I've been doing a lot of thinking about TMFs lately. I like the concept and features in Zorro very much. I'm still learning how to fully utilize them.

This post is about WHEN and WHERE to incorporate a TMF. I have already a process for designing a tradebot that I think makes logical sense. Essentially, you start from a very simply core logic, then add layers on top of that logic. The layers are like filters.

Some steps in the process are more like overlays or add-ons that, in my opinion, should not interfere with core logic triggers. An example of this would be enabling equity-curve trading. I test for and (potentially) activate this as an overlay, only after core logic is optimized/Train'd. The function of using equity-curve trading is a switch on/off that in-and-of-itself does not affect the core logic.

My thinking is that a TMF falls into this same category. If that were the case, then it implies that a strategy would be fully developed without the TMF at first. Later, after Train, a variety of TMF overlays could be tested as filters. I think it is likely that different asset (personalities) will perform better with different TMF logics.

For example, I'm currently trying to build a TMF that can get me out of a trade that has stagnated. I want to identify a trade that has experienced some momentum but then has (too much) consolidation. While that trade is open and moving sideways, it is a risk. I'd rather look at a TMF closing the trade and looking for a new signal in this situation. Alternatively, I'm also looking at a similar type of TMF that would manage drawdown in the same way (ie, original momentum shifted).