Development Roadmap

A living overview of where Freqle is and where it's headed — the search engine and community for freqtrade trading strategies. This page grows with your feedback.

✅ Done

Search thousands of strategies — by name, by indicator or method, or upload your own .py to find similar ones.
A page for every strategy — full source code, automatic tags, community ratings and discussion.
Real tests you can run — sandboxed backtests, forward tests, and lookahead / recursive bias analysis.
Robustness scoring — a plain Robust / Fragile / Unreliable verdict for each strategy, from a Monte-Carlo range of likely outcomes and a significance test, so a big headline number can't hide a fragile edge.
AI Search — describe what you want in plain words ("mean-reversion with RSI, shorts only") and a small local model turns it into precise filters; it runs entirely on our own server.
Interactive charts — zoom and pan through equity, drawdown, profit per pair, and every trade and exit.
Rich metrics — CAGR, Sharpe, Sortino, Calmar, win rate, max drawdown, profit factor, best/worst trade, plus return-vs-market, expectancy, trade excursion (MAE/MFE), rolling consistency and per-month market-regime tags.
Back-tests over key crypto-market periods — one-click tests against notable stretches: the 2021 bull run, the 2022 crash, the current cycle and more.
The Strategy League — a transparent leaderboard ranking the best backtested strategies by a clear, weighted score, weighted towards edges that leverage can't fake. Strategies caught peeking at future candles are removed; weaker code smells only cost score.
Filter & sort the League by any metric — cap the drawdown, set a minimum profit, Calmar or win rate, and sort to surface the safest or best risk/reward strategies.
What's working — a page showing which indicators, concepts and timeframes actually score best across all our real backtests, not self-reported numbers.
Backtests that match live — each strategy's own risk protections (cooldowns, drawdown guards) run during the backtest, so the numbers reflect how it would really trade.
Printable reports — open any backtest as a clean report page and Save as PDF to keep or share.
Around-the-clock testing — Freqle quietly backtests in the background so the League keeps filling itself.
Candle-data dashboard — an admin view of which market data is downloaded, with on-demand download and refresh.
Testable strategies that need extra code — upload helper modules or Python packages so strategies with missing imports can still run.
Home highlights — recently tested, top rated, most popular and newest strategies at a glance.
Freqler community — accounts, ratings, comments, a support page and messages.
A way to support Freqle — a Donate page with free referral signups and sponsor options.
Free beta — right now, every function is free for everyone.

🔨 In progress

Strategy Forge (in testing) — describe an improvement in plain words and an AI proposes a change to a strategy, with a reviewable diff and a private copy you can backtest. Your original is never touched.
Backtesting the whole library so the Strategy League covers more and more strategies.
More testing power — running more backtests at once so results land faster.
Polishing the experience based on your feedback.

💡 Ideas & plans

Custom back- and forward-test settings — pick your own pairs and timeframes, not just the preset periods.
Choose which & how many pairs to test — per market (spot / futures), instead of a fixed pair set.
Compare strategies side by side — line up two or more and see the metrics head to head.
Regime-aware rankings — rank and filter strategies by how they did in bull runs, bear markets and sideways chop (the per-month regime tags are already there; ranking by them is next).
Email support — help and updates by email, alongside the on-site support page.
A better chat / discussion platform — a nicer place for the community to talk strategies.

🛠️ Built with

AI pair programming, in VS Code — Claude (Opus 4.8, Sonnet 5, Fable 5) and Kimi K3 do the implementation; a human does the planning, review and testing.
Infrastructure — Docker, PostgreSQL + pgvector, Ollama (local LLM), Caddy, hosted on GitHub.

🕒 Changelog

2026-07-26 — League ranking reworked to resist leverage tricks: per-trade edge now counts, a near-blowout no longer scores like a great year, confirmed lookahead leaves the board entirely, and the League says how many of its top 20 are probably luck. Bias findings are graded and marked on the code line itself.
2026-07-25 — DCA depth on strategy pages, ML / FreqAI strategies marked as not sandbox-testable, and a full security & performance audit.
2026-07-23 — Backtests run each strategy's own risk protections, a new "What's working" page, League refinable by any metric, printable reports, and early Strategy Forge.
2026-07-22 — Robust / Fragile / Unreliable verdicts backed by Monte Carlo, plus new quality metrics: return vs holding, expectancy, trade heat (MAE/MFE) and consistency.
2026-07-21 — Biased strategies pushed down the League, tags grouped into sections, entry→exit lines on trade charts, and a leaked-API-key scanner.
2026-07-20 — Every backtest month tagged bull / bear / sideways, an indicator-vs-results view, a Donate page, and an admin performance dashboard.
2026-07-19 — AI Search: plain English turned into filters by a model running on our own server. Plus period backtests and a candle-data panel.
2026-07-18 — The Strategy League went live, with a round-the-clock background sweep that keeps backtesting strategies on its own.
2026-07-17 — Interactive equity / drawdown / per-trade charts, monthly breakdowns, dark mode, user profiles and "find similar strategies".
2026-07-16 — Freqle is born: search and auto-tag thousands of freqtrade strategies, sandboxed tests, community ratings and comments, and upload / reverse-search of your own .py.

💬 Got an idea or found a bug? Send us a suggestion or bug report →