Simulating the financial system
Hello,
I am doing an academic project in which I need to implement a financial market system simulation in order to investigate what happens when multiple agents(traders) use certain trading strategies in the market. It doesn't have to be super realistic but I do need to state my assumptions clearly.
I am not that well clued up about the financial world yet and was wondering if I could get some help with this.
Here's the plan I've made so far for the very simple model of a market:
-1-N Market participant(MP) and one market(M).
-(M) has one stock.
-(MP) can buy or sell the stock.
-the stock price is drawn from a normal distribution.
-act of the (MP) buying and selling the stock changes the mean and standard deviation of the normal distribution.(due to demand and supply)
-MP has a strategy or a set of strategy according to which it chooses whether to buy or sell.
-the quantity bought or sold can be decided by a normal distribution, can be set prior or can be decided by the strategy.
-MPs are trying to maximise profits
have to account for:
-A large number of MPs.
-The level up to which an MP may influence the market.
-MPs will have different strategies.
-History of market prices.
-black swan events.
Your assumption set needs work. What sticks out immediately is "the stock price is drawn from a normal distribution" which is absolutely inappropriate. Stock prices can only be positive so the distribution you'd be using is the lognormal, not the normal. Additionally, if the market contains only one stock, all that will occur with multiple agents is that they'll converge onto an equillibrium price. The other assumption that seems peculiar is "the quantity bought or sold can be decided by a normal distribution, can be set prior or can be decided by the strategy." I don't see how this works with your agents having a pre-defined strategy, is the quantity bought/sold a random variable, or is it pre-set by the agents?
Good start, but even a simple simulation of a financial system is a massive undertaking. What is the project for? Graduate work I presume?
It's an undergraduate project. As I said I need to investigate what happens when agents start using certain trading strategies.
How do you recommend I do this? Why would the agents converge to the equilibrium price? if the prices themselves are sort of random, and I introduce some black swan events, won't it mean that the agents won't find an equilibrium?
Off the top of my head, it's probably easiest to form your assumptions around your agents (i.e. their strategies, risk aversion, etc.) and set up your market conditions as deterministic states which would make including black swan events much more straight forward. That would be a pretty simple way to accomplish what you're talking about, but since I'm not sure what the requirements of your assignment are, it may not be appropriate.
This is an information problem. You need bayesian agents and non-bayesian agents. Additionally, you have to include a coalition game. The black swan thing is overreach. You can't 'must' do everything -- you obviously have no idea of how to expand parameters ... the empirical data can be fed into a Nelson and Winter model. They pretty much have the only model where there are 2 traders based on info.
wtf normally distributed? you can't have a black swan model with that... or even a log one.
You need to read the literature, some of your assumptions simply make no fucking sense.
Well the idea was to keep it very simple to start with. And you're absolutely right I need to understand a lot more about this.
This is what I understood from what you wrote, if you can expand on anything I'd be really thankful.
By Bayesian agent, I guess you mean an agent with random parameters, and non-Bayesian an agent with parameters that I specifically set.
What's coalition game? Again after googling my guess is that agents may cooperate with each other, this is an advanced feature that I can add later on, once I get the basic skeleton of the thing working.
For the black swan, I can simply introduce random events that will alter the price by much more then usual.
I have no idea how to expand parameters, what do you mean by that? And what are the Nelson and Winter models?
Again this has to be a very simple simulation of the financial system. And ATM I am trying to find the most basic assumptions for it.
Everyone hold up. Let's start from first principles...
Are you going to use game theory?
Or are you going to use past data (you can make it up... thereby creating the financial system) and project it into the future?
There's either a aye or nye on 1 or 2... you can't do both. I don't think you're sure of the differences involved. fyi, what year are you in college?
We need you to explain the project's difficulty and the expectation of the math involved for an A.
The aim of the project is to investigate what happens when multiple agents in the financial market start using adaptive trading rules that use machine learning. So for example, what happens to the financial system once 60% of market participants use these adaptive trading rules.
My prof suggested that I simply make up the market data from some kind of a distribution. However, this will mean that I can only investigate the performance of these trading rules and can't see what happens to the financial system, when the majority of the market start using those rules, hence I need the market to somehow react to the actions of the agents.
And to start with I need the simplest version of this simulation to work. Once I do that I can always build it up.
Vel et totam et consequatur illo. Alias et ipsum qui cum esse quo placeat. Amet ratione dolorum corrupti vel et adipisci molestiae.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...