A couple questions about your service/models:
1. It seems likely your models work internally with probabilities vs binary buy/sell decisions. Why do you not have the option to expose this probability, vs just simple buy sell signal? I would think this would pair very nicely with asset allocation. Have you investigated performance when using a sliding asset adjustment (even if just sp500 future & cash) that corresponds to model confidence, vs the binary win/lose bet system? Does slippage from frequent adjustment dominate the gains, no matter the adjustment threshold?
2. At a lazy glance, it seems trivial to boost performance with leverage on buy + shorting on sell, assuming the model really retains its backtest-heavy predictive power. Is this ignored just to remove black swan risk, or is there something more fundamental?
Obviously a 0 leverage buy sell signal is very marketable and low friction, fitting for a SAAS product, but it seems you could do even better with skills you obviously would possess if actually capable of making such a "god" level market predictor. Your posts are full of other seemingly more complicated strategies with somewhat contradictory capital levels throughout the years so it further adds to the main crux of confidence here: That you discovered a top 1% hedge fund caliber model individually as a mid tier SWE early retiree and suddenly offer it to all for an attractive price, without already being sure you can fuck off to a private island with the power of 50%+ CAGR, ignoring the rest of the world. You are also very sparse on details regarding your "proprietary anti-overfitting" methodology, when overfitting is of course the well-memed downfall of 99% of algo trading techniques that claim to meaningfully beat the market. To say nothing of the backtesting window conveniently starting after 2008, lack of VIX hourly data or not.
Disclaimer: I've spent the last 8 hours or so digging through your entire SA, reddit, and HN post history. Initially I dismissed your site as 80% likely scam, 19% naïve backfitted and overfitted waiting-to-be-raped-by-bear-market drivel, etc. But now I'd say I'm operating on a more optimistic 5-10% chance there is something legit here. You might actually just have researched sufficiently in all the right places, implemented a legitimate edge through extensive pareto culling via the yet virgin power of code, and had the perfect balance of libertarian desires to motivate you and yet progressive ideals to democratize it. As well of course as this being a genuine passion project that you are excited to share with others.
I currently work overemployed at FAANG, capital accumulation focused, but I've always held attempting to pareto market timing with a ~99-1 EMH assumption as my main FIRE project in the back of my mind for the past couple of years. I came to the weak conclusion with my pareto^pareto research of the idea that attempting to limit drawdown via algorithmic tactical allocation on SP futures (for liquidity) via macro + TA indicators seemed the best method. So, I am very intrigued to see someone who essentially appears to be me born 5 years earlier doing that exact thing.
Rereading this, I don't mean to come as aggressive/hateful; regardless of how the future treats your models this is a very thought provoking and exciting project!