Exchange Interface

Theory

The interface class provides a consistent way of interacting with the supported exchanges.

The idea is that as long as scripts aren't using "exchange specific" information, if all calls are run through here, then it gives the developer the ability to instantly switch between exchanges. The backtesting framework also requires that the calls are made through the interface.

This means a few things about the information through the Interface:

• Pre-typed. Information guaranteed by the interface comes pre-casted. Most information sent from API's comes as strings, but the interface class converts these for easy access.
• Less Detailed. Different exchanges give different data. We try to pick out the most important bits of information to conecrate on maintaining, but some of it will end up in the exchange_specific tag on the dictionary returned.
• Multiple API Calls. Sometimes to accumulate the data required to tell the user everything they need to know, it requires multiple API calls to be run in the background.
• Supports Paper Trading & Backtesting. To use paper trading & backtesting features all calls must be run through the interface.

The interface can grow and change based on user feedback - if you see an important feature we're missing, tell us in the GitHub issues page.

Creation

Interfaces are pre-generated when an exchange object is created. They can then be accessed by using .get_interface() on an exchange:

import blankly

exchange = blankly.CoinbasePro()  # This can be .Binance() or .Alpaca() or anything else supported

interface = exchange.get_interface()  # Use the getter

print(interface.get_account())

Rest API Functions

get_calls()

Get the direct RESTful exchange object that the interface is using.

Where is my IDE linting (autocomplete)?

If you want easy linting, its better to use an exchange object and use .get_direct_calls()

You can also create linting by assigning it one of these types depending on which is accurate context:

• from blankly.exchanges.interfaces.coinbase_pro.coinbase_pro_api import API
• from blankly.exchanges.interfaces.binance.binance_api import API
• from blankly.exchanges.interfaces.alpaca.alpaca_api import API

get_exchange_type()

Determine which exchange an interface is running on.

Response

get_products()

Get all trading pairs currently on the exchange.

Response

[
  {
    "symbol": "BTC-USD",
    "base_asset": "BTC",
    "quote_asset": "USD",
    "base_min_size": 0.0001,
    "base_max_size": 280,
    "base_increment": .01
  },
    ...
]

get_account(symbol=None)

Gets the current account holdings if symbol is None, otherwise will return the symbol holdings. If passed in a specific symbol, it will attempt to get the base asset ('BTC-USD' as symbol will give you 'BTC').

Arguments

Response

If symbol='BTC':

{
  "available": 2.3,
  "hold": 0.2
}

‌If symbol=None:

"BTC": {
    "available": 2.3,
  "hold": 0.2
},
"USD": {
  "available": 4352,
  "hold": 23
}

market_order(symbol, side, size) -> MarketOrder

Create a new live market order.

Arguments

Response

A market_order object. Documentation is pending.

limit_order(symbol, side, price, size) -> LimitOrder

Create a new live limit order on your exchange.

Arguments

Response

A limit_order object. Documentation is pending.

take_profit_order(symbol, price, size) -> TakeProfitOrder

Create a new live take-profit order on your exchange. This is a market sell order that will execute when the price reaches a certain high.

Arguments

Response

A TakeProfitOrder object. Documentation is pending.

stop_loss_order(symbol, price, size) -> StopLossOrder

Create a new live stop-loss order on your exchange. This is a market sell order that will execute when the price reaches a certain low.

Arguments

Response

A StopLossOrder object. Documentation is pending.

cancel_order(symbol, order_id) -> dict

Cancel a particular order.

Arguments

Response

{
    "order_id": "b6d2f951-dae0-89e8-3e79-b460b1e9eead"
}

get_open_orders(symbol=None)

Get a full list of open orders.

Arguments

Response

[
    {
        'id': 'dfa936a4-ea8b-4dbf-bb99-b2b632a5370a', 
        'price': 10000.0, 
        'size': 1.0, 
        'symbol': 'BTC-USD', 
        'side': 'buy', 
        'type': 'limit', 
        'status': 'open'
    },
    ...
]

get_order(symbol, order_id) -> dict

Get info about a particular order. If the objects returned by placing orders are saved, this function shouldn't need to be used.

Arguments

Response

{
  'id': 'dfa936a4-ea8b-4dbf-bb99-b2b632a5370a', 
  'price': 10000.0, 
  'size': 1.0, 
  'symbol': 'BTC-USD', 
  'side': 'buy', 
  'type': 'limit', 
  'status': 'open'
}

get_fees() -> dict

Get the maker and taker fee rates of a particular exchange.

Response

{
  "maker_fee_rate": 0.0050,
  "taker_fee_rate": 0.0050
}

history(symbol, to = 200, resolution = '1d', start_date = None, end_date = None, return_as = 'df') -> pandas.DataFrame

Download historical data with rows of at least time (epoch seconds), low', high, open, close, volume as columns. This is a wrapper for get_product_history(), and exists to allow users to more easily download OHCLV data

Arguments

Response

Pandas dataframe with at least these columns.

Or if return_as is set to list:

{
    'time': [1591110000, 1591113600],
  'low': [9270.0, 9417.38],
  'open': [9583.36, 9464.44],
  'close': [9464.46, 9478.95],
  'volume': [5979.77327365, 1185.12835638]
}

Example Use Case

a = Alpaca()
interface = a.Interface
interface.history('MSFT', to=300, resolution='15m', end_date='2020-05-03') # get 300 points from May 3rd, 2020 as a dataframe
interface.history('MSFT', to=300, resolution='15m', end_date='2020-05-03', return_as='list')['close'] # get 300 points from May 3rd, 2020 as a dictionary of np.arrays

get_product_history(symbol, epoch_start, epoch_stop, resolution) -> pandas.DataFrame

Download historical data with rows of at least time (epoch seconds), low', high, open, close, volume as columns.

Arguments

Response

Pandas dataframe with at least these columns.

get_order_filter(symbol)

Find the limits that the exchange puts on purchases for a specific asset.

Arguments

Response

{
  "symbol": "BTC-USD",
  "base_asset": "BTC",
  "quote_asset": "USD",
  "max_orders": 1000000000000,
  "limit_order": {
    "base_min_size": 0.001,
    "base_max_size": 10000.0,
    "base_increment": 1e-08,
    "price_increment": 0.01,
    "min_price": 0.01,
    "max_price": 9999999999
  },
  "market_order": {
    "fractionable": True,
    
    "base_min_size": .001,
    "base_max_size": 20,
    "base_increment": .001,
                
    "quote_increment": 0.01,
    "buy": {
      "min_funds": 10.0,
      "max_funds": 1000000.0
    },
    "sell": {
      "min_funds": 10.0,
      "max_funds": 1000000.0
    }
  },
}

get_price(symbol) -> float

Get the quoted price of the trading pair.

Arguments

Response

• Returns a float which is the price of the trading pair, such as 53000 or 35000.

Rest API Properties

These properties can be used as shortcuts for making API calls. These are built-in abstractions of the previous methods that provide functionality that is common when developing bots & models.

Each of these performs API requests

.account -> dict

The equivalent of performing .get_account() with no arguments,

Response

"BTC": {
    "available": 2.3,
  "hold": 0.2
},
"USD": {
  "available": 4352,
  "hold": 23
}

.orders -> list

The equivalent of performing a get_open_orders() request with no arguments.

Response

[
    {
        'id': 'dfa936a4-ea8b-4dbf-bb99-b2b632a5370a', 
        'price': 10000.0, 
        'size': 1.0, 
        'symbol': 'BTC-USD', 
        'side': 'buy', 
        'type': 'limit', 
        'status': 'open'
    },
    ...
]

.cash -> float

Get the amount of cash (generally the quote currency) inside the portfolio. This default quote can be set in the settings.json file for each exchange independently. This is used as a shortcut for easily determining buying power when making a transaction. This property is configurable because users may be using a variety of cash quotes such as USD, USDT, USDC or EUR. See here. Usage of this is discouraged if trading more complex assets such as BTC-ETH.

Response

A single float which represents the size value of the account defined in settings.json.

5931.533