencies.
Create a Telegram bot using this tutorial.
Create API secrets for all exchanges.
Create a secrets.py file with your values:
BINANCE_KEY="XXX"
BINANCE_SECRET="XXX"
BITTREX_KEY="XXX"
BITTREX_SECRET="XXX"
BITFINEX_KEY="XXX"
BITFINEX_SECRET="XXX"
TELEGRAM="XXX:XXX"
TELEGRAM_CHAT="XXX"Then run run.py file with Python 3.
You can use the Crypto class to do your own bot or operations:
import crypto from Crypto
import currencies
# Create instance, connects to your exchange accounts
crypto = Crypto()
# Logs in logs.txt
crypto.log("Hello world")
# Logs in logs.txt and as a Telegram notification
crypto.log("Hello world", mode="notification")
# Get your ETH balance on Binance
crypto.get_balance(crypto.binance, "ETH")
# Get the ETH/BTC bid price on Bittrex
crypto.get_price(crypto.bittrex, "ETH", "BTC", mode="bid")
# Get the LTC/BTC price on Bittrex
crypto.get_price(crypto.bittrex, "LTC", "BTC")
# Get the BTC/USDT ask price on Bittrex
crypto.get_price(crypto.bittrex, "ETH", "USDT", mode="ask")
# Estimate triangular arbitrage profit (in %) for given asset and exchange
crypto.estimate_arbitrage_forward(crypto.bittrex, "ZIL")
crypto.estimate_arbitrage_backward(crypto.bittrex, "ZIL")
# Create market buy/sell orders
# Buy 0.3 ETH with BTC at market price
crypto.buy(crypto.binance, "ETH", "BTC", amount=0.3)
# Buy ZIL with 30% of available ETH at market price
crypto.buy(crypto.binance, "ZIL", "ETH", amount_percentage=0.3)
# SELL 10 ZIL for BTC at market price
crypto.sell(crypto.binance, "ZIL", "BTC", amount=10)
# SELL 100% of available ZIL for BTC at market price
crypto.sell(crypto.binance, "ZIL", "BTC", amount_percentage=1)
# Create limit buy/sell orders
# Will buy ETH with BTC at 0.053 or better
crypto.buy(crypto.binance, "ETH", "BTC", amount=0.3, limit=0.053)
# Will sell ETH with BTC at 0.054 or better, if the order is not fullfilled after 3 seconds, we close the order
crypto.sell(crypto.binance, "ETH", "BTC", amount=0.3, limit=0.054, timeout=3)
# Buy with the best possible price
# Will try better prices in order books to buy ETH with 50% of BTC available
crypto.best_buy(crypto.bittrex, 'ETH', 'BTC', 0.5)
# Will try better prices in order books to sell 50% of ETH
crypto.best_sell(crypto.bittrex, 'ETH', 'BTC', 0.5)
# Get order book
crypto.get_order_book(crypto.bitfinex, 'ETH', 'BTC', mode='asks')
# Get compatible alt coins
print(currencies.binance_alternatives)
print(currencies.bittrex_alternatives)
# Get fees for given exchange and market sens. It returns the amount of value remaining after the trade
# If the fee is 1%, it will returns 0.99
print(crypto.get_fees(crypto.bitfinex, 'buy'))
# The crypto class can track balance for multiple exchanges in a local CSV file
# Get the last recorded balance
print(crypto.get_last_balance())
# Record a new loss of 0.2 ETH
print(crypto.save_gain(-0.2))API calls to get_price and get_order_book are cached, if you want to clear cache and fetch updated data:
# Clear cache
crypto.flush_cache()get_price method is great to fetch approximated price, if you want to be more precise over your orders I would suggest you to get the order book and then create a limit order to buy your asset at a precise price. Market orders are fast and simple but can be executed at unexpected price. I stromgly advise you to use limit orders if you want to be successful with arbitrage.
# Wait for opportunities and execute arbitrage if found
python3 run.py binance
# Generate a graph of the evolution of the balance
python3 graph_balance.pyIf you need help setting up the bot on your side, or if you have any suggestions, DM me on Twitter @AirM4rx.
You can also make pull requests.
If you want to thank me: BTC: 3Dh7gEbDKbR6n3VAB1eK16eQgRGBJGU6vD