# L-ETF

> L-ETF is a free, open-source research tool for backtesting leveraged ETF (UPRO, TQQQ, SSO, QLD) SMA (simple moving average) timing strategies over long market history, and for tracking today's live SPX/NDX SMA signals.

L-ETF simulates leveraged ETFs back beyond their real launch dates (S&P 500 data since the 1800s, Nasdaq 100 since the 1980s) by modeling daily leverage reset, expense ratios, and borrowing costs, then calibrates that model against real UPRO/TQQQ/SSO/QLD prices. The core strategy idea: hold the leveraged ETF while price is above its N-day moving average, switch to a defensive "risk-off" asset (e.g. SGOV, VGSH, GLDM, BRK.B) when price falls below it, with a percentage buffer around the line to reduce whipsaw trades.

This is **not investment advice**. It is an educational/research tool; past backtested performance does not predict future results, and leveraged ETFs are high-risk instruments.

## Pages

- [Home](https://l-etf.com/): Overview and links to every tool.
- [Signals](https://l-etf.com/signals): Today's live SPX and NDX SMA signal (in/out of the leveraged ETF), with configurable SMA period and buffer, and optional push notifications on signal flips.
- [Backtesting Tool](https://l-etf.com/backtesting-tool): Single-period backtest with trade log, value chart, drawdowns, and real-vs-simulated ETF comparison.
- [Compare Strategies](https://l-etf.com/compare-letfs): Compare leveraged ETF strategies across many rolling historical windows with percentile return distributions.
- [Compare SMA Periods](https://l-etf.com/compare-sma-strategies): Sweep moving-average lengths and compare faster vs. slower trend-following rules.
- [Compare SMA Buffers](https://l-etf.com/compare-threshold-strategies): Sweep buffer thresholds around the moving average to reduce false switches.
- [Compare Risk-Off Assets](https://l-etf.com/compare-riskoff-assets): Compare defensive assets (SGOV, VGSH, GLDM, BRK.B, VOO, QQQ, and equal-weight mixes) held while the strategy is out of the leveraged ETF.
- [Holding Period Statistics](https://l-etf.com/statistical-analysis): How win rates and returns change as the holding period changes.
- [Futures Tool](https://l-etf.com/futures-tool): SMA strategy simulation using index futures instead of ETFs, with margin scenarios (Normal/Stress/Crisis/Extreme) and futures-specific fees.
- [FAQ](https://l-etf.com/faq): Methodology — what leveraged ETFs are, how the SMA rule works, how borrowing-cost calibration works, trading-cost assumptions, and the full daily-return formula used in the simulation.

## MCP server (for AI agents)

L-ETF exposes its simulation engine over the Model Context Protocol so agents can run real backtests instead of estimating figures. Add it to any MCP client (Claude Desktop/Code, Cursor) or inspect with `npx @modelcontextprotocol/inspector`.

- Endpoint: https://l-etf.com/mcp (MCP over Streamable HTTP)
- Machine-readable descriptor: https://l-etf.com/.well-known/mcp.json
- Tools: `list_presets`, `get_market_data`, `get_sma_signals`, `get_sma_signal_history` (crossover log, current regime, whipsaw stats), `get_sma_calibration`, `run_backtest`, `compare_backtests` (simulated and/or real ETFs, either index), `run_rolling_window_analysis` (optional percentiles and per-window rows), `run_holding_period_analysis`, `compare_strategies` (SMA on/off, risk-off assets, SMA periods, SMA buffers, 2-D asymmetric upper/lower buffer grid), `compare_letfs` (percentile distributions), `run_futures_backtest` (index futures), `get_box_spread_apy`, `get_precomputed_analysis` (canonical precomputed runs, no simulation cost)
- Resources: `letf://methodology`, `letf://data-coverage` (data freshness)
- Prompt: `analyze_strategy`
- Backtest and sweep tools take `smaExecutionMode` (`trigger-day-close` | `next-day-close` | `next-day-open`, default `next-day-open`) so a result can be re-checked under a different fill assumption.
- The long-running sweep tools emit MCP progress notifications when the request carries a `progressToken`.

Every tool returns the not-investment-advice disclaimer — preserve it in any summary. The endpoint is rate-limited per client IP (stricter for the rolling-window sweep tools).

## Data & source

- Underlying market data: S&P 500 and Nasdaq 100 total-return history, real leveraged-ETF prices (UPRO/TQQQ/SSO/QLD), interest-rate series, and inflation (CPI) data, refreshed on a schedule.
- Source code: https://github.com/ravelab/l-etf