Verstats
Turn clunky MT4/5 reports into institutional quant metrics.
I just officially shipped the core engine for my side project, Verstats.
The Problem I Was Trying to Solve. By day, I work as a Risk Analyst at a liquidity provider, and my background is in infrastructure software engineering. I have been building my own automated trading algorithms on the side, but the default reporting tools in the industry standard software (MetaTrader 4/5) are terrible. They only give you a single historical path and hide massive tail risk. To get proper institutional-grade metrics, I had to manually export multi-million row log files into Python to run Monte Carlo simulations. I got tired of that friction, so I automated the whole pipeline.
What is Verstats? It is a web app where algorithmic traders can drop their raw legacy MT4/MT5 HTML reports and instantly get deep quant metrics, trade distribution charts, and Monte Carlo stress testing.
The $0/Month Architecture. Since I am starting with zero users, my primary goal was engineering the backend to handle massive data parsing without spending a single dollar on server costs.
Frontend & Landing Page: React + Astro and Vite + TSX, hosted entirely on Cloudflare Pages (Free)
Backend: Python FastAPI and Polars (migrated from Pandas for memory efficiency), running on Google Cloud Run (Free Tier)
Database: Supabase (Free Tier)
File Storage: Google Cloud Storage (Free Tier) to handle the massive report uploads
Payments: Paddle (Merchant of Record, purely transactional so no fixed costs)
The Pricing Experiment: Tokens > Subscriptions. I deliberately avoided the classic SaaS monthly subscription model. Backtesting is inherently bursty. You might run a hundred tests over the weekend and then do nothing for a month while you tweak your code. Charging a recurring monthly fee for that felt forced.
Instead, I built a pay-as-you-go token system. Everyone gets a free allowance of tokens that refills every hour. If you are doing a massive backtesting session and run out, you can just buy a one-off top-up pack to keep going.
What I'm looking for right now: Since the core parsing engine is finally live, I am shifting entirely to marketing and UX.
If anyone has experimented with token or credit-based pricing for "bursty" utility tools, did it work for you? Do you think subscription model still have its place for this type of SaaS?
I am terrible at marketing copy. If anyone wants to roast my landing page (verstats.com) and tell me if the value proposition is clear, I would massively appreciate it.
1 Comment
The bursty usage pattern makes the token model an interesting bet.
Have you seen enough real usage yet to know whether customers actually prefer pay-as-you-go, or is that still an assumption you're testing?