Once Profitable Hedging EA needs rescuing!

Hello Traders/Developers/Strategists

I worked with a developer 2 years ago to create a hedging strategy.  For 9 months, I made great money with it, and made a nice return (97%) and thought everything was working great!  (see screen shot of myfxbook performance)  Unfortunately, my strategy hit a snag and in a matter of 3 weeks, I lost all of the money I had made and then some!

After taking a lengthy break to get my motivation and confidence back, I decided to go back to testing and try to see if there is a way I can correct the issue that occurred.  I am fully aware of the limitations of using a hedging strategy... they work great in trending markets, and falter in sideways markets.  To protect against this, we built in a mechanism inside the EA to expand the parameters once the EA detects it is trapped inside a sideways market.  This has its pitfalls though as well because each trade gets larger and larger.

To keep it simple, my trades are initiated based on several conditions trying to detect momentum.  If the conditions are met, the EA places a trades in the hopes of course that the profit target is met.  If the market goes against me a certain amount, I place a trade in the opposite direction (obviously, this is what hedging is all about).  Based on my testing of several currencies from 2008-2017, 97% of the time my EA creates only 4 hedges or less.  So, getting trapped in a sideways market is statistically rare for my EA, but it does happen.  And it can get really nasty if I can't get out of this hedging trap.

So, my options are the following:

1) Scrap hedging altogether because it is impossible to figure out a way around the 3% chance that too many hedges will happen

2) Evaluate my entry strategy a bit more and determine if there are additional ways to filter out trades which are being placed during a sideways market.  This is very tough because there are many times a market can be trending with lots of momentum, and can flatten out quickly afterwards.

3) Accept that sometimes getting trapped in a sideways will happen, and determine if there are ways to better mitigate the issue.

4) Do a combination of #2 and #3

One thing that has interested my lately is the concept of AI machine learning and genetic algorithms.  Instead of trying to wrack my brain to come up with the answers on my own, maybe the EA can utilize machine learning and produce the answer on it's own?  I am certainly open to this approach and welcome anyone who knows anything about this approach.

Also, if there are any traders or developers who have faced the issues that I mention above and has experienced overcoming them, please let me know!

Trey