Why Edit the Formula
|Top Previous Next|
When ChaosHunter creates a formula, the optimizer chooses the formula components from your initial selection of inputs and operations, based on the optimization settings and formula parameters you choose such as population size, training objective, max equation size, max lookback period, etc. When you see the final formula produced by ChaosHunter, you might wonder what results it might produce if you changed one or several of the formula terms. The ChaosHunter Formula Editor allows you to make those changes and apply the model to your data file.
Edit a ChaosHunter Model
You can use the Formula Editor to make slight changes to a ChaosHunter formula. For example, if the ChaosHunter formula includes a term that is the slope of the close over 3 periods, you might want to experiment with a slope calculated over 10 periods to determine if the model is more effective if it looks farther back in time. Since ChaosHunter will not let you modify models during the optimization process, you could not play these type of "what if" games without the Formula Editor.
Another example would be if you want to fine tune the terms inside a ChaosHunter formula for long and short trades. (When ChaosHunter creates a model for both the long and short side, it uses the same formula and evolves different buy/sell threshold values.) The Formula Editor gives you the power to change and test different terms inside of the formula itself, in addition to modifying the threshold values found by ChaosHunter.
Add a Missing Element to a ChaosHunter Model
If you believe that a particular operation such as a neural net would be beneficial to your model, but the optimization process has not included neural nets in your formula so far, you could let ChaosHunter continue to evolve models and hope that a new model includes neural nets, or you can use the Formula Editor to tweak an existing ChaosHunter formula or write a new formula that includes neural nets. ChaosHunter can immediately test this new model on the optimization data set, the out-of-sample set, or both. With immediate feedback you can then either return to the Formula Editor or accept the model.
Quickly Prototype a Model
If you already have a set of rules in mind for your model, you can use the Formula Editor to create a formula with those rules, even if you have not yet created a formula with ChaosHunter. If you run a number of tests, you could feed the successful inputs and operations to ChaosHunter to evolve an even better formula.