CHARTROOMS

The evidence

What we tested — and what failed

The rules this course teaches were measured against real historical tape wherever a measurement could be made. 133 of those measurements are below, 16 of which refused to confirm something the internet repeats confidently. The failures stayed in the course, next to the evidence that killed them.

Measured on the house runner census — 5,925 runner-days, 2021–2026 — unless a finding says otherwise. How we know explains the method, and what it cannot tell you.

Things that did not survive measurement

Popular, widely taught, and not supported by our data.

Refuted: “Gap size is a rule you can trade by”

From our own backtests: gap size behaves like a dial, not a rule. Sub-20% gappers went green from the open 78% of the time; ≥200% gappers went green only 23% — and of those monsters, the share that filled the gap the same day was zero. Worse for the "short the open" instinct: across every gap bucket, including the ones that faded all day, price popped past the open print 68–84% of the time first.

Room 00 · Chart Literacy — Gaps

Refuted: “Round numbers are walls”

From our own backtests (US stocks $0.50–$20, 2023–2026, about 5 million first touches): price turned back 1% from the first touch of a whole dollar 21% of the time, against 14.5% for an arbitrary price on the same charts, and it held out of sample. At $2–5 it was 43.9% against 30.9%. A minute bar's high landed exactly on the dollar 25.9% of the time (18.5% at an arbitrary price) and one cent through it 7.4% (11.2%).

Room 01 · Levels — Whole numbers

Refuted: “Stacked, rising EMAs are bullish on a runner”

From our own backtests. On extended small-cap runners, the classic bullish EMA tells all pointed the wrong way: • Stacked EMAs (9 over 50) → weaker forward returns, not stronger. • A steep, rising EMA → weaker still. • The test-and-hold "bounce" → worse than losing the line outright. • A touch of the 1-minute 50 EMA marked a top two times out of three.

Room 02 · Trend — The 9 & 20 EMA as dynamic levels

Refuted: “Short every double top”

From our own backtests — the honest version: on runner days, a mechanical "short every double top" looked great without a stop (64% resolved down)… and lost money the moment we costed it with a real stop — 54% of trades got stopped out by the breakout tail before the fade paid. The read is real; the autopilot trade is not.

Room 03 · Reversals — Double top

Refuted: “Buy the flush for a full reclaim”

From our own backtests: after a parabolic flush, the dead-cat bounce is nearly universal — but the prior peak is almost never reclaimed. On the backside, the double bottom marks a BOUNCE you can rent, not a reversal you can own; the sellers stacked overhead usually get their turn.

Room 03 · Reversals — Double bottom

Refuted: “Flag logic works on already-extended stocks”

From our own backtests — a caution the textbooks skip: flag logic has a range. On stocks already up 50–100%, coils leaned slightly upward, as taught. But on stocks extended +400% or more, the same tight flag broke DOWN twice as often as up — at that altitude, a "high tight flag" is usually distribution wearing a bullish costume.

Room 04 · Breakouts — The bull flag

Refuted: “It'll come back”

From our own backtests — why "it'll come back" is the most expensive sentence in trading: measuring thousands of underwater short entries, 92% were green at some point — which is precisely why people learn to hold losers. But past ~30% of adverse move, the median trade never saw green again. The comeback is real just often enough to teach the habit that eventually ends the account.

Room 05 · Risk Definition — Max loss per trade

Refuted: “The opening-range breakout is a free long”

From our own backtests — the opening-range breakout is not a free long on runners. We measured 64% of runner days breaking the OR high — but only about half of those held it, and just 32% closed above it. Runners broke the OR low more often (69%). The break fires constantly; it resolves down more than up. Use the range as a map of who is winning, not as a trigger by itself.

Room 06 · Timing — The opening range · paid room

Refuted: “Thin names hold up better”

From our own backtests — liquidity is the single most powerful segmentation we have found, and it points the opposite way to intuition. Runners trading only $1–3M a day deep-faded 77% of the time; the $100M+ monsters only 26%, and those kept running and squeezed shorts. Thin is the fade side; heavy is the trend side.

Room 07 · The Tape — Liquidity · paid room

Refuted: “A low float means it won't fade”

From our own backtests: (1) Low float does not mean "won't fade." We segmented thousands of runner-days and the tiniest floats actually faded MORE, not less. The hold-versus-fade axis was dollar volume, not float. (2) The reported numbers are frequently wrong. In our own database, 34.6% of runner-days reported a float larger than shares outstanding — an impossibility.

Room 08 · Structure — Float vs. shares outstanding · paid room

Refuted: “Short the ignition candle”

From our own backtests — three numbers that ARE this lesson: (1) Shorting the ignition candle had a profit factor of 0.11 with median heat of 75% — and the vertical first candle (the FCUV type) was the worst in the study: 97% stopped out. Waiting improved everything monotonically, with the elbow around 15–20 minutes — the "don't touch the first 15 minutes" rule is measured, not folklore.

Room 11 · Multiday — Gap-and-crap and the day that refuses to crap · paid room

Refuted: “Float and short interest predict the squeeze”

From our own squeeze-tail research: the names that produced the violent short-killing moves were dominated by dollar volume, not float or short interest. Runners trading $100M+ produced mega-moves (≥300%) about 23% of the time versus roughly 3% for the thin ones, and were far more likely to close at their highs.

Room 12 · Short Side — Squeeze risk · paid room

Refuted: “Having news means it will hold”

The honest caveat from our own data: we tested whether having news at all separates the runners that hold from the ones that collapse, and it did not: in our 5,925-day runner census, days with news deep-faded about 79% of the time versus 76% without, against a base of about 77%. News is a non-signal for direction. What it buys you is an explanation of the crowd, and a prior about how long the crowd stays.

Room 13 · Catalysts — PR quality tiers · paid room

Refuted: “Coils resolve in a predictable direction”

From our own backtests — we measured tight intraday coils (≤10% range) on runners: about half resolved into chop, roughly a quarter broke up, a fifth broke down. The coil reliably stores energy; it does not reliably pick a direction. Trade the resolution, never the prediction — and remember the extension flip from Unit 4: the same coil above a doubled stock leaned into breakdown, not breakout.

Room 16 · The Pattern Dictionary — Triangles and wedges · paid room

Refuted: “Short every triple top”

From our own backtests — the warning transfers intact from the pattern's smaller sibling: mechanically shorting every double top looked great uncosted (64% resolved down) and lost money with a real stop — 54% were stopped by the breakout tail first. Adding a third top does not repeal that arithmetic. The read is real; the autopilot trade is not.

Room 16 · The Pattern Dictionary — Triple tops, triple bottoms, and the range · paid room

Refuted: “The 9 EMA tells you something price doesn't”

From our own backtests — the three measured facts that shaped this course's indicator minimalism: (1) the 9 EMA is price restated — correlation 0.967 with price itself; it adds nothing a clean chart doesn't show. (2) The 200 EMA is dead intraday on runners — correlation +0.017 with forward returns, pure decoration.

Room 16 · The Pattern Dictionary — The indicator aisle — why this course has no RSI · paid room

Things we measured and kept

What the same tests found worth teaching.

From our own backtests (5,925 runner-days, 2021–2026): dollar volume was the sharpest lever in the entire study — and it points the opposite way from intuition. Thin $1–3M names deep-faded 77% of the time; ≥$100M monsters only 26%, ran further, and squeezed shorts (forced them to buy back at rising prices — Unit 12's whole subject).

Room 00 · Chart Literacy — Volume

From our own census (5,925 runner-days): runner tops cluster at three moments — the premarket ramp, the 09:30 open, and the 16:00 print — and 37% of day-highs are already in by the open bell. The fade's low lands late: median 16:35, which is after the bell, with 70% of bottoms after 3pm. The typical runner day has a shape: top early, bleed all day, bottom at the close.

Room 00 · Chart Literacy — Pre & post-market sessions

From our own backtests (US stocks $0.50–$20, 2023–2026, about 5 million first touches): the prior-day high turned price back 2 to 3 points less often than a random nearby price, in sample and out. On $1–5 stocks it did 1 to 2 points better than random, which is still a small effect.

Room 01 · Levels — Prior day high & low

From our own backtests (US stocks $0.50–$20, 2023–2026, about 5 million first touches): the premarket high turned price back 1 to 3 points less often than a random nearby price, in sample and out. On $1–5 stocks it did 2 to 3 points better than random, still a small effect.

Room 01 · Levels — The premarket high

From our own backtests: when a high was going to be exceeded, it happened within 15 minutes 82% of the time. The day's final high is another matter. A high that had stood unbroken for 90 minutes was still only the real HOD 46% of the time, and a quarter of final HODs printed in the last 30 minutes. "This leg is over" is knowable; "the top is in" mostly is not.

Room 01 · Levels — HOD & LOD

From our own atlas (43,738 runner-days — the full map): a runner spends only ~27% of its time above VWAP but transacts ~51% of its money there; the median runner crosses VWAP 19 times a day (a bare touch or cross is noise — every reliable read needs the state to hold); 98% touch VWAP at some point, median within one minute of leaving it; and ~65% close below it.

Room 02 · Trend — VWAP — the day's average price

From our own backtests: exhaustion is the one reversal condition that survived every honesty test we threw at it. Shorting climax breaks gated on extension — price stretched ≥1.8–2× above VWAP, the parabolic blow-off case — won 75–80% in every single year we tested (2023–2026). The same entry without the extension gate was roughly breakeven.

Room 03 · Reversals — The exhaustion wick

From our own backtests — the same pause means opposite things depending on extension. A stall on a stock up less than 100% was consolidation: it resolved into a hold or grind 76% of the time (and a stall while holding above VWAP was the single strongest "don't short this" signal we measured). The identical stall on a stock up 100%+ flipped to exhaustion — nearly half fast-faded from there.

Room 04 · Breakouts — Consolidation length

From our own runner-day data: 64% of runners broke their opening-range high, only about half of those held it, and 32% closed above it. More runners broke the OR low than held the high.

Room 04 · Breakouts — The false breakout

From our own backtests: 37% of runner day-highs are already in by the open bell — the top is frequently printed before most participants have finished their coffee. But the clock is a filter, not a trigger: even after waiting 90 minutes, a blanket short was still a loser. Time removes the worst entries; a setup still has to do the work.

Room 06 · Timing — The 9:45 fade · paid room

From our own backtests — this shows up as a hard number in our best short signal: the same setup taken between 10:30 and 14:00 ran a profit factor of about 0.57–0.61, versus about 2.31 near the open. Skipping midday was one of only two changes that turned a marginal edge into a robust one. The tape does not owe you a trade every hour.

Room 06 · Timing — Midday chop · paid room

From our own census (5,925 runner-days): the fade's LOW lands late — median 16:35, with 70% of bottoms printing after 15:00. Note what 16:35 means: the median low is after the bell, in after-hours tape that carries about 1% of the day's dollars. So power hour is where the bleed is still running, and the final low often prints when almost nobody can trade it. We haven't yet recomputed the low on regular hours alone.

Room 06 · Timing — Power hour · paid room

From our own backtests — the same principle appears on the short side, inverted: the best short entries were failed bounces into resistance, not chases of the breakdown. Across every study we ran, entering after the move had already travelled was the worse half of the trade.

Room 06 · Timing — First pullback vs. chasing extension · paid room

From our own backtests — we raced the candidates for predicting a stock's forward range: dollar volume (correlation 0.45) beat raw share count (0.43), and both crushed RVOL (0.28). Normalizing by float or by average volume actively hurt prediction. The finding is counterintuitive and consistent: absolute size wins. Use RVOL to notice that something is happening; use dollar volume to judge what it can support.

Room 07 · The Tape — Relative volume · paid room

From our own fill modeling: once your order approaches 5% of the minute's volume, expect meaningful slippage; past that it escalates fast. The drill engine charges you for exactly this, because real tape does. Buying 20,000 shares in a minute that trades 45,000 is not "getting filled" — it is the minute.

Room 07 · The Tape — Size vs. minute volume · paid room

From our own segmentation (5,925 runner-days with a ≥100% run-up): float rotation and RVOL moved the hold-versus-fade rate by only 2–4.5 points across their buckets. Structure is sizing information, not a direction call.

Room 08 · Structure — Float rotation · paid room

From our own research — this repetition is measurable, not anecdotal. We maintain a census of 525 names that run the same pump-and-decay script repeatedly. Serial behavior is the norm in this corner of the market: the same tickers, the same script, different dates.

Room 08 · Structure — Share structure history · paid room

From our own backtests (3,641 reverse splits, 2023–2026): about 10% became ≥100% runners within 20 trading days, and fast — a median of 2 days. Buying the split and holding lost at every horizon we tested, at roughly a 0.5 profit factor. Shorting the day-one pop was the one positive-expectancy simple trade in the whole census (PF ~1.4–1.8, and it survived realistic borrow costs).

Room 08 · Structure — Reverse splits · paid room

From our own alert work: the readable lever at the moment the headline hits is registered direct / public offering versus private placement. A registered deal is stock that can hit the tape now; a private placement is typically restricted, so the supply arrives later, through a different door.

Room 09 · Dilution — The 424B5 — the offering itself · paid room

From our own segmentation: dilution-heavy names deep-faded meaningfully more often than the base rate. The sample was modest, so treat it as a prior, not a promise.

Room 09 · Dilution — ATMs — the invisible seller · paid room

From our own research on the extreme end: sub-penny warrant spikes — where warrants themselves become the vehicle — resolved brutally, with roughly 85–95% fading back to their 20-day low. The mechanism generalizes: when the instrument's whole purpose is conversion into sellable stock, strength is the exit, not the destination.

Room 09 · Dilution — Warrants · paid room

From our own alert work — this is why the offering-on-runner alert was built at all: the combination "big runner + fresh offering headline" recurs often enough to be worth monitoring in real time. Our segmentation also found dilution-tagged names deep-fading more than the base rate (though on modest sample sizes — a prior, not a promise).

Room 09 · Dilution — Filing timing patterns · paid room

From our own tape (20,403 published bands, 2026-08-19 to 08-24, bucketed by prior close): sub-$0.75 names got the flat $0.15 in 97.9% of opening bands, $0.75–$3 names got 20% in 98.6%, and over-$3 names got 10% in 92%. The misses were mostly leveraged ETPs, reverse-split artifacts and doubled closing bands.

Room 10 · Halts — LULD mechanics · paid room

From our own halt research (60% of runner days contain at least one halt): after a halt-up, the stock made a NEW HIGH on the reopen 64.5% of the time — roughly two in three. Only about a third faded to the close. Halt-down reopens were far weaker at making new lows (about 18%).

Room 10 · Halts — Resume behavior · paid room

From our own halt research: after a halt-up, the reopen made a new high 64.5% of the time, and a third halt in the chain still did 63%. What shrank was the payoff: median follow-through fell from about +0.6% to +0.2%. The halt times were inferred from gaps in the tape, so news halts and LULD pauses are mixed together in that sample. That was runner days only.

Room 10 · Halts — Halt-and-go · paid room

From our own multi-day research — the bleed is measurable and it is one of the most consistent directional results we have. Following ≥100% runners forward from their day-1 close: the overnight gap was a median −5.9% with 66% gapping down, day+1 about −6.3%, and by day+5 about −13.7%, with 69% of names red.

Room 11 · Multiday — The Day 1 / 2 / 3 lifecycle · paid room

From our own backtests — we tested exactly how much this matters, and it is the difference between a business and a hobby. Our best backside short signal ran a profit factor of 1.55 with no locate cost, 1.31 with a locate fee of 3% of the position per trade, and 1.10 at 6%. Those trades covered the same day, so this is the one-time locate fee, not the annualised borrow rate.

Room 12 · Short Side — Borrow · paid room

From our own census — how much heat "being right eventually" actually costs: across 3,500 random short entries on runner days, the median trade sat through 55% adverse excursion before resolving, and one in ten saw 171%+. On this tape, deep heat is not the unlucky case — it is the normal case, which is why the sizing question below is asked about the tail.

Room 12 · Short Side — Asymmetric loss framing · paid room

From our own alert research — the distinction you can actually read in real time is in the headline: registered direct / public offering versus private placement. That single word choice tells you whether the new stock can hit the tape now or is restricted for later, and it is the load-bearing lever in our shipped offering-on-runner alert.

Room 13 · Catalysts — Offering risk by catalyst type · paid room

From our own census — this is a repeat business, not a series of one-offs. We catalogue 525 names running this script repeatedly, and our halt-chain data shows the same concentration: 1,101 chains across just 625 symbols. The single most useful question about a chart that looks like this is whether this exact ticker has done it before — and the answer is usually yes.

Room 13 · Catalysts — Structural tells of a ramp-and-dump · paid room

From our own multi-day research: for ≥100% runners, the overnight gap ran a median −5.9%, with 66% gapping DOWN. Holding a runner overnight is not a neutral coin flip that gives you more time — it is a position in a distribution that leans against you, taken during the hours when you cannot act.

Room 14 · Swans & Tail Risk — Overnight and weekend gap risk · paid room

From our own census: the median runner peaks around +149% above its prior close, but the distribution has a long right tail with the 99th percentile past +1000%. A "typical" day and a career-ending day live in the same dataset, and they look identical for the first thirty minutes.

Room 14 · Swans & Tail Risk — Fat tails vs. normal assumptions · paid room

From our own backtests: shorting every double top on autopilot, costed, lost money, with 54% of trades stopped out by the breakout tail. Entries at failed bounces against broken structure were the ones that paid.

Room 16 · The Pattern Dictionary — Head and shoulders · paid room

From our own backtests — the sequence is the edge, on both sides of the same fact: the first VWAP loss on a runner resolved up more often than down (the break itself is a bounce, not an entry), while the failed reclaim after it was measurably more bearish. In plain flag language: don't short the pole — short the flag failing.

Room 16 · The Pattern Dictionary — The bear flag · paid room

From our own backtests: shorting the ignition ran a 0.11 profit factor with a 75% median adverse excursion. The extension-gated climax break won 75–80% of trades in every year we tested, and the same entry without the stretch gate was breakeven. A win rate is not a profit factor, so read the 75–80% as how often it worked, not how much it paid.

Room 16 · The Pattern Dictionary — The parabolic arc · paid room

From our own census: 525 names show the ramp-and-dump structure (quiet controlled markup, a crowd recruited late, record volume on the collapse) as repeat behaviour.

Room 16 · The Pattern Dictionary — Wyckoff — the old names for our mechanics · paid room

From our own backtests: 2,878 days where a stock ran ≥100% (2021–26, forward daily bars, so names that later delisted are counted). The overnight gap was down 66% of the time, median −5.9%. Day+1 closed a median −6.3% below the run-day close. By day+5 the median was −13.7%, and 69% were red. The names that held their gains into the close did worst: −18.2% by day+5.

Room 17 · The Long Side — The base rate a buyer is up against · paid room

From our own backtests: we wrote the rule down before running it. Buy every stock gapping ≥20% in premarket at the 09:30 close, −10% stop, 15% trailing stop, 0.5% cost per side, 2023–26. PF 0.72, 28% winners, n=4,515. It lost money in every year and under every exit we tried. Splitting out the serial pumpers changed nothing: both halves came out at 0.72.

Room 17 · The Long Side — Buying the gapper at the bell: PF 0.72 on 4,515 trades · paid room

From our own backtests: on days that ran ≥100% (n=5,925, 2021–26), the opening-range high broke 64% of the time. About half of those breaks held, and 32% of days closed above it. The range low broke more often, 69%. As a costed trade the 5-minute ORB long came to PF 1.01 once survivorship was removed. An earlier 1.43 came from testing only stocks that went on to become runners.

Room 17 · The Long Side — The open: gap-and-go and the ORB long · paid room

From our own backtests: the history, in order. - Naive test: PF 3.7 (gross winnings ÷ gross losses; 1.0 is breakeven). - Obvious biases removed: PF 1.6. - Five filters, tested out-of-sample on 2023–25: PF 3.84 on 38 trades, better than the in-sample period. A yearly walk-forward passed too.

Room 17 · The Long Side — Red-to-green: the edge that was never there · paid room

From our own backtests: same detector, same bars, universe including the voided days (2024-01 → 2026-07): 37% winners, PF 0.65, −0.21R a trade, n=2,380. Days in the cache: 72% / PF 3.07 (n=429). Days not in it: 30% / PF 0.44 (n=1,951). Nothing visible at entry (leg size, price, leg volume, prior-day volume) told the two groups apart.

Room 17 · The Long Side — 86% win rate → 37%: the fib long that looked ahead · paid room

From our own backtests: every US stock up 10%+ on 5× relative volume, 2019–26. Arm on a new high of day, wait for a one-to-three-bar pullback, buy the break of the prior bar's high, stop under the pullback low, take 2R or get out after five minutes. 24,859 trades, resolved on 1-second data. Before costs it breaks even: PF 1.00 in 2019–23, 0.91 in 2024–26.

Room 17 · The Long Side — The micro pullback: 24,859 trades, PF 0.34 after costs · paid room

From our own backtests: runners ≥10% above the 1-minute 50 EMA for an hour, long on the first tap, 2023–26: PF 0.48 after 1.5% costs on 1,112 trades, 0.73 even at zero cost. On an outcome-independent universe (221 trades): PF 0.29, 13.8% winners. All 36 parameter combinations finished below 1.0.

Room 17 · The Long Side — Two in three EMA taps come after the high. Those paid PF 0.04. · paid room

From our own backtests: days that ran ≥100% (n=5,925, 2021–26), descriptive. - A 10-minute stall above VWAP: grind or hold 75%, fast fade 8.6%. - Extension flips it. Stall with the stock up under 100%: 76% grind. Stall at ≥100%: 48% fast fade, 78% deep fade. - Distance from VWAP. 1–15% above was the healthy zone: 35–43% made a new high, next-hour median 0 to +0.4%.

Room 17 · The Long Side — Reading real strength (context, not entries) · paid room

From our own backtests: runner days, 60% of which have at least one halt. Halts are inferred from 5–60-minute holes in the tape, so LULD pauses and news halts can't be told apart. The third halt and later still made new highs about 63% of the time. But the median move over the next 30 minutes fell from +0.61% to +0.17% as halts stacked, and the odds of fading into the close rose from 29% to 39%.

Room 17 · The Long Side — Halts and the crowd: when the tape is on your side · paid room

From our own backtests: 967 listed reverse splits, Jan 2025 – Sep 2026, 1% costs, filing-cancel rule on. - Every trade: n=347, PF 1.16. Close to nothing (PF = gross winnings ÷ gross losses; 1.0 is breakeven). - No reverse split in the prior 12 months: n=187, PF 1.92, 57% winners, average +12%, worst trade −87%. By year: 2025 PF 1.47 (n=90), 2026 PF 2.38 (n=97).

Room 17 · The Long Side — The split-day low bid · paid room

From our own backtests: at 0.5% per side, PF 2.71, 65% winners, PF 2.13 without the best three trades, positive in all four years. Day+1, the obvious day, was 0.94. Add cost per side: +0.5% → 1.79, +1% → 1.18, +2% → 0.51. Breakeven is about 1.6% per side all-in.

Room 17 · The Long Side — The day+2 flush bounce: PF 2.71 at 0.5% a side, 0.51 at 2.5% · paid room

From our own fill modelling (Unit 7): once an order gets near 5% of a minute's volume, slippage becomes meaningful and then climbs fast. The square root explains the shape of that curve. We have not yet fitted the law to our own fills; the regression is first on this Room's test list.

Room 18 · How Prices Form — Bouchaud and the square-root law of impact · paid room

From our own backtests: within a day, trade count and volume move together almost perfectly (correlation 0.96), so counting trades adds little. Its one independent clue is average trade size. Among runner-days with a 10,000-trade minute, those averaging under 150 shares a trade went on to become 500%-plus swans about 8.5% of the time, against 1.1% for those averaging 300 shares or more.

Room 18 · How Prices Form — Reading Level 2 and time & sales honestly · paid room

From our own backtests: entry at the first minute with ≥20,000 trades, 551 runner-days, held to the close. Long: 20% win, median −18.6%. Short: 79% win, median heat −31%.

Room 18 · How Prices Form — 20,000 trades in one minute. Buyers won 20% of the time. · paid room

From our own backtests (US stocks $1–20, 2023–2026): a minute bar's high landed exactly on a whole dollar 25.9% of the time against 18.5% at an arbitrary price, and one cent through it only 7.4% against 11.2%.

Room 18 · How Prices Form — Stops sit a few cents past the obvious price · paid room

From our own halt research (Unit 10): after a halt-up on a runner day the reopen made a new high about 64.5% of the time (56% across all 5,389 halt-ups in the market, 2023–26), and a halt cycle carried price about 1.5× the band width (median, across well-populated price buckets). The auction usually continues the move.

Room 18 · How Prices Form — The halt reopen auction, step by step · paid room

From our own VWAP atlas (43,738 runner-days): extension above VWAP is the only reliable forward predictor, and it weakens with the clock. Runners more than 20% above VWAP went on to return −21.4% measured from 09:45, −15.8% from 10:30 and −9.6% from 12:00. Deeply-below runners went +0.1% to +0.6% forward and faded only 47–49% of the time. Thin runners below VWAP drifted up, +1.6%.

Room 18 · How Prices Form — VWAP as a benchmark — why it pulls, and when it doesn't · paid room

From our own backtests (255 trading days): at first glance it looked like a slam dunk. Dormant stocks that armed SSR ripped the next day 7.7× more often than a flat-stock control. Then the decisive test: if the rule caused it, the rip rate should jump right at −10%. It doesn't. Closes down 9.5–10% (SSR off) ripped 1.02% of the time; closes down 10–10.5% (SSR on) ripped 0.95%.

Room 18 · How Prices Form — Barardehi et al. — what Rule 201 does to the book · paid room

From our own backtests: early beats late almost everywhere we've looked. A short on the first touch of a stretched-above-VWAP reading scored a profit factor of 1.42 when the touch came by 10:00, 1.06 by 11:00 and 0.84 across the whole day. But "it's been running too long" is backwards: of runs that had lasted 60 minutes and were still going, the median one ran 220 more minutes. The chance of topping falls with age.

Room 18 · How Prices Form — Admati & Pfleiderer — why the day has a shape · paid room

From our own data: one premarket in 2025, a stock trading around $2.65 showed minute bars with highs up to $5.415. The wicks were late-reported trades from the 4–6 a.m. price range, stamped minutes or hours later. On that stock and one other the same morning, about half of all prints between 09:15 and 09:30 carried the flag. Both appeared to "bounce to VWAP" only through those prints.

Room 18 · How Prices Form — Data artefacts — when the chart lies · paid room

From our own backtests: we pulled every company with three or more reverse splits out of 19,070 split records since 2010. Of the 124 exchange-listed serial splitters under $5, none trades above the price of its first split, and the median is −99.9%. Yet 79% traded back above that first-split price at some point, and the median name's best 60-session window was +239%. A staircase down, with violent rips on the way.

Room 19 · The Small-Cap Machine — The life cycle of a small-cap · paid room

From our own backtests: S-1 and F-1 filings, 2023–26, 1,576 filings. An initial S-1 by a listed name was followed by a ≥50% drop within 20 sessions 17% of the time, against a 0.95% base rate. Cash on hand was the lever: 25% for companies with under $1M, 4% at $20M or more. And it kept going. From the first −50% close, the median was another −27% over the next 20 sessions.

Room 19 · The Small-Cap Machine — The dilution clock · paid room

From our own research: on 2026-09-25 a Hong Kong micro-IPO ran from a $1.97 close to an $11.42 high with no press release. The catalyst was in its filings. Weeks earlier it had reverse-split 1-for-8 and issued 1,377,000 restricted shares at $3.40 in a share swap, about a 39% increase in share count. At $8 those shares were worth roughly $11M to holders with a $3.40 cost.

Room 19 · The Small-Cap Machine — Who is on the other side of a runner · paid room

From our own backtests: 2026 small-cap one-minute bursts (≥20× the quiet-base volume, a +5% bar). About two-thirds had no wire release or SEC filing at all. The burst gave back half its move within an hour 79–84% of the time with no news and 86–92% with fresh news. Shorting after the peak lost money after costs in every version we tried (PF 0.2–0.8), out of sample too.

Room 19 · The Small-Cap Machine — The touter economy: the alert is the exit liquidity · paid room

From our own backtests: the house data agrees on direction and adds timing. Across 17,545 runner-days (2023–26) joined to company filings, runners on non-earnings 8-Ks faded just like runners with no catalyst, and about 90% of the fade from the peak to the next day's close happened the same day. The attention story plays out in hours. A fade rate is still not a fade edge, as lesson 10 shows.

Room 19 · The Small-Cap Machine — What social attention does to a small cap · paid room

From our own research: the vehicle is often hidden. Our China-nexus detector looks for a statutory disclosure an issuer can't opt out of (mentions of the Holding Foreign Companies Accountable Act). It catches 97.6% of China/HK issuers in our validation set and found 18 names our own data had tagged as US companies. China/HK names were 17–25% of our poppers in spring 2026 and 29–38% in June–August 2026.

Room 19 · The Small-Cap Machine — QMMM, September 2025: the first suspension of the wave · paid room

From our own backtests: buying the split-day low after a reverse split (Jan 2025 – Sep 2026, 967 listed splits, float ≤2M, +30% target) returned PF 1.16 across all splits and PF 1.92 when the company had no reverse split in the prior year. The names the new rule targets are the ones that bounce least.

Room 19 · The Small-Cap Machine — The Nasdaq delisting clock, 2025–26 · paid room

From our own backtests: our day-2 gap fade went from PF 1.13 to 0.97 once a 1.5-point locate fee was charged. Our serial-splitter pop short held up at a 1% round-trip cost (PF 1.21) and was dead at 3% (PF 0.99). A small account trading 1,000-share lots in sub-$2 names pays a higher percentage cost than either test assumed.

Room 19 · The Small-Cap Machine — PDT is gone. The arithmetic isn't. · paid room

From our own backtests: the names most likely to be restricted are where the squeeze tail lives. Among 5,925 runner-days that ran ≥100%, the $100M+ dollar-volume names ran further (median +184% vs +120% at peak) and held better (deep fade 26% vs 77% for thin $1–3M names).

Room 19 · The Small-Cap Machine — Why your broker stops letting you buy · paid room

From our own backtests: a lot of small-cap action already happens outside the regular session. Of 47 pre-merger SPAC trust squeezes since 2023, 60% peaked outside regular hours, and for 57% the peak was a price no regular-session trader could reach; from peak to that day's regular-session close the median move was −52%, negative in all 47.

Room 19 · The Small-Cap Machine — The overnight session, from December 2026 · paid room

From our own backtests: our day-2 gap-fade signal went from PF 2.24 in 2021–23 to 1.13 in 2024–26, and to 0.97 after locate fees. Most gappers still fade. The trade stopped paying. A later rebuild by era found the plain gap-open fade made its money in 2019–20 and has run near break-even after costs since 2021. The day-2 fade held up through 2022 and was at break-even by 2024–26 (Claims Court has the chart).

Room 19 · The Small-Cap Machine — Twice the gappers, a quarter of the tailwind · paid room

From our own backtests (double-extension study, several thousand runner-day setups): the variable that separated continuation from fade on both sides was whether price held a level or false-broke it. Most first-leg highs (83% in the loose version of the detector) were where the run died. When price held a projected rung for 10 minutes, the odds of reaching the next one rose step by step: 62%, 83%, 91%, 97%.

Room 20 · Why Patterns Work (and Fail) — A pattern is a story about who is trapped · paid room

From our own backtests (US stocks $0.50–$20, 2023–2026, about 5 million first touches): price turned back 1% from the first touch of a whole dollar 21.0% of the time, against 14.5% for an arbitrary price on the same charts. That is about 1.4× as often, and it held out of sample. Half dollars did about two-thirds as much.

Room 20 · Why Patterns Work (and Fail) — Osler's order book, or why levels stall and when breaks run · paid room

From our own backtests (failed-breakout wick study, 2021–2026). On 1-minute candles the setup is dead: shorting a long-wick failure above the high of day did worse than a control that shorted any red candle. On 5- to 30-minute candles with a run-up of ≥100% it turns real. At 5-minute with a wick of at least 15% of price it won about 53%, for roughly +0.11R per trade after costs.

Room 20 · Why Patterns Work (and Fail) — Failed breakouts are failed auctions · paid room

From our own backtests (Full-Range Leg study, 44k runner-days scanned): a 3–6 candle green impulse that covers the whole prior move's dollar range was only a modest short as a bare shape (PF 1.26, n=468). What separated the good cases was the dollar volume traded before the leg started, and the ladder ran mostly one way: $1–5M PF 1.25, $5–10M 0.93, $10–20M 1.58, $20–50M 1.80, $50M+ 2.33.

Room 20 · Why Patterns Work (and Fail) — The breakout as a liquidity event · paid room

From our own census (runner-days up >50% with a coil of ≤10% range): 63% of runner-days coil at some point, and most coils go nowhere. On a 30-minute coil, 52% never moved ±10% in the next hour, 25% broke up and 23% broke down. Extension set the lean. At +50–100% coils leaned slightly up (26% vs 21%). At ≥+400% they broke down about twice as often as up (30.5% vs 15.3%).

Room 20 · Why Patterns Work (and Fail) — Coils — absorption or distribution? · paid room

From our own census (runner-days up ≥100%, causal intraday VWAP): the first loss of VWAP was more often a bounce. 39% went on to a new high within an hour, with a median of +1.46% over the next 60 minutes. A reclaim after a loss was weaker: 33% new high, median −0.53%. Over the next 30 minutes, reclaiming VWAP (−2.3%) was more bearish than losing it (−1.7%), against a −0.3% baseline.

Room 20 · Why Patterns Work (and Fail) — The reclaim paradox · paid room

From our own backtests (runner-shape study, two years): held rates ran from 28% for runners under $20M of market cap to 56% above $200M. Predicting hold vs fade from the intraday shape scored an out-of-sample AUC of 0.76. Pre-open setup alone scored 0.66, and adding it to the shape model moved it to 0.767, because by then the shape already carried the setup.

Room 20 · Why Patterns Work (and Fail) — Same shape, different story · paid room

From our own backtests (move-dynamics study, n=5,918 runner-days): holders had 80% of their volume before the top, faders 32%. The holders' climax bar came about 100 minutes before the high; the faders' came at the high. Across quintiles of volume-before-top the held rate climbed from 4% to 55%. Across quintiles of average speed, slow to fast, it fell from 45% to 2%.

Room 20 · Why Patterns Work (and Fail) — Tempo beats shape · paid room

From our own backtests (3-leg exhaustion study): we replayed the whole 1,008-cell parameter search on shuffled outcomes 2,000 times and kept the best cell each time. Noise alone produced a median best PF of 1.75 (95th percentile 2.67). The bare 3-leg shape, with no selection at all, ran PF 1.07 on n=1,643.

Room 20 · Why Patterns Work (and Fail) — What sixty years of pattern research found · paid room

From our own backtests (double-extension study): one Wyckoff-flavoured idea that did measure. The more often price had tested the first-leg high before the setup, the harder the later fade. With one touch, 27% went on to double and the short's PF was 1.53. With five or more touches, 8% doubled and PF was 3.05. Three or more touches happen about twice a month, so use it as a conviction tell, not a gate.

Room 20 · Why Patterns Work (and Fail) — Wyckoff, sorted into what you can test · paid room

From our own backtests: a daily 120% fib-extension short (402 events, 2023–2026) lost to a plain "+46% over prior close" trigger at matched extension, PF 0.75 vs 0.98. Going long the 61.8% retracement tag after a first leg won about 28% and lost money on average.

Room 20 · Why Patterns Work (and Fail) — Every pullback lands near a Fibonacci line · paid room

From our own backtests: our closest real version, a first-pullback scalp with a +5% target and −20% stop, printed 86% winners and PF 1.93 over 2021–26 at 0.3% slippage. The edge is 1–2% a trade, less than a realistic small-cap spread plus fees. We file it as cost-bound.

Room 21 · Claims Court — A 26% win rate · paid room

From our own backtests: in a premarket-runner short study, stocks flagged as on the short-sale restriction (SSR) scored PF 1.54 and unflagged ones 0.88. The tilt passed a 2,000-subset null, confidence intervals and replication in two separate years. Then we read the code behind the flag. It marked a stock as SSR if it had dropped 10% below the prior close during the session being traded.

Room 21 · Claims Court — How to read a backtest · paid room

From our own backtests: in a study of dormant stocks that suddenly spike, keeping only events where the stock peaked within 30 minutes and ran less than 150% printed PF 3.3–8.8 with 86% winners. Both filters are measured from the peak, which you can't see until it's behind you, and together they quietly cap every loss. With filters you could have known in real time: PF 0.37–0.63.

Room 21 · Claims Court — Look-ahead, small n and the grid · paid room

From our own backtests: 2019–26, whole market, 1-minute bars, a universe picked before any outcome was known; 1,186 gap-open fades and 156 day-2 fades. Gross PF fell along with net (the gap-open fade went from 3.73 before costs in 2019–20 to between 1.6 and 2.2 in each era since 2021), so costs alone can't explain it.

Room 21 · Claims Court — PF 2.40, then break-even · paid room

From our own backtests: Gap-open fade: PF 1.68 in-sample, 1.22 out of sample, about break-even after locates. Retired, and removed from the scanner and the bot. Day-2 fade: 2.24 in 2021–23, 0.97 after locates in 2024–26. A kill switch went in, and whether to retire it outright is still an open decision on our desk.

Room 21 · Claims Court — Building and killing a playbook entry · paid room

From our own backtests: a 1-second replay through the production detector, market-wide, 2023–26: n=128, PF 3.17, win 77% [69–84], average +14.9%, stopped out 12%, about 36 a year, by year 2.65 / 2.69 / 3.58 / 3.71. The $2 price floor is what fixed 2026, and it was chosen after seeing 2026: the $1–2 group was good for three years and broke in one (n=19).

Room 22 · The Short Book — Card: the VWAP stack short (gated off) · paid room

From our own backtests: 2019–26, every gapper day, prior close $20 or less, ranked by premarket dollar volume at 09:29 and shorted plain at the open (+20% stop, cover at the close, 1.5% fee). PF by rank: #1 1.50, #2 1.36, #3 1.13, #4 to #10 0.93, #11 and lower 0.62, on 1,945 days. #2 beat #1 in 2 of 8 years.

Room 22 · The Short Book — Card: short the day's volume leader · paid room

From our own backtests: 2020–26, the 2×-in-5-minutes trigger fired 178 times. Waiting for the VWAP loss with a +50% stop: PF 3.00 (first period 3.13, second 2.89), worst trade −52%, about 2.2 a month. Using candle bodies for the target instead of wicks, PF 3.24.

Room 22 · The Short Book — Card: the 4am double, after the VWAP loss · paid room

From our own backtests: January 2025 to August 2026, n=67, PF 2.87, win 54%, average +10.6%, about 3.4 a month, against a control of 0.75 on the same days. Every period was positive (2025 H1 4.45, H2 1.90, 2026 H1 3.08, July on 2.99). Rebuilt for 2026 on a universe with no "it ran 100%" condition (5,909 candidate days), the rule got better: PF 3.53 to the close on 29 trades, control 1.30.

Room 22 · The Short Book — Card: the afternoon blast off a dead morning · paid room

From our own backtests: 1-second replay through the production detector, 2023–26: n=89, about 25 a year, PF 2.74, win 75% [65–83], average +10.9%, every year 2.20 or better. Now the discount. The same search run on shuffled outcomes 2,000 times found a median PF of 1.75 on pure noise, so 1.75 is what a search like this buys for free.

Room 22 · The Short Book — Card: three even legs (gated off) · paid room

From our own backtests: a 1-second replay through the live detector. Shipped settings, by year: 2024 (out of sample) PF 1.04, 2025 4.24, 2026 (January–July) 4.95, win 65 / 80 / 85%, 43 / 41 / 34 trades. The 2024 floor is breakeven after the 1.5% haircut. Before we added the 70% fade ceiling, 2025 had a high win rate and a PF of only 1.25, and four trades explained it.

Room 22 · The Short Book — Card: short the bounce, not the flush · paid room

From our own backtests: 5-minute failed breakouts on 100%+ runners, 2021–26. With the wick at 15% or more, win 53% and about +0.11R a trade after costs. The wider 7%-wick version was positive gross in every year on 5-, 15- and 30-minute candles. It beat a matched control (same run-up, same wick, but the candle never reached the high), so the break of the high carries information of its own.

Room 22 · The Short Book — Card: the failed push above the high · paid room

From our own backtests: March to September 2026. 3-day gate: n=47, 74% winners, median +14.2%, PF 6.1. Without the gate, 830 headlines made PF 1.75. Priced deals: 36 trades, 83% winners, PF 10.8. Proposed deals: 11 trades, 45%, PF 0.24. After-hours releases traded the next morning: 15 fires, 40% winners, PF 0.27. That last group is where the losers lived.

Room 22 · The Short Book — Card: the priced offering on an elevated name · paid room

From our own backtests: 2023–26, 1,576 filings, 267 knives. Short at the first −50% close, cover 20 sessions later: 75% winners, PF 3.17; F-1 filers 7.98. From that close the median kept falling 27% (knives without an S-1 fell 7%). Borrow and halts are not modelled, and on a stock that just lost half its value, borrow is the question.

Room 22 · The Short Book — Card: the dilution knife and the bounce after it · paid room

From our own work: in the warrant-exercise montage (15 induced exercises, April–July 2026), the five deepest fades all added at least 43% to the float. Float-add, not dollar size, is the number to carry. Fifteen events is a small montage.

Room 23 · The Dilution Desk — Discount, coverage, and the dilution nobody headlines · paid room

From our own backtests: on 437 registered offerings, the 30 whose headline said "priced at-the-market under Nasdaq rules" (every one a registered direct or follow-on, none an ATM program) fell a median 8.2% from the first minute after the release to the close, 63% of them down. The other 407: 0.3% and 55%.

Room 23 · The Dilution Desk — Three things called "at-the-market", told apart · paid room

From our own alert work: two of our offering matchers read press-release bodies until August 2026. In one of them, over eight weeks, 279 hits came from the body alone, and they were close to 100% junk (one was a travel guide). The trigger case was MMA: "Completes US$4.0M Private Placement at US$1.00 Per Share", priced far above the prior close with no warrants. It was flagged as an offering.

Room 23 · The Dilution Desk — Form D and the Reg D trap · paid room

From our own backtests: registered offerings on runners, n=49: the stock fell a median 16.8% to the close, 73% of them down. Private placements, n=30: fell a median 0.5%, and 50% down, a coin flip. Through a 25% stop the private short lost a median 10.7%. The worst private shorts went 76.7%, 74.2% and 61.6% against the entry.

Room 23 · The Dilution Desk — Why a discounted PIPE can pop (Hertzel & Smith, 1993) · paid room

From our own screen: of 77 names under a $5M float value rated Low on both scores, 25 carried a High rating for dilution history, with a median three-year share growth of +2,643%. The screen we built on that finding uses both capacity and behaviour, the behaviour leg being three-year share growth and the reverse-split count. One split can be a compliance fix. Two or more is the fingerprint.

Room 23 · The Dilution Desk — Rated low dilution risk. Share count up 223,350%. · paid room

From our own backtests: 1,576 S-1 and F-1 filings by listed companies, 2023–26, 267 knives (a close 50% below the pre-filing close within 20 sessions). Everything below is measured on those 267.

Room 23 · The Dilution Desk — When an S-1 knife keeps cutting · paid room

From our own backtests: no stop, win 73%, PF 4.93. With a +25% stop: win 62%, PF 2.56. The stop saved one trade (SOPA, −53.5% cut to −25%) and turned four into losers, including HKIT, a +80.0% winner stopped at −25%. 28.5 points saved, 183 given up. A +100% stop never triggered once in 26.

Room 23 · The Dilution Desk — The stop that pays for the squeeze · paid room

From our own work: we set SPRC next to 19 warrant and note inducements. The six that combined a deep discount (19% or more under the market) with an elevated price (40% or more over the three-day low) closed the day a median 31% lower, and every one was 40–57% lower three days later. SPRC missed the elevation cut and fell further than any of them. It was also the deepest discount in the set.

Room 23 · The Dilution Desk — SciSparc, 15 October 2026, a note converted at 65% off · paid room

From our own research: runner count from exchange daily bars for every listed stock, 1,400 sessions. Depending on the cut-off the correlation lands between 0.49 and 0.65. Our runner census printed 0.73 because it counts only ≥100% runners, sees premarket highs, and leaves zero-runner days out. Twenty sessions out it is about 0.2. The 0.12 comes from the clearouts study (6,990 events, 2021-01 to 2026-08).

Room 24 · Tape, Regime & Crowd — Hot tape follows hot tape: 0.6 · paid room

From our own research: crowding study, 2026 YTD plus a frozen-spec 2025 holdout (395 sessions); monster-vacuum study, 416 sessions; same-day swan study, 901 days. The crowding study also tested the obvious trade, shorting every lit name once five were lit: PF 1.09 before costs in 2026, and a loss after a 1.5% locate fee.

Room 24 · Tape, Regime & Crowd — Why every tape signal ends up a tilt · paid room

From our own research: Bitcoin's own tape barely predicts any of it. On ignition days BTC was up 53% of the time, against 51% on all days. The ignition rate was 31% when the coin was in a ten-day uptrend and 23% when it was in a downtrend, a tilt. And June 2026 produced 25 ignitions across 14 names while Bitcoin sat at its $58.5k low, 54% off the top. Trade the names, not the coin.

Room 24 · Tape, Regime & Crowd — The relay: one name at a time · paid room

From our own research: an indirect check from a wider angle, measured on how followers behaved rather than on who followed. After a foreign swan, followers that matched its float tier faded 5.4 points more often than the base, matching market cap 4.3, matching country 3.7, matching sector nothing. And in a separate study of midday pumps, Asia-domiciled names behaved like everything else (16.9% vs 16.5%).

Room 24 · Tape, Regime & Crowd — Sympathy runs on structure, not sector · paid room

From our own research: lead and lag from 1-minute bars in the polygon-aggs-1min cache; ranks from the point-in-time rebuild (prices, share counts, splits and filings dated on or before 24 September). The first pass at this screen used a data snapshot pulled on the Sunday, which already contained Friday's move. Rebuilding it from as-of data is what made the 2nd-of-417 result count.

Room 24 · Tape, Regime & Crowd — INLF, MSGY, TDIC: a sympathy chain, minute by minute · paid room

From our own research: clearouts study, 6,990 events from exchange daily bars, 2021-01 to 2026-08; the dollar-volume split uses the scanner's runner rows for the same months, which see premarket highs and so count more clearouts than the exchange-bar series above.

Room 24 · Tape, Regime & Crowd — Regimes describe. They don't forecast. · paid room

From our own research: multi-day follow-through on 2,878 runners of 100% or more, from each ticker's own forward daily bars, so names that later delisted or vanished from our runner table are still in. Across all of them: overnight gap −5.9% median (66% gap down), day one −6.3%, day five −13.7% with 69% red.

Room 24 · Tape, Regime & Crowd — The horizon flip: by full-day volume, the best multi-day short is the worst intraday one · paid room

From our own research: the 1,101 chains are a floor, not a count. Halts before August 2026 had to be rebuilt from holes in the minute bars, spot-checked against the LULD limit-states in the quote feed, and we threw out every name trading under $10M a day because there neither the detector nor its check could be trusted.

Room 25 · The Museum — 625 symbols, 1,101 halt chains · paid room

From our own research: this shape repeats. We have catalogued 525 names whose charts ran a spike-then-decay cycle more than once, and the most useful question about a chart like this is whether this ticker, or one built like it, has done it before. In August 2022 the closest precedent was three days old, one ticker over.

Room 25 · The Museum — Magic Empire Global: the rerun opened three days later · paid room

From our own research: DryShips is an extreme case of a common one. Of 664 tickers with three or more reverse splits since 2010, the 124 exchange-listed dilution splitters (under $5 at their last split) we could measure cleanly are all below their first-split price, median −99.9%. Yet 79% of them traded back above that first-split price at some point, and the median name had a +239% best 60-session window.

Room 25 · The Museum — DryShips, 2016–17: eight reverse splits in sixteen months · paid room

From our own research: in our census of the biggest sub-$1-base run-ups from 2021 to mid-2026 (367 episodes of 4× or more), FFIE's May 2024 run is the largest we have, measured on prices that actually traded rather than single-print wicks. Across the census the median name kept 15% of its peak-day close a year later.

Room 25 · The Museum — Faraday Future, May 2024: four cents to $3.90 · paid room

From our own research: we took an exchange symbol-change feed (623 changes: 460 small-cap renames, the rest bankruptcies, OTC demotions and unit splits) and measured day one for the 509 changes with clean data. 80% were noise: median day −2.3%. The other fifth was a tail that peaked early, 70% of the day highs printed before 09:30, and then faded: median −49.5% off the high, 58% closing below the 09:30 open.

Room 25 · The Museum — New ticker, same company: four in five do nothing · paid room

From our own research: Eightco's day is one of our seven "anchor days" since 2023 (800%+ and $300M+ traded). The rename to ORBS came three sessions after the run, on a gap down, so the rename record doesn't apply to it. And we keep a separate check for treasury companies, because the token pile is often not what the share price assumes it is: locked or vesting tokens are worth less than spot to the company that…

Room 25 · The Museum — Eightco, 8 September 2025: the treasury template · paid room

From our own research: we built a recurrence badge into the scanner (↻) because recycled catalysts are common enough to need one: another milestone payment, another warrant exercise, another distribution agreement. We haven't yet measured whether repeats fade harder than novel news. VIVS is one case.

Room 25 · The Museum — VivoSim, January and July 2026: the same pop twice · paid room

From our own research: we went back and asked whether TDIC could have been found before 25 September, using only what was public at the close of 24 September. A structural screen of 417 Asia-based names, scored on price, share count, domicile, reverse-split recency, IPO vintage, dead tape and prior no-news spikes, ranked it 2nd, behind only Masonglory itself.

Room 25 · The Museum — Dreamland, 2026: the same ticker three times in five months · paid room

From our own research: INHD was the seventh of our anchor days, and it came in the week our sticky-tape study dates as the turn from the stickiest three months in our sample (March–May 2026, every month 44%+ of the run retained after five days) to one of the deadest (June–August, clearout rate up from single digits to 24%).

Room 25 · The Museum — Inno Holdings, 8 June 2026: halted at 17:18, reopened 36 sessions later · paid room

From our own research: the anchor cycle comes from a weekly regime query over our daily runner table (runners per day, 100%+ count, giant days, dollar volume, by week since the anchor). The 20-session test used 6,990 outcome-independent pop events from 2021 to August 2026, built from grouped-daily bars rather than from our runner list.

Room 25 · The Museum — The anchor-day cycle, and how far to trust it · paid room

▪ Read the rooms behind these

Six rooms — chart literacy, levels, trend, reversals, breakouts and risk — are free forever, no card. Start the course.

Six rooms, free

Free account · no card