I built two algo trading strategies that topped the leaderboards. Then I rebuilt the platform they should've run on.

We're live on Product Hunt today — wanted to share the build honestly here first, the good and the not-yet.

Backstory. In 2024 I built two strategies, MFRider and GhostRider, on a third-party platform. Over the months they consistently topped the leaderboards — people made real money running them. But operating them was miserable: clunky testing, duct-taping exchange APIs to deploy, something breaking at 3am. The strategies were good; the platform was the bottleneck. TradeStaq is my answer.

What it is. Describe a strategy in plain English — "buy ETH when the 4h RSI drops below 30 and the daily trend is bullish, trail a 2% stop" — and FORGE writes the actual strategy code, backtests it on real historical data, shows the metrics, and deploys it to your exchange. Minutes, not weeks.

The parts that were actually interesting to build:

• FORGE generates real code, not presets — against our td.* API (indicators, raw OHLCV, multi-timeframe, persistent state). "Beyond the basics" just means more code; there's no ceiling at a preset menu.

• Assume, don't interrogate. For a vague prompt like "buy the dip," FORGE doesn't ask 20 questions — it picks a sensible interpretation and turns every ambiguous choice into a labeled, tunable parameter with a default (dip %, lookback, threshold), hands you working code, and lets you refine in plain English. The parameters + backtest are the safety net.

• Strategy Lab — an optimizer that mutates a strategy, backtests each variant, keeps what improves the score, discards the rest. A quant researcher iterating 24/7.

• Safety as a hard floor — the platform supports a stop-loss/take-profit even if the strategy omits it; a validation pass rejects code that can't exit or won't compile; and paper trading needs no API keys, so you can try everything at zero risk.

Honest limits (still building): funding rate, open interest, and mark/index price are modeled in the futures engine but not yet surfaced to custom strategies, and cross-exchange liquidation data isn't in the strategy context yet. Those are next.

Where we are: free tier, 15+ exchanges (Binance, Bybit, OKX, Hyperliquid, Kraken…), built largely with Claude + OpenAI. Launching today.

Biggest lesson: the wedge isn't "a bot" — it's collapsing test → ship into one motion and letting the AI discover better strategies. Everyone else sells presets.

Would love the IH crew's take on: (1) activation — what gets you from signup to first paper-traded strategy fastest? (2) trust — what would make you let software trade real funds? (3) if we surface funding/OI to strategies, what would you key off — funding flips or OI divergence?

Product Hunt: https://www.producthunt.com/products/tradestaq

Site: www.tradestaq.com