AI - Gold (3) After surviving the hellish market conditions of V1 and V2 last week, I did a review. The reason V1 and V2 performed poorly last week was because gold basically went on a one-sided trend with almost no rebound... For these agents, it meant most of the high short orders didn’t get triggered, and all the low long orders got wiped out... So starting last week, I added an improvement—a regime filter. Using H4 to apply a trend filter... Filtering out counter-trend orders... Let’s see if this makes things better... Still testing. As for Gold V3, it’s the result of a sudden inspiration this week... Since we already have 5 weeks of V1 data and 4 weeks of V2 data... Can we treat this as a horse race and pick out the winning horses (agents) to form a better portfolio? That’s what V3 is about... Not sure how effective this simple method of bundling agents will be... Will the agents that performed well over the past 5 weeks continue to perform well over the next 5 or even 10 weeks? This is probably the problem V1-V3 will need to keep solving going forward... At least for now, V3 looks pretty good...