Premium selection of algorithms
- Self-optimizing ARIMA expert
- Finite Impulse Response Neural Network
- Finite State Markov Automation
- Stepwise Best Regression
- Square Root Regression
- Square Regression
- Logistic Regression
Self-optimizing ARIMA expert
- Full ARIMA(p,d,q) implementation
- Unlimited order of mixed modeling
- Conditional error estimates
- Chi-square statistics on residuals
- Expert inference for optimal parameters
- Automatic trend adjustments
- Prediction on multiple future horizons
FIR Neural Network
- Finite-Impulse-Response (FIR)
- Optimal selection of filter parameters
- Adaptive neural network training
- Temporal back-propagation algorithm
Finite State Markov Automation
- Market data flow exploration
- Dynamically construct Markov models
- Building state transition graph
- Predict future market states
Stepwise Regression Algorithm
- Enter and remove predictors, in a stepwise manner, until there is no justifiable reason to enter or remove more.
- At each step, enter or remove a predictor based on partial F-tests.
- Stop when no more predictors can be justifiably entered or removed from the stepwise model.
History Prophet
- Dummy predictor for strategy testing
- Predicts every point with its future value
- Imitates a “prophet” knowing the future
- Delivers 100% of profitable trades
- Explicitly uses forward info
- Not suitable for practical trading
- Analog of “Maximum Profit System”
Extensible algorithmic API
- Modular algorithmic server
- Extendable calculation engine
- Real-time C++ core framework
- Open standard development API
- Universal DLL interface
- Compatibility with development tools
- Multiple sample models