lcfp / agent-based-stock-market-model Goto Github PK

I think I found a bug in the market matching mechanism. It returned the wrong agent set. And it appended to a list of matched agents which I believe is no longer necessary. I added a TODO there. Mark could you have a look to check that I did not make any mistakes?

Make sure the model runs with traders using either fundamentalist and chartist strategies.

Including the new file structure for parameter config files and databases.

@joerischasfoort do make sure you include the /experiments folder into the .gitignore! :)

Automate testing

Test all current model code

Julian's existing code for a new, more aggressive rule.

Dependency on #25

Minutes Meeting 10/03/2017: To do list
Minutes.docx

In a new notebook. Dependency on #33.

I describe two models that offer different ways of modelling the endogenous determinate number of chartists and fundamentalists in the stock market. This has important consequences for price and market volatility.
FundamentalistChartistBehaviour.docx

Feel free to spend as much time as you like on this @ChristopherStockermans :)

We added a market matching mechanism function which had as input an agentSet and wanted its output to be a list of matched agents. Then in the simulation.py we could let these agents have a transaction. However, we now feel that this might lead to a very weird market. Because suppliers can only serve demanding agents once or too often. We also do not take agent inventory of stocks into account.

Therefore, we are now leaning toward a new type of function with input agentset and output an agentSet after matching AND TRADING. Mainly because matching depends on trading.

Work in progress.

Formulate both a chartists and fundamentalist valuation function. Ideally the function has as input an asset and returns a valuation.

See #12, #14 for more info.

Dependency on #25