We need a feed of ETH prices to drive simulation dynamics.

Our options here are to use a random process (Ornstein-Uhlenbeck) to completely model the ETH price, or to sample from a historical distribution.

We should start with the latter, though it will be confounded by ETH market idiosyncrasies from the sample time frame. This is preferable when investigating Black Thursday, but not for generalized hypotheses.

• add keeper_bid and keeper_bid_generator to policies.py
• create new models.py to hold bidding models
• assumed that state for keepers would change to object withkeeper id as key and bidding model value

We need to decide on system-wide and keeper-specific behavior regarding opening and closing Vaults.

We will definitely need a warmup period of Vault opening to start the simulation. However, we may also need a random process to model the opening of arbitrary Vaults over the course of the simulation, and potentially separate logic governing how Keepers interact with their Vaults.

Similarly, we will have to decide on the system semantics for Vault closure.

Since the Dai system doesn't explicitly react to price, this will likely be our price-reactive model component. At a high level, one is incentivized to open a Vault if they believe Dai price will go up, and to close a Vault if they believe Dai price will go down.

Implemented Frontend:

• using Victory, Material-UI
• merged Andrew's frontend demo code for connection
• needs testing

Includes landing page (optional), Graph component, Form component for parameters), Dark theme

Heya --

I stumbled upon your project late at night looking at the dai.js dependents and thought I'd drop by to share this jupyter-like dai.js notebook: https://observablehq.com/@jparklev/dai-js-interactive Feel free to fork it or edit it there. My sense is that it might be useful for what you're doing, but if not, no worries -- don't mind me

What stablecoin implementation will we use? Do we need to tweak any existing ones? Need to consider what implications this bears on the economic modeling platform and agent/asset classes.

Not sure if this is absolutely necessary for a simplified economic model, since traders will transact using just stablecoin, but we will need control over some asset classes / stablecoin value derivatives to accurately model crises (e.g. simulate asset bubble for 2008 Great Recession).

This includes defining how the assets are priced, and perhaps asset-specific trader behaviors.

We should definitely ask some experts who have worked in / understand the fields we're pushing into, i.e. economics and cryptocurrency.

We'll try to get a sense for which parts of the project are most feasible, and what directions to pursue for the most interesting and value-additive work.

To start this effort off, I think we should definitely hop on a call with Tanay Kothari and Catherine Gu from Stanford, who wrote the Simulating Stablecoin Systems with Latent Market Confidence Index paper.

Talking to Dev would also probably help, he'll probably be able to tell us what's feasible and what isn't, and generally poke holes in our project early on so we don't waste time unnecessarily.

Any other ideas?

We need to read up on context-building resources to gain a solid grounding in popular stablecoin implementations, the economic principles at play, and the details of the crises we want to model.

We should also see what previous work we can find in the realm of economic simulations, and ideally, stablecoin simulations.

As we do our preliminary research, we should add all pertinent resources that we find to the resources doc.

Let's aim to get this done by the end of Spring break. Of course, we'll be doing more reading and research as we go through the project, but we need to establish a footing now.

Oopsies, was using spot as the actual ilk price in a couple places - made the appropriate changes. Also, swapped out c_ratios param in keepers for spot_padding. Same idea, they have some risk tolerance dictating how collateralized the vaults they open are, but now spot_padding is multiplied by spot (the maximum drawable debt per unit of ilk) to determine how much debt to draw when opening a new vault.