Hi everyone,on a daily basis i need to travel a lot. So to overcome the time in the train i implemented a zero-intelligence model of agents in kdb+/q to simulate a limit order book.Actually i took the thesis ASYNCHRONOUS SIMULATIONS OF A LIMIT ORDER BOOK from Gilles Daniel and implemented the algorithm 1 in kdb+/q.Now all the code is uploaded on the contribution site of kx.http://code.kx.com/wsvn/code/contrib/kuentang/zia/If you want give it a try you need to start 3 q processes using the following lines:C:\q\w32\q.exe qx.q -p 2009C:\q\w32\q.exe zia.q -p 2010q) \t 100q) obkC:\q\w32\q.exe user.q -p 2011The first q process is there to act as an exchange. The implementation is inspired from the site gbkr.com.The second one is the implementation of the zero-intelligence model of agents.The last one gives you the opportunity to interact with the exchangeIf you have questions, comments or even ideas to improve it, please feel free to reply.Why did you do it?Please read the beginning again. Moreover using the zia i believe you can measure the market impact of your strategy. This is something that you cant backtest with historical data.As you said there is already an implementation from Gordon Baker of an exchange. Why did you implement it again?Actually in the beginning i implemented the zia on top of the code from Gordon Baker. But later on i want to try the idea of using a finite state machine to build an exchange. So that’s why i reimplemented it using the fsm.What is a finite state machine?You can google it.Regards,Kim
Excelent idea and great work.Nice example for me how to use many process.Some interesting work for OB is natlab nad work from PHD groupBut idea with FSM …very intresting..On 10 Lut, 23:18, Kim Kuen Tang <kuent…> wrote:> Hi everyone,>> on a daily basis i need to travel a lot. So to overcome the time in the> train i implemented a zero-intelligence model of agents in kdb+/q to> simulate a limit order book.> Actually i took the thesis ASYNCHRONOUS SIMULATIONS OF A LIMIT ORDER> BOOK from Gilles Daniel and implemented the algorithm 1 in kdb+/q.>> Now all the code is uploaded ?on the contribution site of kx.http://code.kx.com/wsvn/code/contrib/kuentang/zia/>> If you want give it a try you need to start 3 q processes using the> following lines:>> C:\q\w32\q.exe qx.q -p 2009> C:\q\w32\q.exe zia.q -p 2010> q) \t 100> q) obk> C:\q\w32\q.exe user.q -p 2011>> The first q process is there to act as an exchange. The implementation> is inspired from the site gbkr.com.> The second one is the implementation of the ?zero-intelligence model of> agents.> The last one gives you the opportunity to interact with the exchange>> If you have questions, comments or even ideas to improve it, please feel> free to reply.>> Why did you do it?> Please read the beginning again. Moreover using the zia i believe you> can measure the market impact of your strategy. This is something that> you cant backtest with historical data.>> As you said there is already an implementation from Gordon Baker of an> exchange. Why did you implement it again?> Actually in the beginning i implemented the zia on top of the code from> Gordon Baker. ?But later on i want to try the idea of using a finite> state machine to build an exchange. So that’s why i reimplemented it> using the fsm.>> What is a finite state machine?> You can google it.>> Regards,> Kim</kuent…>