meli / What works on MELI
Historical hit rates for every rank-panel rule on MELI, measured on daily bars (2024-08-29 → 2026-08-28, 501 bars, 335 firings). Not a buy or sell list.
What works on MELI
| Rule | Win 5d | Win 1mo | Win 3mo | Count |
|---|---|---|---|---|
| SMA 50/200 death crossSELL | 100.0% | 100.0% | 100.0% | 1 |
| SMA 50/200 golden crossBUY | 100.0% | 100.0% | 100.0% | 1 |
| MFI cross 80SELL | 40.0% | 60.0% | 80.0% | 5 |
| BB upper band rejectSELL | 57.1% | 66.7% | 63.6% | 15 |
| ADX -DI/+DI cross (strong)SELL | 50.0% | 100.0% | 50.0% | 2 |
| Williams %R cross -20SELL | 64.9% | 54.5% | 60.0% | 38 |
| CCI cross 100SELL | 61.5% | 60.9% | 52.4% | 27 |
| MACD bearish crossSELL | 46.2% | 48.0% | 66.7% | 26 |
| EMA 9/21 death crossSELL | 37.5% | 53.3% | 66.7% | 16 |
| BB lower band bounceBUY | 71.4% | 71.4% | 35.7% | 14 |
| ADX +DI/-DI cross (strong)BUY | 100.0% | 50.0% | 50.0% | 2 |
| Stoch %K/%D cross (overbought)SELL | 59.3% | 56.0% | 50.0% | 28 |
| RSI(14) cross 70SELL | 33.3% | 83.3% | 40.0% | 6 |
| RSI(14) cross 30BUY | 83.3% | 50.0% | 33.3% | 6 |
| Stoch %K/%D cross (oversold)BUY | 57.7% | 46.2% | 34.6% | 26 |
| MFI cross 20BUY | 60.0% | 40.0% | 40.0% | 5 |
| Williams %R cross -80BUY | 65.8% | 40.5% | 36.1% | 38 |
| CCI cross -100BUY | 64.0% | 37.5% | 34.8% | 25 |
| MACD bullish crossBUY | 48.1% | 44.0% | 27.3% | 27 |
| EMA 9/21 golden crossBUY | 58.8% | 31.2% | 26.7% | 17 |
| Squeeze fire (bearish)SELL | 33.3% | 33.3% | 33.3% | 3 |
| Squeeze fire (bullish)BUY | 16.7% | 50.0% | 20.0% | 7 |
Interactive chart with the same engine: /?symbol=MELI. Fresh triggers across mega-caps: /firing. Rule write-ups: SMA 50/200 death cross · SMA 50/200 golden cross · MFI cross 80.
Why it moved
TA on this site tells you which classic rule printed and whether it historically paid. It does not say why the stock moved. Options positioning is often why. SpotGamma maps that.
Optional: SpotGamma free tools. Nothing here is trade advice.
Honest caveats
- Past ≠ future. Everything here is a historical measurement on daily bars, not a prediction and not investment advice.
- No costs modeled. Forward returns are close-to-close with no commissions, slippage or borrow.
- Overlap bias. Signals that fire in clusters get credit for the same move more than once.
- Small samples. A row with fewer than 10 firings is anecdote, not evidence.
Full method: help manual. Interactive chart: terminal.