In the Prediction wizard, you can select different input optimization levels above the inputs tab.
If you see an arrow circle beside an item, it means that it can be optimized. If the arrows are red, an optimization range has been set, while gray arrows means that an optimization range has not been set. A black lock means that the input won’t be optimized.
For example, No Input Optimization builds a prediction and backtests it using the exact inputs and parameters you entered. It is marked with black locks next to each input.
Input Parameter Optimization keeps all of the inputs you’ve entered, but changes parameters inside of the inputs, such as the number of periods used to compute the Relative Strength Index or the time series. This is noted with a large black lock on the inputs and a small red arrow circle if you set parameter ranges. If you don’t set parameter ranges for an input, you’ll only see a black lock.
Input Selection Optimization picks which inputs to use but doesn’t change any of the parameters or time series used in the inputs. This option displays a red arrow circle beside each input but the input parameters are marked with a small black lock.
Full Input Optimization picks the inputs and changes the parameters, and is denoted with red arrow circles beside each input and parameter range. Time series in the inputs are marked with a gray arrow circle, but may be optimized if more than one time series is chosen.
Custom Input Optimization lets you select some inputs and/or parameters to optimize. This option is marked with red arrow circles beside each input. You can lock entire inputs just by clicking on the red arrow circle until it turns to a lock. Only the inputs with red arrow circles will optimize.
You can optimize some parameters such as a time series and lock others.
Once you’ve decided on which inputs and parameters you want to optimize, the next requirement is to select an objective function. The objective determines the best entry, stop, and exit conditions during optimization of the Prediction. Click on modify Prediction parameters, then the Training Tab. Click on the drop down list to view all of the difference objective functions. The help file will provide the details.
Move to the Optimization tab, where you can select other optimization parameters such as the maximum number of inputs to the model, or how long to optimization. You can also set the desired shortest and longest trade span. The Trader attempts to satisfy this requirement, but the objective function is the primary goal.
If you choose to optimize the model across all chart pages, the optimizer is forced to choose the same inputs and parameters for each chart page.