Options For Coffee

Options for Coffee is a simplified backtesting tool (or toy) for trading algorithms. It works by randomly generating stock prices (from specified parameters) and running the strategy on it.

Installation

Download the .jar file from the releases page, and execute it. It may probably want permissions depending on your system (This type of file can harm your computer etc..).

Usage

OfC uses a Geometric Brownian Motion to simulate stock paths, that is,

S(t) = S(0) exp(Q(t))
Q(t) = theta * t + sigma * W(t)
W(t) = Standard Brownian Motion
theta = mu - sigma^2 / 2

(Please note that, this model is the most basic of all models used to simulate stocks).

So, in the interface, under Stock Properties, you need to enter your desired value of mu and sigma. You can check how a sample path looks like using Show Sample Path. For good results, it is recommended to use values of mu and sigma so that theta above is pretty small. Try out mu = 0.006, sigma = 0.015 for inspiration.

Now once you're happy with your paths, you need to write a strategy. Strategies for OfC are written in JS (ES5) and are explained below. Once you write a strategy, select it, and set Iterations, and the Risk Free Rate. By default the Risk Free Rate is set to 2.41% a year, which is a good default. But ofcourse, you're free to experiment.

When you test your strategy, OfC will take some time (Issue: for many days, and iterations, specifically when days * iterations is large, the execution blocks. I need to put the evaluation in a separate thread or something). After that, it will display on the left Results pane a graph plotting the average position of your strategy, by day. On the left pane you'll find the distribution of final returns of your strategy. You can also view logs using Console Logs. However, note that the total log size is limited to 100,000 characters for now, so please be frugal in your console output.

Strategy

Strategies in OfC are written in JavaScript. Your file (name it whatever), has to implement exactly one function, with the exactly following signature.

// You can change "agent" to whatever but 
// your function should be named "strategy"

function strategy(agent) { 
 // your strategy here
}

The interactor will call your function, by passing it an object which it can use to retrieve data about the market. The list of actions / functions are listed below. Note that your function will be called every'day', so your strategy should only implement the action for a day (depending on today's and past information received from the agent).

Here is a sample strategy, that is extremely bad:

function strategy(Q) {
	var curday = Q.getCurrentDay();
	
	if (curday >= 1 && Q.getStockPrice(curday) < Q.getStockPrice(curday - 1)) {
		Q.stock(0.1);
	} else {
		Q.stock(-0.1);
	}
}

Interactor functions

The agent object is referred to as Q below.

Method	Description
Q.getStockPrice()	Returns current day stock price.
Q.getStockPrice(day)	Returns stock price on given day. `day <= Q.getCurrentDay()`
Q.getCurrentDay()	Returns current day.
Q.stock(amount)	Amount of stock to buy. Set amount to negative if you want to sell.
Q.callOption(direction, strike, maturity)	Buy/sell a European Call Option, with given strike rate and days to maturity. Set direction to Q.LONG or Q.SHORT for the direction.
Q.putOption(direction, strike, maturity)	Buy/sell a European Put Option, with given strike rate and days to maturity. Set direction to Q.LONG or Q.SHORT for the direction.
Q.getCallPrice(strike, maturity)	Returns call option price for given strike and maturity.
Q.getPutPrice(strike, maturity)	Returns put option price for given strike and maturity.
Q.log(loggingInfo)	Output loggingInfo to console. It will be logged as [Iteration # : Day #] : loggingInfo
Q.getCurrentCash()	Returns the current cash position. Maybe negative if borrowing.
Q.getRiskFreeRate()	Returns the current risk free rate. This is the risk free rate you select divided by 360.

Bugs