Methodology

How these reports are counted

Version 2026-09-21

Market terms

  • Cryptocurrency trading signal — a dated long or short on a USDT spot pair (Bitcoin BTC, Ethereum ETH, or an altcoin).
  • Take-profit (TP1–TP5) — staged profit targets. A target hit counts only after the signal is decided.
  • Stop-loss (SL) — risk exit. A stop is a loss, not a timeout.
  • Hit rate / TP vs SL — wins among decided signals. Open or no-result (NR) rows are not wins.
  • AI Assistant — in-app suggestions. Modes include default mix, technical analysis, social sentiment, and AI/on-chain data.
  • Telegram Official Signals — published channel alerts, often a rolling last-40 window on t.me/cryptocrispy_ai.
  • Core vs tactical — investor-book split: persistent hold versus a smaller rotating sleeve.
  • ROI — return after pairing BUY then SELL on the same symbol, fees subtracted, on trading pages.
  • Risk appetite — separate risk modes; the headline is sample-weighted, not the best bucket.

Counting rules

  • Each prediction is evaluated once at a declared horizon.
  • Risk modes are reported separately. The headline is sample-weighted, not the best mode.
  • Training or validation accuracy is shown only when a labelled holdout set exists. Otherwise it is “Not available”.
  • Trades are paired BUY then SELL on the same symbol. Fees are subtracted from ROI.
  • Reports that fail sample or quality gates stay in Historical with a label. Poor performance is not a deletion rule.
  • Telegram Signals hit rate is (full take-profit + any target hit) / closed signals. Open rows are excluded. Each Official Signals recap is a rolling last-40 window; overlapping days are not independent samples. The 20% plan figure is the channel recap, not platform trade ROI. Source: t.me/cryptocrispy_ai.

Publication gate

A current Insights, AI Assistant, Telegram Signals, or Trading report needs at least 30 scored observations, valid timestamps, and required fields. Historical pages can be smaller; they are labelled.

Limitations

  • Overlapping windows are labelled; they are not independent samples.
  • Dataset B (quality score 0) is excluded from trading headlines.
  • Fees, pairing rules, and the 15-minute evaluation horizon are disclosed on each generated page.