Why this exists
Most retail forex tools sell certainty: "98% accurate AI signals," "guaranteed daily profits," "millionaire-maker indicator." The math doesn't survive a backtest, the marketing doesn't survive a regulator's letter, and the user account doesn't survive the month.
Aetherion takes the opposite stance: state every rule, version every change, and stand down by default. The methodology — three pillars, multi-timeframe alignment, engulfing at AOI — is classical price action. It's been written about by BabyPips, ForexTrainingGroup, FXOpen, and countless others for decades. It's not new. It's not proprietary. It works because the discipline works.
The contribution Aetherion makes is operational: turning the discipline into a checklist a machine can run consistently against your live chart, with an audit trail you can read line by line.
The version timeline
v3.0 — The institutional baseline
Started as a codification of classical price-action principles with custom institutional weekly-trading filters. Three pillars, five timeframes, OANDA broker, basic grading scale. Worked on paper. Survived two months of live use. Got patched.
v3.5 — Conservative Mode 2026
Trump-era macro volatility broke the original 1% risk model. Conservative Mode dropped risk per trade to 0.5%, raised the minimum trade grade to B+ (80%), required 1:3 R:R, and capped concurrent positions at one. The philosophy line that shipped with this version: "In uncertain markets, the best trade is often no trade."
v4.0 — Head-and-shoulders patch
Promoted head-and-shoulders to the primary reversal pattern for high-volatility regimes — a classical price-action setup used by professional traders for over a century. Got integrated as a high-confluence pattern with explicit invalidation rules (close beyond the head, neckline retest failure, etc.).
v4.1 — Single-confirmation engulfing doctrine
Introduced the discipline of refusing to stack indicators. If you have trend + AOI + engulfing, you have enough.
Adding RSI, MACD, BBands, stochastics — they don't increase win rate, they increase hesitation. The doctrine became
a setups.SINGLE_CONFIRMATION_ENGULFING_AT_AOI block.
v4.4 — Live-analysis framework
After multiple weeks of live testing, five new rules got added: all-time-high exclusion (no historical reference = no AOI), multi-week patience expectation (week 1 retest, week 2 rejection, week 3 push), trend > AOI tiebreaker, round-number psychological confluence, and market-open entry blackout (no Sunday/Monday entries). Plus a scanner-behavior fix: "range at AOI" is setup-pending, not noise.
v4.5 — H1 execution-timeframe directive
User directive on 2026-04-21: shift from H4 to H1 as the primary execution timeframe. Earlier pattern detection, finer AOI precision, better entry/SL placement. Trade-off: H1 has more noise. Counter-measure: HTF alignment requirement became a hard veto, not a downgrade.
On the GBPNZD trade the next day, this exact rule prevented a bad short. H4 looked perfect for the entry; H1 had quietly flipped bullish at the support retest. The v4.4 system would have entered. v4.5 stood down. Price rallied 80 pips before reversing. The discipline paid.
v4.6 — Split top-down framework
Formalized the two-phase top-down architecture: W/D/H4 determine direction (the "trend phase"), H2/H1/M30/M15 determine entry (the "execution phase"). Added the explicit 2-of-3 majority HTF tiebreaker, structure-point durability quantification (a weekly pivot is ~30x more durable than an H1 pivot), and the "market shows hand first" entry veto.
v4.7 — Day-trading specialization
Locks the framework to three explicit timeframe roles: TREND from W/D/H4 (2/3 majority), ENTRY SIGNAL from H1/M30/M15/M5 (engulfing detection across the whole entry phase), AOI from H1 only (single source of truth — no conflicting levels from multiple TFs). Trade execution chart: H1. Scanner now collects 7 timeframes per pair. Annotated charts draw the H1 AOI zone explicitly; W/D/H4 levels appear as faint context lines for reference only.
v4.8 — Live watchlist sync
The website now reads the user's TradingView watchlist panel live via Chrome DevTools Protocol and uses it as the source
of truth for scans. Add a forex pair in TradingView, the website picks it up within ~60 seconds (or instantly via the
🔄 button). Only forex pairs are kept; equities/crypto/futures are silently filtered out. The static array in
rules.json becomes a fallback when the panel is closed or unreachable.
v4.9 — Predictive scenario mode
Deliberate softening of the original "no predictions" doctrine. The chat bot now provides probabilistic scenario analysis with a confidence percentage capped at 75%, and the annotated chart includes a forecast banner showing the bias direction with a faded forecast arrow drawn on the chart. Five hard rails kept it honest: confidence cap, both bull and bear scenarios always shown, invalidation level always stated, mandatory ⚠ disclaimer on every predictive output, and conservative_mode_2026 still gates execution.
v5.0 — M15 execution + H1 must-match gate
Shift in execution timeframe: H1 → M15. M15 fires entry signals faster than H1, but it's noisier — so v5.0 added a critical brake: H1 and M15 must BOTH be colored AND in the same direction before any recommendation. If they disagree, no trade. A new "H1+M15 match" badge appeared above the Three Pillars showing whether the gate passed.
v5.1 — Dual-chart annotation: H1 + M15 side-by-side
Refinement of v5.0: instead of showing only the M15 chart, every analysis now shows BOTH H1 and M15 annotated charts side-by-side. Each has its own AOI (drawn from that timeframe's swing high/low), its own forecast bias, and its own entry / SL / TP levels — H1 with wider stops (~50 pips, 1:2 / 1:3 RR for daily swing holds) and M15 with tighter stops (~25 pips, for faster intraday entries). Philosophy: "Trade on H1 daily; use M15 for the fastest possible entry timing; both inform the decision." The H1+M15 must-match gate is unchanged.
v5.2 — Statistical edge + risk calculator + chat memory
Four additions:
- Statistical edge panels: every analysis now shows Edgeful-inspired probability stats for H1 and M15 — range distribution histogram, break-probability bars (% of last 40 bars that broke prior high/low), bidirectional momentum strength gauge, and a volatility regime badge (normal/elevated/extreme).
- Risk calculator page at
/calculator: Polymarket-style outcome preview for any forex trade. Inputs: balance, risk %, pair, entry, direction, SL pips, R:R, leverage. Outputs: lot size, $ risk, margin, and side-by-side scenarios for SL hit / TP1 hit / TP2 hit with bar visualizations. - Chat-bot trade-log memory: the AI now reads
trade_log.jsonand includes your win rate + recent trades in its context. Ask "what's my win rate this month?" or "how have my B+ trades performed?" and it answers from your actual data. Honest framing: this is statistical pattern memory, not machine learning. Real ML/RL training requires labeled data + GPU compute we don't ship. - Site-specific web search: chat bot recognizes 16 trusted trading sources (ForexFactory, DailyFX, Investing, Edgeful, Reuters, Bloomberg, etc.) and auto-restricts search to those domains when mentioned. "Search ForexFactory for EURUSD news" → results only from forexfactory.com.
v5.3 — Trade journal + AI pattern analyzer + active-trade flag (current)
This closes the "AI self-learning" feedback loop honestly. Rather than fake ML training (which the system repeatedly declines), v5.3 ships a real workflow:
- Trade journal at
/journal: log every trade you take with full context (pair, direction, entry/SL/TP, lot size, verdict tier, notes). Mark outcomes when they close (TP1 / TP2 / SL / breakeven). Backed bytrade_log.json. - AI Pattern Analyzer: once you have ≥3 closed trades, click "Analyze patterns." The AI
reads your real win rates broken down by pair, grade, verdict tier, and direction — then proposes 1-3
specific rule changes with empirical reasoning ("Your MONITORING tier shows 28% win rate on 8 trades; 1:3
R:R needs ≥25% to break even but you're not capturing 1:3. Recommend disabling MONITORING entries.").
Auto-application is intentionally NOT supported — review and manually edit
rules.json. - Active-trade flag: when a pair has an OPEN trade in the journal, the scan summary table shows a 📌 marker so you don't accidentally double-trade it. The row gets an amber accent.
- Quick-log button: every analysis result with a tradable direction now shows "Took this trade?" with a one-click deeplink to the journal pre-filled with entry/SL/TP from the analysis.
What Aetherion is not
- Not a broker. It does not hold funds, place orders, or interact with your trading account in any way.
- Not a registered investment advisor. Nothing here is investment advice, a solicitation, or a recommendation.
- Not a black box. Every rule lives in
rules.json. You can read it, edit it, audit it, and roll it back. - Not a guaranteed-profit system. No such system exists. Anyone selling you one is committing fraud.
- Not a replacement for your own judgment. The tool gives you a structured second opinion. The trade decision is yours.
The credibility stance
Aetherion's mechanical rules — engulfing at AOI, HH/HL structure, three pillars, multi-timeframe alignment — are independently documented by classical price-action sources: BabyPips, FXOpen, ForexTrainingGroup, and decades of published trading literature. None of this is proprietary or new — it's discipline applied consistently.
We ignore unverified marketing claims (specific win-rate boosts like "25-30% boost," student income figures like
"$1000-1500/week") that circulate around forex education content. Those are unverifiable by definition.
Aetherion's stance: adopt the mechanics, ignore the hype, always backtest before risking real capital.
Every caveat is recorded in the JSON's credibility_caveats sections.
Open architecture
Aetherion is cloud-hosted. The frontend is plain HTML/CSS/JS. The backend is a set of serverless functions deployed on Vercel. The strategy is a single rules file. The TradingView connection uses Chrome DevTools Protocol — no proprietary APIs, no third-party data feeds, no telemetry.
You run it on your machine. Your data stays on your machine. Your trades stay between you and your broker.