How to use WinAnt to run JTAP

Transcription

1 Table of Contents How to use WinAnt to run JTAP... 2 Running JTAP game server and clients separately... 3 Trading Strategies... 4 Bidding strategy:... 4 Investment strategy:... 4 JTAP Market Manual... 6 Configuration In cat.params... 7 Traders Design Terminology... 8 Trading Strategies... 9

2 How to use WinAnt to run JTAP Step 1: Download and Install the latest version of SUN JDK from Step 2: For Windows system, download and install the latest version of WinAnt from you will need to install JDK first, because WinAnt will ask you to allocate the JDK folder when being installed, and you will need to log off / reset the computer after WinAnt is installed. For Mac system, Ant should be already installed and can be run from your terminal. Step 3: Download and unzip the latest version of JTAP from i.e. under E:\JTAP_Traders The manual is located in the doc folder and called JTAP Traders Design Manual Step 4: Open your terminal, and then navigate to the folder where you have unzipped JTAP, for example E:\JTAP_Trader Step 5: Once there, you can use the command $: ant all to build the.jar file in the same directory, and then use command $: ant run if you are under main JTAP folder OR $: java -jar jtap_0.17_all.jar, The above commands will run JTAP with default parameters (.params files under params folder) If you want to use other parameters to run JTAP, you will need to specify which.params file you want JTAP to run with. i.e. $: ant run -Dparams=params/another.params under main JTAP folder OR $: java -jar jtap_0.17_all.jar params/another.params

3 Running JTAP game server and clients separately All the above methods launch a complete JTAP game, including a JTAP game server, a set of competing specialists, and a set of trading agents. In certain scenarios, you may need to run the game server and the clients separately, e.g., during a JTAP tournament. To run a JTAP server only, run $ ant server To run a JTAP server and a set of trading agents only, run $ ant servermarkets This is typically when the JTAP tournament organizers run the JTAP game server and provide the trading agents themselves. To start a set of trading agents only, run $ ant traders Again, additional -Dparams=params/JTAP.params can be added in these commands when you choose to use a configuration file other than the default params/jtap.params. NOTE: When you run a JTAP game with the game server and clients separately, make sure that the configuration files used by the game server and the separated clients all use the socket-based method to communicate with each other since the alternative methods do NOT support interprocess communication.

4 Trading Strategies The trading strategy is divided in to two classes Bidding Strategy and Investment Strategy: Bidding strategy: This strategy is associated to each market, i.e. trader should decide one bidding strategy for each market. Note that, trader can use a single bidding strategy for different markets. We have provided one abstract bidding strategy and a simple example in the source code. There is one key function in the AbstractBiddingStrategy class: public boolean timetoshout(); This function helps traders decides whether or not to shout and if it is time to shout, the price and the quantity of the shout. Investment strategy: This is the section of the program where the investment decision goes. For example, how to allocate investment to all available markets and what bidding strategy should be used for each market. AbstractInvestmentStrategy class proposed a basic structure of the strategy. It contains the following two key functions. public void revise(); Revise function will be used to decide the trader's market portfolio, i.e. market selection, bidding strategy selection, etc. This function can be called anytime during the game to revise the trader's decisions. public void addinvest(double amount); This function decides how to distribute investment among all selected markets. Note: Participates are free to choose their own strategy structure design. The sample strategies offered in the code are just samples, which may not necessarily meet your need.

5 Traders JTAP Trading Agent Strategy 1 Strategy 2 Strategy 3 Investment Strategy Bidding Strategy Revise() Buy/Sell AddInvest() Investment Amount Timetoshout() Price Timetoshout() Quantity

6 JTAP Market Manual STG consists of a number of days (500 in the trail run), each day consists of 10 rounds, and each round lasts for a number of milliseconds. Traders submit shouts at the beginning of each round, and at the end of the round, markets will clear the shouts it received during that round. Market will also send out the average clearing price (market price) for that round's matching. Traders can use this information to analyse the market and make decisions. This information is stored in the MarketData for traders. Traders can access these data by using the following command inside their trading strategies: agent.getmarketdata(marketid).getmarketpricehistory(numofdays); Will return the market price for the specific day, which sets back the number of \numofdays". agent.getmarketdata(marketid).getmarketpricehigh(numofdays); Will return the highest market price in the past number of \numofdays". agent.getmarketdata(marketid).getmarketpricelow(windowsize); Will return the lowest market price in the past number of \numofdays". agent.getmarketdata(marketid).getmarketmeanpriceoflast(windowsize); Will return the mean market price in the past number of \numofdays". agent.getmarketdata(marketid).getmarketmeanprice(); Will return the mean market price so far. agent.getmarketdata(marketid).getmarketprice(); Will return the latest market price. agent.getmarketdata(marketid).getclosingprice(); Will return the latest closing price.

7 Configuration In cat.params The following segment defines trading agents configuration. The parameter base for trading agents is JTAP.agent. ############################################################### # trading agents # trading agents JTAP.agent.n = 5 size = 1 JTAP.agent.0 = edu.cuny.jtap.traderclient JTAP.agent.0.n = &size; cat.agent.0.id = trader1 cat.agent.0.investmentstrategy = edu.cuny.cat.trader.strategy.sampleinvestmentstrategy cat.agent.0.investmentstrategy.biddingstrategy = edu.cuny.cat.trader.strategy.samplebiddingstrategy cat.agent.0.investmentstrategy.biddingstrategy.windowsize = 20 cat.agent.1 = edu.cuny.cat.traderclient cat.agent.1.n = &size; cat.agent.1.id = trader2 cat.agent.1.investmentstrategy = edu.cuny.cat.trader.strategy.sampleinvestmentstrategy cat.agent.1.investmentstrategy.biddingstrategy = edu.cuny.cat.trader.strategy.samplebiddingstrategy cat.agent.1.investmentstrategy.biddingstrategy.windowsize = 30 The trading agent can behave in two different ways that is defined in the trading strategy as part of java class file; the first type is buyer and the second seller. Both of the above trader sub-populations has 50 agents and each agent is implemented as an instance of edu.cuny.cat.traderclient. TraderClient, inheriting edu.cuny.cat.gameclient. Cat.agent.id defines the identity of the traders, which is used to implement the trader in the program. Cat.agent.investment strategy has the parameter which will describe which type of investment strategy is to be used by this certain trader.

8 Traders Design Terminology There will be a number of inputs that will be provided to design a trading agent strategy. The following definitions describe the information that will be required by the participants of STG to design trading agent. Fund: Fund is the amount of capital available to the trader. There will be a specific amount that will be provided at the start of the game. Tips: the funds can be split evenly between the trading strategies so that there is a limit to invest in the markets. E.g. Allocate 50% fund to Stock A and 50% fund to Stock B so that the risk of losing your investment due to bad performance of one is reduced. Inventory: It is the quantity of various assets (stocks) that a certain trader holds at a specific time in the game. Profit/Loss: Value representing the amount a trader gained as compared to the initial capital held by that trader. In case of a loss it would be the difference of Capital minus the current fund, resulting in a negative value. Profit/loss = Capital Current funds available High Price: The highest price that a stock reached in a certain day. Low Price: The lowest price a stock reached in a certain day. Price: A certain value generated in-between the Market High Price and the Market Low Price of the stock. Closing Price: The price of a stock at the end of the day. Transaction Cost: Cost incurred by the trader during the completion of a certain transaction. This is the amount every market will charge to complete a transaction. Quantity: Number of stocks to be bought or sold by a trader during a certain transaction. Note: Quantity will not be provided as an input it has be part of the decision of investment for the trading agent. Shout Price: Price that a trader will quote for a specific ask/bid in the market.

9 Input Fund Market Price Closing Price Inventory Transaction Cost Trader Trading strategy 1 Trading strategy 2... Trading strategy N Output Buy /Sell Shout price Quantuty This Process diagram demonstrates how the inputs will be provided to the trader and then using an appropriate strategy the trader will produce an output which will be sent to the market visa the server. Trading Strategies To design our strategy we have decided to divide it into two parts, so that the first one is more of the analysis section called market indicator. Market indicator will help find the market trend and help make a decision. Second part of the strategy is where the decision of investment will be made based on the analysis performed earlier, so that is how much to invest and at what price. In-case trading agent owns the stocks then he can decide to sell these stocks to the respective market. Market Indicator: This section of the strategy can help the trader decide the trend of the market. i.e. Bearish, Bullish or Neutral. There are various types of indicators available to help the trading agent make investment decision. Trading Actions: The ability of the trader to invest in a certain stock will be given in this section. Once the trading agent identifies a specific trend in the market it s time to decide if trading agent is going to buy or sell the stock or decide on the quantity to and how much of his fund he wants to invest.