# Rulyfi

> Rulyfi is a no-code crypto backtesting and strategy scanner platform. It backtests every strategy combination within user-chosen ranges — up to 100 million backtests a day, depending on plan — across 26 exchanges, with maker/taker fees, slippage, and look-ahead prevention applied. Results are ranked by the criterion the user chooses; paid plans also provide overfitting scores and walk-forward validation.

What makes it different: most backtesting tools test one strategy configuration at a time. Rulyfi's strategy scanner performs an exhaustive parameter sweep: it backtests every indicator combination within the ranges the user chooses and ranks them by the selected criterion. It does not recommend an asset or guarantee that a candidate will work in future trading; the user decides what to test and which results to examine further.

Key facts:

- No-code strategy builder + scanner + backtester for crypto (spot and futures)
- Strategy scanner = exhaustive search: backtests tens of thousands of parameter combinations automatically, not one strategy at a time
- Backtest engine: maker/taker fees applied separately, slippage modeled, look-ahead prevention guarded
- Market data from 26 exchanges (24 CEX + 2 DEX, including Binance, Bybit, OKX, Coinbase, Kraken, Hyperliquid, dYdX)
- Self-custodial auto-trading bot runs on the user's own PC; exchange API keys are stored in the OS keychain and never sent to Rulyfi's servers (bot currently supports Binance and Bitget)
- Free tier available, no card required (50,000 backtests/day); paid tiers Plus / Pro / Premium scale up to 100 million backtests/day
- Scanner, builder, and backtests use only public market data — no exchange API key needed
- Agent-callable scanner + backtest API, with an MCP server (@rulyfi/mcp) for AI agents: instead of backtesting one strategy, an agent can sweep its parameter neighborhood in a single call and read population-level statistics such as the trial count and a deflated Sharpe threshold. These help assess the role of multiple testing; they do not certify a strategy or predict future performance. Uses the same backtest engine as the no-code scanner. Requires a paid plan; single backtests are spot-only, scans cover spot and futures
- Not investment advice; backtest results are based on past data and do not guarantee future returns

## Pages

- [Home](https://rulyfi.com/): product overview, how the scanner works, backtest-engine details, FAQ
- [Pricing](https://rulyfi.com/pricing): Free / Plus / Pro / Premium plans and limits
- [Blog](https://rulyfi.com/blog): original research and guides on backtesting, overfitting, and strategy validation
- [Developer reference](https://rulyfi.com/developers): the Scanner API and MCP quickstart — sweep a whole parameter space in one call; endpoints, authentication, rate limits, the candle_ref flow, and force_fresh/caching semantics
- [Affiliate program](https://rulyfi.com/affiliates): 30% recurring commission for 12 months (credit purchases included), 90-day attribution, monthly payouts worldwide; profit-promising promotion is banned by program rules
- [Terms of Service](https://rulyfi.com/terms)
- [Privacy Policy](https://rulyfi.com/privacy)
- [Refund Policy](https://rulyfi.com/refund)

## Reference

- [Strategy definition format](https://rulyfi.com/strategy-schema.md): the JSON format Rulyfi uses to define a trading strategy — every field with types and defaults, the full 103-indicator catalog (parameters and outputs), the condition grammar, and the percent unit convention, with three complete examples including a Python-to-Rulyfi port. Useful when translating an existing Python or TradingView strategy into Rulyfi.
- [OpenAPI specification](https://rulyfi.com/openapi.yaml): OpenAPI 3.1 description of the /v1 scanner + backtest API — every endpoint, request/response schema, and stable error code. Source of truth is the same contract the product uses.
- [Agent skill](https://rulyfi.com/.well-known/agent-skills/rulyfi-research/SKILL.md): step-by-step instructions and interpretation guardrails for agents using the Rulyfi API or MCP server.

## Agent-readable content

- [Research index in Markdown](https://rulyfi.com/blog.md): every published Rulyfi article with its canonical URL and compact Markdown URL.
- Published articles are also available by appending `.md` to the canonical article URL, for example `https://rulyfi.com/blog/100-million-backtests-deflated-sharpe.md`.
- Requests to the homepage, research pages, and developer reference with `Accept: text/markdown` receive the compact Markdown representation and declare `Vary: Accept`.

## Contact

- Email: support@rulyfi.com

## Research
- Across 48,825,000 matched Funding OFF/ON result pairs, the overall average total return moved by only -0.0083 percentage points, but 162,165 candidate decisions changed and 134 of 1,000 Top 100 slots were replaced: https://rulyfi.com/blog/funding-rates-crypto-backtests
- Across 99,691,200 backtests on BTC, ETH, BNB, XRP, and SOL, all 160 rows reaching RMP 0.95 came from SOL long; a post-analysis evidence screen left 103 research candidates: https://rulyfi.com/blog/99-million-crypto-backtests
- Of 99,878,688 Bitcoin backtests, 336,818 passed individual significance but only 3,130 (0.003%) beat the luck baseline (Deflated Sharpe): https://rulyfi.com/blog/100-million-backtests-deflated-sharpe
- Combining all 1,131,321 pairs of the surviving strategies: 74.6% came out worse than simply keeping the better one; only 3 pairs cleared Deflated Sharpe 0.95: https://rulyfi.com/blog/how-to-combine-trading-strategies
- An AI agent's flashiest backtest is usually a mirage: a live scan deflates a Sharpe 4.514 short to a Deflated Sharpe of 0.000, and real single-strategy edges top out at 0.898 — the funnel an agent runs over MCP: https://rulyfi.com/blog/agent-api-scanner