Ideas and results
Historical paper ideas and their results. No funded portfolio return is available. These results do not simulate a $1,000 account with 3× leverage.
Mainnet experiment · $1,000 · 3×
Variable sizing with total positions capped at 3× current equity, up to $3,000 initially. Fees and slippage included; funding excluded.
No learned exit policy passed the evidence check. New entries remain disabled.
Tested 1, 4, 24 and 48-hour policies with stop and target distances learned from earlier price paths. This study is separate from the historical idea sums and trial 3. Study results and settings
Weekly idea sums
Sum of closed ideas’ returns in percentage points (pp), after estimated fees and slippage, excluding funding. Position sizes, leverage, shared capital and open P&L are not applied. These sums are not portfolio returns or trial 3’s score.
Ideas
Newest first. Each idea closes at its target, at its stop or after 24 hours, and is graded after 0.2% round-trip costs.
| Time (UTC) | Coin | Call | Entry | Stop | Target | Result | Net |
|---|---|---|---|---|---|---|---|
| Loading ideas… | |||||||
How to read this
The model takes a side on a flagged coin, LONG or SHORT at a size %, with an entry, a stop and a target. It closes at whichever level is touched first, or after 24 hours.
Unusual volume and positioning, but the model takes no side, and the reason is shown. There is no trade, so nothing to grade.
Sent when an idea closes: its net result after 0.2% round-trip costs, plus the week's tally. This page shows the same numbers.
- The goal is ideas that make money every week. Trial 3 (19 Oct 2026 to 11 Jan 2027) decides whether the model is real. Its scorecard is computed once, at the end, so this page shows no interim trial returns.
- A screening tool, never a trade trigger. A flag earns a manual look at the chart, nothing more.
- Unproven. Every earlier directional strategy tested here looked good at first and failed a fair test.
- Hindsight. The thesis comes from one case picked after the fact, and most flags are noise.
- Legal risk. Acting on tip-shaped information can carry legal risk.
Research archiveThe old follow-the-crowd scoring and every raw detection. Loads when opened.
Follow-the-crowd baseline · last 30 days
This is not what the ideas do. It trades every flag in the crowd's direction (long after new longs, short after new shorts), holds it 1h, 4h or 24h at equal size, and subtracts costs.
All detections
The last 100 flags, whether or not they became an idea. Nothing is graded here; results live in the ideas table.
| Time (UTC) | Coin | Crowd read | Model | Z | ×med | Px 1h | OI Δ | Scope |
|---|