This repository holds the cadCAD model created by BlockScience for Olympus DAO. It can be used to test a wide variety of simulations and scenarios.
.
├── README.md
├── Single Example.ipynb - Example of a simple single run
├── Sweep Example.ipynb - Example of a parameter sweep run
├── legacy - Folder with legacy notebooks no longer used
├── model
│ ├── behavioral - Folder holding the behavioral functions of the system
│ ├── classes - Folder for the classes used
│ ├── mechanism - Folder for the mechanism functions (state updates)
│ ├── policy - Folder for the policies used in the model
│ ├── psub.py - Python file that creates the partial state update blocks
│ ├── psub_functions - Folder holding the functionalities for PSUBs
│ ├── run.py - Python file for code involved in running the model
│ ├── signals - Folder of functions for creating random signals (i.e. a pattern of bond disbursements)
│ ├── types - Folder holding the types utilized in the system
│ └── utility - Folder with utility functions for ease of use modifications
├── requirements.txt - Python packages required
└── research - Folder for all research notebooks/analysis
└── 20230210 Exploratory Research - First cadCAD based model research notebooks
1. Pre-installation Virtual Environments with venv (Optional):
It's a good package managing practice to create an easy to use virtual environment to install cadCAD. You can use the built in venv package.
Create a virtual environment:
$ python3 -m venv ~/cadcadActivate an existing virtual environment:
$ source ~/cadcad/bin/activate
(cadcad) $Deactivate virtual environment:
(cadcad) $ deactivate
$Requires >= Python 3.6
Install Using pip
$ pip3 install cadcad==0.4.28Install all packages with requirement.txt
$ pip3 install -r requirements.txtYou can see a simple example of how to use the model with just one set of parameters in "Single Example.ipynb".
You can see an example of how to run parameter sweeps in "Sweep Example.ipynb"
demand_factor: range (0,+inf), marking how much random demand will be at each time step.
supply_factor: range (-inf,0), marking how much supply demand will be at each time step.
panic_sell_on: True or False. If panic sell is going to happen.
panic_param: Range (0,+inf). Bigger the larger panic sell amount.
initial_reserves_volatile: range [0,+inf]. The reserve has two parts, stable and volatile and this markes the value of the latter. Not interacting in the current model.
max_liq_ratio: range(0,1), marking the ideal ratio of liquidity_stables / treasury_stables. Every 7 days the treasury will take actions according to whether this ratio is reached (detailed in model.policy.treasury.p_reserves_in).
max_outflow_rate: ranging (0,1), marking the max reserve outflow ratio for the 7-day liq_ratio adjustment behavior.
reward_rate_policy: default Flat". Does the protocol change their reward rate for staking, which impacts the supply.
target_ma: number of days for moving average price target.
lower_wall: range [0,1]. Ratio of the lower wall price to the target price.
upper_wall: range [0,1]. Ratio of the upper wall price to the target price.
lower_cushion: range [0,lower_wall). Ratio of the lower cusion price to the target price.
upper_cushion: range [0,upper_wall). Ratio of the upper cushion price to the target price.
reinstate_window: range [0,+inf). How many days before the capacity get refilled.
ask_factor: range [0,+inf). Ratio of the reserves that the treasury can deploy when price is trading above the target
bid_factor: range [0,+inf). Ratio of the reserves that the treasury can deploy when price is trading below the target
cushion_factor: ranging [0,+inf). The percentage of a bid or ask to offer as a cushion
min_counter_reinstate: range [0,reinstate_window]. Number of days within the reinstate window that conditions are true to reinstate a bid or ask
with_reinstate_window: default 'Yes'.
bond_create_schedule: a list of all ohm bonds.
bond_schedule_name: name of each bond schedule.
bond_annual_discount_rate: range (0,1), for estimating the bond price based on the tenor.
ohm_bond_to_netflow_ratio: range (0,1), for estimating how the buy and sell of ohm bonds will reflect in the netflow towards the liquidity pool.