Time and time again, it is easy to make simple, single indicator strategies (or more complicated) that have great optimization results but completely fail when forward tested. Attached is a screenshot of one such simple strategy's optimization (it's basically just the d9ParticleOscillator with money management) that has excellent results but does terrible going forward.
This leaves me to conclude that one of the most important parts in being successful with this endeavor is setting it up right, not just the strategy itself.
From my researching it, there are 3 principles to consider:
The time interval used may or may not be more generally applicable. Ex: 1, 2, or 5 minute increments.
The variables that are optimized. For example, SQN, Max Expectancy, and Net Profit.
The amount of prior time used for the amount of future time. Ex: 1 month for the next week. Past 6 months for future 1 month, etc. How often it's re-optimized.
The answers to these I am currently researching. These answers would be beneficial to all and with community input, we can no doubt reach a better answer than otherwise. What are your thoughts/experiences in this matter?