I have been trying to work out which automated trading software to use. I was leaning towards Ninjatrader but then read a bit about Tradelink. Does anyone have any information on which to use/another option? Or can answer any of these questions: Do either have the ability to perform monte carlo backtesting? Risk Management - sizing positons based on a % risk? Ability to use fundamental data? Portfolio Testing Multiple Strategy testing Multiple entries/exits (pyramiding etc) Thanks guys
Do either have the ability to perform monte carlo backtesting? There is a random tick/data generator in tradelink. The bundeled backtesting apps (Gauntlet and Kadina) do not natively support monte carlo methods. However gauntlet and kadina are just GUI wrappers to the same historical engine. Since tradelink is open source, you could just copy the code and add in an option/call to random data generators. Risk Management - sizing positons based on a % risk? Yes. This is about 1 line of code in a response/strategy. Ability to use fundamental data? Ability yes, you can call any 3rd party fundamental source you may have access to from your strategy or application. There is no bundled fundamental library as part of tradelink though. Portfolio Testing Yes tradelink gauntlet supports this natively. Multiple Strategy testing The off-the-shelf applications (gauntlet/kadina) do not support running multiple strategies simultaneously. The simulator/engine does support doing this however. So you could roll your own app if you like. Additionally, There is a user on the mailing list who has also rolled his own version of a similiar idea that works inside of the off-the-shelf tools. Multiple entries/exits (pyramiding etc) Yes http://tradelink.googlecode.com
What is the point of "monte carlo backtesting"? DO you want to avoid buying hisorical data? Or do you believe you can create a trading system that is profitable on purely random data? What does that mean?
Thanks TradeLink. Does anyone know the answer to those questions with regards to Ninja? @LeeD It's not about having to buy historical data, its that this does not provide a large enough sample size. Also most (all) Monte Carlo simulators have data specifications so are not fully random. I think bootstrapping is nearly as good. Multiple Strategy testing involves backtesting multiple strategies at the same time on the same data. The reason why this is important to me is you can see why/when strategies work together, but most importantly get a more advanced system that has preferential trades based on profitability etc.