# AI Learning Benchmark 2026

Public-safe benchmark summary for NoahAI learning operations.

## Core summary

- Data scale: 140,000+ signals
- Trend: stepwise improvement over time
- Risk control: dynamic downshift in weak performance windows
- Exit policy: early-exit contributes to downside defense in volatile sessions

## Operational loop

- The public operating loop is based on a user-level 7-day window.
- Day 1: establish the baseline for win rate, RR, average win, average loss, and max drawdown.
- Day 2 to Day 6: make one small change at a time and observe the result.
- Day 7: decide to keep, pause, or roll back based on RR and drawdown.

## Asset scope

- The optimization concept is not blockchain-only.
- The same button can be used across crypto, stocks, and ETFs.
- Each asset class maps to different adjustable parameters.

## Responsibility boundary

- NoahAI supports decision quality, not profit guarantees.
- Final judgment remains with the user.
- Public pages show aggregated, anonymized indicators only.

Disclaimer: This content is an aggregated technical summary and does not guarantee returns.