Research
Every scanner on trial, failures included.
Every WarRoom scanner carries a research status. Some survived a historical test against a control group. Some describe the market without predicting it. Some did not work as signals, and we say so. This page lists them all and explains how they are tested.
Every scanner and its status
| Scanner | Status | What the research says |
|---|---|---|
| Stealth Accumulation / Distribution | Experimental | The Scout view is currently unavailable pending a verified production feed and first snapshot. Distribution held up on held-out BTC rally data; Accumulation fell short. |
| Trap | Historically supported | Tracks a range’s trap-and-reverse sequence. Much of the move happens before the final breakout; the holdout was encouraging but inconclusive. |
| Leverage Expansion (daily) | Historically supported | A one-day open-interest build plus a volatility transition was followed by larger price expansion than matched states. Direction is unresolved. |
| Leverage Expansion (hourly) | Descriptive | Exceptional hourly open-interest moves mark higher-volatility states. Not a direction signal. |
| Breakout | Descriptive | An event to look at: the 1m, 5m, 1h and 4h break their range together. A fixed test rule gained where it was found (2024–26) and lost on held-back 2022–23 data, so it did not pass its confirmation test. |
| Fair Value Gaps | Descriptive | A first touch of a fair value gap did not beat comparison periods over 3 years and 512 coins. Useful as a map, not an entry. |
| Ignition | Historically supported | Signals on 1h and 15m reached a 5-ATR move 34–35% of the time on unseen data, against 26% for ordinary big candles. Not a direction signal. |
| Whale Spotting | Research in progress | A neutral observation of large flow meeting a wall. No forward outcome study of the current version yet. |
| Saturation & Void | Descriptive | A relative positioning-state monitor, not a directional signal. |
| Imbalance | Experimental | Live order-book depth against open interest on Hyperliquid. Not backtested; it describes the book now, not direction. |
Research records
Stealth Accumulation & Distribution: Research RecordThe research rules and held-out evidence for WarRoom’s experimental Stealth lists. The Scout view is currently unavailable pending a verified production feed and first snapshot.Trap Scanner: Breakout Trap Backtest ResultsHow WarRoom’s Trap scanner (formerly Trapdoor) tracks failed breakouts that reverse across a range, and what two years of research on 95 coins showed.Open Interest Build-Ups: Volatility ResearchWhat WarRoom’s Leverage Expansion research found about open-interest build-ups, volatility expansion, and why the direction of the move stays unresolved.Breakout: Multi-Timeframe Breakout ResearchWhat happened after a coin broke out on the 1-minute, 5-minute, 1-hour and 4-hour charts at once: WarRoom’s test, its confirmation run on older data, and why Breakout makes no outcome claim.Do Fair Value Gaps Work? 3 Years of DataWarRoom tested fair value gaps on 512 coins over 3 years with a sealed holdout. The first touch did not beat comparison periods. Here are the numbers and limits.Ignition: Big-Move Odds ResearchHow WarRoom rebuilt Ignition, tested it on data it had never seen, and what the odds of a big move really are on 1-hour and 15-minute charts.
What each status means
- Historically supported
- The preset’s definition survived a recorded historical evaluation against a comparison group. It is still monitored live, and its stated limits still apply.
- Descriptive
- The preset describes a market state or event. It is shown for context and must not be read as a directional or predictive signal.
- Experimental
- A research candidate that can be observed in the product, without enough evidence for a validated claim.
- Research in progress
- The definition is frozen, but no adequate outcome study of the current version has been published, so no performance claim is made.
How WarRoom tests a scanner
- Freeze the definition first. Results describe the version that ships, not a tuned variant.
- Compare against a control group or matched comparison states from the same coins and timeframe, so a result is measured against the ordinary market, not against zero.
- Hold data back. Where the study allows, the most recent period is sealed and checked once, after the rules are fixed.
- Report medians, intervals and sample sizes, not just averages or win rates.
- Publish what failed. Negative and null results stay on the record and change the product: the experimental Stealth view stays unavailable until a verified production feed is ready, Fair Value Gaps is labelled descriptive, and Ignition shows no direction because every direction idea we tested fell short on unseen data.
- Keep the recipe private, publish the evidence. Features, weights and thresholds stay confidential; samples, controls, results and limits are public. Each record says whether fees were included.
Research results are historical distributions under stated rules. They are not probabilities of profit, trade instructions or financial advice.