Adaptrade Builder
- Features
- Six code output options
- Build over multiple markets
- Build with multiple data series (data2, data3, etc.)
- Bar types: intraday, tick bars, range, daily, weekly, monthly
- 21 order types
- 44 built-in indicators
- Custom indicators
- Neural network indicator
- 100+ build metrics
- 13 adjustable parameter ranges
- Stress testing
- Ideal trade matching
- Out-of-sample testing
- Build quality tracking
- Optional dark user-interface mode
- Multi-core processing for fast build times
- Includes Adaptrade Editor
As an example, let's say you want a trading strategy for a small portfolio of stocks – it could just as well be futures, forex or other instruments – and that you want your strategy to have a profit factor greater than 2, no more than 20% drawdown, and a nice straight equity curve. You want winning trades to last, say, five days on average. To have Builder generate such a strategy, you would load the price data for the symbols of interest; enter your requirements for profit factor, drawdown, correlation coefficient, and average bars in wins; then press the Build button. Builder will use an advanced genetic programming algorithm – a type of artificial intelligence (AI) – to evolve your strategy while you watch. When it's finished, you can review the performance results and strategy code and either accept it as is or modify one of the settings and build again.
Builder has a variety of settings that allow you to customize the results, such as specifying the types of trading orders used for entry and exit, the set of indicators that can be considered for use in the strategies, and whether the strategies are long-only, short-only, or include both long and short trades. You can also specify the range of available values used for indicator inputs and order price calculations. Other options let you limit the number of entries per day and specify the entry and exit times for intraday strategies. You can also add a neural network as an entry condition, control the strategy complexity to limit the risk of over-fitting, and specify how much data is set aside for out-of-sample testing.
