Totally forgot I was weighting by volatility. Furthermore, I intended to use inverse volatility, not normal! So the above results are somewhat expected.
I quickly tested with equal weighting and got better results. I also tried filtering my fundamentals by "free cash flow (FCF)." Long story short, filtering for positive FCF seems to slightly outperform not filtering it -- when excluding MGT (27925).
MGT gives ~2000% returns, but it's negative free cash flow. Thus, the FCF strategy misses the phenomenal returns. However, I consider MGT to be anomalous -- thus until research proves otherwise, it seems like FCF is the better strategy.
Unfortunately, I can't find how to attach multiple backtests (not even in a notebook??)
Below is an algorithm that tries it's hardest to make independent trades every month and hold for a year -- with fundamentals filtered to include negative EV and positive free cash flow. Any feedback is welcome.