hummingbot-strategy-dsl

Domain-Specific Language (DSL) for creating hummingbot AMM strategies, aka hbot strategies.

This package contains the metamodels and the grammar of the language, and a code generator for automated generation of software components required for the creation and deployment of new strategies.

Furthermore, a CLI interface is provided so as to validate and generate hbot strategy projects from input model files.

Minimize Required Effort to Start Developing Strategies

As mentioned in hummingbot#4715, currently to create a new strategy, a user needs to understand various inner workings of Hummingbot and to maintain a minimum of 5 files

[strategy]_config_map.py
Details configurations needed for a strategy.

conf_[strategy]_TEMPLATE.yml
Formats a configuration file with comments and version number.

conf_[strategy].yml
A configuration file, created automatically by the 2 files above. A user can change the setting either by using config command or by updating the file manually

start.py
Initialises the strategy, connectors and market_info objects.

[strategy].py
The main strategy file which users need to care about the most.

By digging into these files, we noticed that the same information is used, which is essentially the core attributes of a Strategy:

• Name
• Description
• Markets
• Parameters

By providing this information, the aforementioned files can be auto-generated from input models describing the strategies. In this context, the hbot-strategy-dsl provides means of auto-generating strategies from input models.

The set of these files define an Hbot Strategy Project!

Installation

Install as a simple python package.

pip install .

Or install in development mode:

python setup.py develop

The installation includes an executable cli which can be used for the validation and generation of strategy model files.

User Guide

Model Creation

The grammar of the DSL is developed using the textX framework.

The grammar definition is located in the grammar/strategy.tx directory of this package.

Models are defined within a simple file with the .strategy extension. The grammar is json-like, with extra buildins for defining several concepts.

Below is an example of a strategy which connects to 2 markets and has 2 parameters.

name: SimpleStrategy
version: "0.0.1"
description: "Example of a simple strategy model"
markets: [
    Market(Binance('MY_API_KEY', 'MY_API_SECRET'), Pair('ADA-USDT')),
    Market(AscentEX('MY_API_KEY', 'MY_API_SECRET'), Pair('ADA-USDT'))
]
parameters: [
    Param(
        name=max_spread,
        type=float,
        default=1.0,
        description="The maximum spread value",
    ),
    Param(
        name=order_refresh_time,
        type=int,
        default=10,
        description="How far away from the mid price do you want to place bid and ask orders",
    )
]

The DSL currenntly provides buildins for defining Markets,Exchanges, Pairs and Params.

Note that at least one market and one parameter must be defined for the model to be considered valid.

A Market can be defined using the Market building, which requires 2 arguments;

1. The Exchange and 2) the Market Pair. Currently, the following exchanges are supported:

• Binance(API_KEY, API_SECRET)
• AscentEX(API_KEY, API_SECRET)

The Pair buildin has a simple argument; Pair("<MARKET_PAIR>").

To support definition of strategy parameters, the language provides the Param buildin. The following arguments are supported:

• name (str)
• type (One of: int|float|str|bool|list|dict)
• description (str)
• prompt_msg (str)
• default (int|float|str|bool|list|dict)
• keyword (str)
• dynamic_reconfigure (bool)
• prompt_on_new (bool)

Strategy Project Generation

To generate a new strategy project from an input model, use the hbot cli:

hbot generate <PATH_TO_MODEL_FILE>

The generator will create a new directory with all files included.

current_dir
  - my_strategy_model.strategy
  - gen/
    * start.py
    * my_strategy.py
    * __init__.py
    * conf_simple_strategy_config_map.py
    * conf_simple_strategy_strategy_TEMPLATE.yml

The generated strategy, in current version of the Hbot DSL, is dummy and does not include any logic. Though it prints the timestamp on each tick 🤘

Build the generated strategy

First, the HUMMINGBOT_REPO_DIR environmental variable must be set, pointing at the hummingbot client repo directory.

export HUMMINGBOT_REPO_DIR=~/Development/hummingbot

Next use the hbot build command.

hbot build <GEN_SRC_DIR> <STRATEGY_NAME>

And then recompile hummingbot client:

cd ${HUMMINGBOT_REPO_DIR} && ./compile

Examples

Simple Strategy Example

A simple strategy defined using the DSL is located in the examples/simple_example directory.

TODO

• Market Initialization in generated strategy
• Parameter Validator Concept in DSL
• Support parameter validators in code generator