There are always going to be scenarios where profit is missed or the entry/exit isn’t called ideally, but I think you’ll find that over time, the math really starts working (at least historically).
In the case of this recent turn down, we saw some different behavior across the signals:
– BTC closed due to hitting a protective stop loss (i.e. didn’t predict the downturn)
– ETH predicted a further downturn and closed the position
– BCH predicted a further downturn and closed the position
There is one additional mechanism (trailing stop) to help protect profitable positions, but this didn’t kick in on any of these.
The models generally don’t call perfect bottoms or tops of trades, but they do usually grab a good portion of the way up and avoid most of the way down.
Regardless, I agree with you @juliandwillett…it’s tough to watch a profitable position vaporize. I’ve experienced this on my own trade decisions and when I follow a system.
We are constantly pushing on our ML capabilities. As I alluded to in this post, our 5.0 models are actually ready. But we need an infrastructure overhaul to handle them.
We’ll be sharing a post on this here shortly. 5.x won’t be perfect either, of course…but we’re committed to continuous improvement.
@Conwayddd, thanks for the note–I 100% agree. We certainly recommend everyone watch several trades to get a feel for the system and see if it makes sense for them or not. It’s easy to get excited about the great trades, but also easy to be disappointed with any loss.