Skip to content

Connector Architecture

Connector Architecture Diagram

The Architecture Diagram, given above, depicts the high-level design of a Connector.

Tip

Notice that for Derivative connectors, we have a multiple inheritance to ExchangeBase and PerpetualTrading.

Connector Component Overview

Connector Architecture Diagram

Each connector is comprised of the following components. Below are the detailed descriptions of tasks for each component and its corresponding files.

Exchange/Derivative.py

File: *_exchange/derivative.py — REQUIRED

Connector modules are centered around an Exchange/Derivative class, which are ultimately children of ConnectorBase. Each Exchange/Derivative class contains an OrderBookTracker and UserStreamTracker, and they are responsible for maintaining order books and user account information.

Exchange/Derivative instances also contain a ClientOrderTracker which tracks the connector's InFlightOrders, which are orders placed by Hummingbot currently on the order book. Typically, it is also helpful to have an exchange-specific Auth class, which generates the necessary authentication parameters/headers to access restricted REST endpoints and WebSocket channel, such as for placing orders and listening for order updates.

The Derivative class in particular inherits functions that are specifically used in perpetual markets. See the PerpetualTrading class for more info.

ConnectorAuth.py

File: *_auth.py — OPTIONAL

This class generates the appropriate authentication headers for the restricted REST endpoints to be used by the Exchange/Derivative and UserStreamTrackerDataSource classes. Generally, this would mean constructing the appropriate HTTP headers and authentication payload(as specified by the exchange's API documentation)

Some arguments tend to include:

  • HTTP Request Type
  • Endpoint URL
  • Mandatory parameters to pass on to the exchange (e.g. API key, passphrase, request body)

Depending on the specific exchange, different information may be needed for authentication. Typically, the Auth class will:

  • Generate a timestamp/nonce. Generate a signature based on the time, access method, endpoint, provided parameters, and the user's private key.
  • Compile the public key, timestamp, provided parameters, and signature into a dictionary to be passed via an http or ws request.

Note

This module is typically required for centralized exchange only. Generally, auth for DEXs is handled by the respective wallet.

OrderBookTracker

File: *_order_book_tracker.py — REQUIRED

Each Exchange/Derivative class contains an OrderBookTracker to maintain a real-time order book of one/multiple trading pairs and is responsible for applying the order book snapshots and diff messages to the corresponding OrderBook.

  • An OrderBookTracker contains a Dictionary of OrderBook for each trading pair it is maintaining.
  • APIOrderBookTrackerDataSource class contains either API requests or WebSocket feeds to pull order book data from the exchange.
  • The OrderBook class contains methods that convert raw order book snapshots/diff messages from exchanges into usable dictionaries of active bids and asks.

UserStreamTracker

File: *_user_stream_tracker.py — OPTIONAL

Each Exchange/Derivative class contains a UserStreamTracker, to maintain the current state of the user's account, orders and positions.

  • APIUserStreamTrackerDataSource class contains either API requests or WebSocket feeds to maintain user balance and order data from the exchange.
  • The Auth passed from the Exchange/Derivative class contains methods to construct the appropriate authentication requests for REST API calls or WebSocket channel subscription requests.

OrderBookTrackerDataSource

File: *_order_book_data_source.py — REQUIRED

The OrderBookTrackerDataSource class is responsible for order book data retrieval. It simply collects, parses, and queues the data stream to be processed by OrderBookTracker. Generally, this would mean pulling data from the exchange's API/WebSocket servers. For Perpetual connectors, the OrderBookTrackerDataSource is also tasked with maintaining the funding information of the active market.

It is necessary to track the timestamp/nonce of each message received from the exchange API servers to maintain a consistent and up-to-date order book. Depending on the exchange responses, we can keep an order book in the following ways:

  1. Presence of Timestamp/Nonce
  2. In this ideal scenario, we will only 'apply' delta messages onto the order book if and only if the timestamp/nonce of the message received is above or +1 of _last_diff_uid in the order book.
  3. Absence of Timestamp/Nonce
  4. In this scenario, we would have to assign a timestamp to every message received from the exchange and apply the delta messages sequentially only if it is received after the snapshot message and greater than the _last_diff_uid.

Note

It is important that the order book being maintained reflects all changes and is consistent with the order book on the exchange. As a safeguard/fallback, in the event when Hummingbot is unable to adequately maintain the order book, executing periodic order book snapshot requests can help to ensure that any deltas missed would be corrected.

UserStreamTrackerDataSource

File: *_user_stream_data_source.py — OPTIONAL

The UserStreamTrackerDataSource class deals with user data retrieval. It simply collects, parses and queues the data stream to be processed by UserStreamTracker.

Unlike OrderBookTrackerDataSource, UserStreamTrackerDataSource only retrieves data about user account balances and orders.

InFlightOrder

File: /hummingbot/core/data_type/in_flight_order.py

Stores all details pertaining to the current state of an order.

It is important to keep a consistent and accurate state of all active orders placed by the user. This ensures that the strategies are given the correct information and are able to perform their tasks accordingly.

ClientOrderTracker

File: /hummingbot/connector/client_order_tracker.py

An instance of ClientOrderTracker holds and manages InFlightOrders by calling the connector's trigger_event method.

Provides utilities for connectors to update in-flight orders and to handle order errors.

For more details on how to begin implementing the components, please refer to the Connector Tutorial

Protocol Connector Components Overview [TBD]

Coming soon.

Fee Accounting

The BudgetChecker uses the information from a TradeFeeSchema to generate a specific instance of TradeFeeBase that is then applied to an OrderCandidate in order to asses the order's effects on account balances.

TradeFee

The TradeFee object contains the necessary information to account for fees when estimating an order's impact on account balances.

TradeFeeSchema

Contains the necessary information to build the TradeFee object. For both makers and takers specifies percent and fixed fees, and tokens in which the fees are paid. Exchanges must specify their respective default schemas inside their [exchange]_utils.py files:

DEFAULT_FEES = TradeFeeSchema(
    maker_percent_fee_decimal=Decimal("0.001"),
    taker_percent_fee_decimal=Decimal("0.001")
)

  • percent_fee_token: str
  • maker_percent_fee_decimal: Decimal
  • taker_percent_fee_decimal: Decimal
  • buy_percent_fee_deducted_from_returns: bool
  • maker_fixed_fees: List
  • taker_fixed_fees: List

TradeFeeBase

A specific instance of the TradeFeeBase class defines the fees to be applied to an order - their types, amounts and assets.

  • fee_amount_in_quote(): calculates a total fee in quote asset units as a combination of a percentage fee and fixed fees
  • get_fee_impact_on_order_cost(): returns order cost for a particular position opening OrderCandidate with fees accounted for
  • get_fee_impact_on_order_returns(): returns order returns for a particular position closing OrderCandidate with fees accounted for

AddedToCostTradeFee

Extends TradeFeeBase, implements get_fee_impact_on_order_cost(), get_fee_impact_on_order_returns(). Fees of this class are applied on top of the cost of a buy order (e.g. a buy order of 10 COINX at 9 USDT with a fee of 1% means that the user's account will be deducted 90.9 USDT and added 10 COINX — this is most exchanges' approach to fees).

DeductedFromReturnsTradeFee

Extends TradeFeeBase, implements get_fee_impact_on_order_cost(), get_fee_impact_on_order_returns(). Fees of this class are deducted from the returns of a buy order (e.g. a buy order of 10 COINX at 9 USDT with a fee of 1% means that the user's account will be deducted 90 USDT and added 9.9 COINX — this is Binance's approach to fees).

Example - TradeFeeSchema

trade_fee_schema = TradeFeeSchema(
    maker_percent_fee_decimal=Decimal("1.0"),
    taker_percent_fee_decimal=Decimal("2.3")
)

Example - TradeFee

from hummingbot.client.settings import AllConnectorSettings

trade_fee_schema = AllConnectorSettings.get_connector_settings()[exchange].trade_fee_schema

percent = trade_fee_schema.maker_percent_fee_decimal if is_maker else trade_fee_schema.taker_percent_fee_decimal
fixed_fees = trade_fee_schema.maker_fixed_fees if is_maker else trade_fee_schema.taker_fixed_fees

trade_fee = AddedToCostTradeFee(percent, trade_fee_schema.percent_fee_token, fixed_fees)

Last update: 2022-03-10