Signal results

• A signal counts in a period by the time it was published, not when it closed — results of the latest signals are still coming in.
• Numbers are shown only from 20 results; below that the sample is too small to mean anything. For exchange events the floor is 5, the same as on the listings radar and research pages.
• “In profit” is directional for ideas, presets and ignition (a short wins when the price falls). For exchange events it is the share of coins trading higher 24 h later — a description of what happened, not a trade.
• “Median” is the typical result of one signal: half did better, half did worse.
• “Sum R” is the ideas' total result in risk units: 1R is the loss had the stop been hit, so +2R means two such risks were made in total.
• Baseline: for preset signals — BTC buy & hold over the same days (average); for ignition — the median move of the other coins over the same window, signed the same way. Ideas and exchange events have no baseline here. Preset and ignition moves are before fees; idea results are net of fees, slippage and funding.
• Forward only: a signal is recorded on the day it fires, from the data available at that moment. Nothing is backfilled or re-scored.
• No cherry-picking: every built-in preset is tracked. Each day the journal records up to ten of its largest matches by market cap whose trade setup is an idea, not watch (it agrees with the coin's overall state, R:R is at least 1.5 and the market is liquid enough), and never the same coin both long and short. Since the restart on 19 Oct 2026 rows are never edited or deleted. In practice most presets record nothing on most days — the gates are strict — so expect a handful of signals per day in total, often on the same few coins (about 1–3 independent bets a day).
• Entry is the close of the last completed daily candle (00:00 UTC); the scan runs in the first hours of the UTC day. Exits are the daily closes 1, 7 and 30 days later.
• Universe: the 100 most-traded coins with a USDT perpetual on Binance or Bybit (ranked by 24h turnover), on the daily timeframe, excluding stablecoins and tokenized assets. It is the same scan as the public preset pages.
• Two scores per signal. Close-to-close returns at 1, 7 and 30 days are gross. The setup result follows the recorded stop and target 1 on hourly bars from the moment the levels were recorded to the end of the signal day, then on the daily bars after it: target first, stop first (a bar touching both counts as the stop; a gap past the stop exits at the open), or closed after 30 days. It is net of 0.3% round-trip fees and slippage plus funding for the days held.
• Numbers are shown only once a horizon has at least 20 settled signals. Below that the sample is too small to mean anything.
• Benchmarks: BTC buy & hold, and an equal-weight basket of that day's 20 largest altcoins (from the same universe), over the same days.

Every number on this page counts only method-v2 rows (recorded setup levels, one direction per coin per day, scanned on the liquid-perps universe), which start on 2026-10-01.

The screener's preset cards also show an in-sample historical replay next to random entry. It is not evidence of an edge; this forward journal is the real test.