Built-in forecasting expert works on unique collection of elaborate predictors, technical indicators, digital filters and statistical tests to achieve top forecasting accuracyand reliability in full automated unattended mode. Not offered in any other product.
Trading performance criteria
Free open API enables unlimited system extensibility both in data access and additional forecasting technologies which then are native consumed in system kernel expert reasoning. SOAP, DCOM and C++ bindings give easy direct integration with virtually any enterprise infrastructure including Java, NET, Delphi, MS Office, online portals and interactive web services.
Below we provide the full table of algorithms available in ForeStock. Algorithms are grouped according to their packages. Packages are separate modules in common algorithmic space and are licensed individually. You can watch names of packages in License Manager. Licensing any package implies all algorithms contained inside it.
|ARIMA with expert model fit||Seasonal Auto-Regressive Integrated Moving Average forecasting model with automatic expert inference on all model parameters.||Predictor|
|Finite State Markov Automation|
|Finite State Markov Automation||We dynamically construct Markov models that describe the characteristics of Market data flow. Such models are used to predict future market states.||Predictor|
|Finite Impulse Response NN|
|Finite impulse response neural network||The finite impulse response neural network is a neural network, where scalar weights are replaced with moving average filters. These filters compute a weighted average of past values presented to the network, as opposed to the feed-forward network, which only computes a weighted “average” of the current value. These networks are trained using a variation on the backpropagation algorithm.||Predictor|
|Forecast with average value||Classical moving average with period 20||Predictor|
|Linear regression||Linear regression liney = at + bcalculated over 20 last points||Predictor|
|Exponential Fit||Exponential regression curvey = eat + bcalculated over 20 last points||Predictor|
|Logarithmic Fit||Logarithmic regressiony = log(at + b)calculated over 20 last points||Predictor|
|Logistic Fit||Logistic regressiony = c / [1 + e-(at + b)]calculated over 20 last points||Predictor|
|Square Fit||Parabolic regressiony = (at + b)2calculated over 20 last points||Predictor|
|Square Root Fit||Square root regressiony = (at + b)1/2calculated over 20 last points||Predictor|
|History Prophet||Emulates “ideal” predictor. Forecast is set to real next observed value, which ensures 100% forecasting accuracy on historical data. It is very useful to calibrate performance of trading strategies in “ideal” conditions. In no case, it should be used as predictor in real trading.||Predictor|
|Naive Predictor||Forecast with the previous price. Dummy forecast to evaluate performance of other algorithms.||Predictor|