{"ok":true,"nota":{"id":"70f106d06dfaf4a5","slug":"un-filtro-de-confirmaciones-no-cambio-ni-una-ganancia-en-las-operacion-70f106","titulo":"Un filtro de confirmaciones no cambió ni una ganancia en las operaciones de prueba","resumen":"Un grupo evaluó si exigir al menos cinco confirmaciones mejoraba un sistema de operaciones financieras automáticas. En el período analizado, el filtro no habría cambiado ningún resultado cerrado.","fuente":"zenodo","fuenteNombre":"Zenodo · Investigación abierta","color":"#1682d4","autor":"Caldin, Andrew Stewart","url":"https://doi.org/10.5281/zenodo.23490699","imagen":"/hoy/portada/70f106d06dfaf4a5","tema":"economia","subtemas":["trading algorítmico","backtesting","mercado de divisas","gestión de riesgo","rendimiento de inversiones"],"entidades":["E8 Intelligence Research"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-20T05:23:19.000Z","indexTs":"2026-10-22T21:54:18.823Z","hash":"fc038d36a59dd589d7f1ba5db0f948e5e84eb83b7e62ce810783f732c7b18015","tsa":{"time":"Oct 22 21:56:07 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"motor","registro":"hallazgo","porQueImporta":"Las mejoras que aparecen en simulaciones históricas no siempre se confirman con operaciones reales, y el propio trabajo lo reconoce al mostrar que la ganancia prometida no apareció en el período observado. Para alguien que sigue inversiones por su cuenta, es una señal de que una cifra de rendimiento ajustada con datos pasados no garantiza nada. Cuidado con quien vende un método como infalible. Los autores son independientes y la muestra es muy chica, con apenas ocho operaciones cerradas.","quienFirma":"E8 Intelligence Research (investigación independiente)","confianza":"preprint","consulta":{"pregunta":"¿Confiarías en un sistema de inversión con resultados solo simulados?"}},"sello":{"hash":"fc038d36a59dd589d7f1ba5db0f948e5e84eb83b7e62ce810783f732c7b18015","sello":"95ff50370203166d8298a04e83ef842c0bd8cb265510741b5bb94ca861f6bcf5","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6575838,"aleatorio":"0a3876e39172152042339137ab5e3b393dcc0b9424256c5af20c090e9045fc73","tiempo":"2026-10-22T21:56:00.000Z"},"nist":{"cadena":2,"pulso":1995118,"salida":"F0E0C41C949133434038AD3665131BA50B8741AD6763A98DA9C9F98A82572C39F0270A26E31EB0E30B2F4719DE6ED82AA401EE3E9EB6515A5780BD147A7BE091","tiempo":"2026-10-22T21:55:00.000Z"},"noAntes":"2026-10-22T21:56:00.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 22 21:56:07 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"LAB #4425 NEUTRAL: LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.6% to 33.1% (+1.4pp, — E8 Intelligence Research","descripcion":"<p>IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: Min confirmations >= 5 — improves win rate from 31.6% to 33.1% (+1.4pp, n=1903) SAME-WINDOW EFFECT: Applying a minimum of 5 confirmations would have filtered out the two REJECTED/open trades (EURJPY SHORT and AUDUSD LONG) that entered with fewer confirmations, but those are not yet resolved, so their exclusion is ambiguous. Among the eight CLOSED trades, all appear to have met the 5-confirmation threshold based on their signal sources (rizzy5m/7m and wkjudas), so none of the closed wins or losses would have been removed. Therefore, over this specific window, the filter would have changed nothing in realized PnL — the +1.4pp backtest gain does not manifest here, and the live win rate remains 5/8 (62.5%) regardless. The only effect is a reduction in open exposure, which is neutral to slightly positive ","pubDate":"2026-10-20T05:23:19Z","url":"https://doi.org/10.5281/zenodo.23490699"},"indexado":"2026-10-22T21:54:18.823Z"}}