The bot's Fixed Scan, Daily Consistency and Super Search, running entirely in your browser from research.db. Same kernels: EMA cross at candle close, ATR(14) Wilder SL, TP = SL×RR, Stop & Reverse or Hold, smart trailing (BE at +1R → chandelier), one-trade-at-a-time. All times UTC (Dhaka = UTC+6). Session windows are fixed UTC buckets with no daylight-saving adjustment — self-consistent across all tabs, and matched to the EA via the session time strings printed in Tomorrow's Plan.
Loading database…
🕐 Upload history (last 10)
Loading…
📜 Trade Log
💡 Why this ranked here
ℹ️ How this tab works
🌙 Tomorrow's Plan — the nightly answer sheet for your EA. One click analyzes
BTC M1 · BTC M5 · XAU M1 · XAU M5 · ETH M1 · ETH M5 (whatever exists in your database) and for each one answers:
(1) a summary table of the Global Best + each session's best pair with a comparable Quality Score, then BOTH plans built out — Adaptive (session-specific profiles, the statistically-preferred default) and Simple (Global Best in all 5 profiles) — with a data-driven verdict: Adaptive is recommended only when the improvement is meaningful (≥15% net, or PF ×1.2, or ≥30% fewer whipsaws), because a 1–2% gain never justifies extra complexity ·
(2) which EMA cross to run tomorrow ·
(3) full parameters in your EA's exact input names (Tky_/Tlo_/Ldn_/Lno_/Nyc_): Fast/Slow EMA, Exit mode, RR, SL = ATR×, TP = SL×RR, ADX band, RSI band ·
(4) a WHY explanation from past statistics — text only, no graphs.
How it decides: Stage 1 scans all EMA pairs on a base config and shortlists the strongest per session + all-sessions. Stage 2 refines the shortlist across Exit mode (Hold/S&R) × RR (2/3/5) × ATR (1.5/2/3). The winner must be net-positive over the full window AND non-negative over the recent check window (the "still working now" guard — tomorrow resembles last week more than last month). ADX (band X–60) and RSI (band) filters are then tested on the winner's actual trade list and recommended ONLY if they improved past net by >5% — otherwise OFF. Filters are tested for Hold mode only (in Stop & Reverse, skipping entries changes the whole chain, so an honest test isn't possible from the same trade list — they're set OFF there).
⚠ Honesty notes: every figure here is NET of your spread inputs (per-trade cost = spread × contract × lot) — this alone kills many fake M1/low-timeframe edges whose per-trade profit is smaller than the spread. In walk-forward tests on real data (select on the past, measure on the unseen next day), picked plans kept roughly HALF of their backtest daily average — the card shows this discounted realistic expectation. Covers all nine combos: BTC/XAU/ETH × M1/M5/M15, and opens with a QUICK LOOK table (tomorrow's date, each combo's EMA cross(es), config, yesterday result and a TRADE/SKIP action) so one glance answers what to run. M15 is Simple-only: walk-forward testing proved per-session optimization fails at M15 (too few candles per session — XAU adaptive lost $241 while the global plan made $93 on unseen days). The One-trade checkbox re-runs everything in flat-until-resolved mode: tested on XAU M5, false earned +$187 vs +$68 for true over 15 unseen days — false makes more, true caps risk at one position (and disables filter testing, which becomes dishonest when entries are chain-dependent). Each combo also runs a YESTERDAY CHECK: the identical selection is re-run on data ending before the last day, and what that plan actually earned on the last day is shown — if it lost, the card tells you to skip or halve size (this meta-momentum gate cut a losing BTC regime from −$52 to ≈$0 in walk-forward tests, at modest cost to winning regimes). Doing this doubles the scan time; it is worth it. An EMA-neighborhood check flags isolated parameter spikes (likely curve-fit) vs plateaus (more trustworthy). M1 uses a different selection method than M5: candidates are ranked by the RECENT slice (full-window profit is only a gate) — in walk-forward tests, normal ranking LOST money at M1 while recent-ranked roughly broke even; treat M1 plans as experimental and M5 as primary until your own live logs prove M1. IMPORTANT session alignment: your EA's session windows run on BROKER time and its defaults do NOT match this tool's UTC sessions — each plan therefore also prints the exact Inp*Start/End time strings (converted by your broker-offset input) to paste into the EA once; without them, the EA trades different hours than the ones these numbers were computed on. XAU/forex weekends are closed, so a plan generated Friday night for Saturday applies to Monday for Gold. M1 backtests ignore spread — with many trades, real spread can cut the shown net substantially. Past statistics can't promise tomorrow; this picks what has been most consistent lately, nothing more.
⚙️ Settings
cap risk to 1 position
Load the database first
⚙️ Settings
/
Load the database first
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
ℹ️ How this tab works
Daily Consistency, scoped to whichever day(s) of the week you tick — Saturday through Friday, any
single day or any combination. "Occurrences" = N means the last N actual matching calendar dates found in your
data, not N calendar days back. Tick only Sat+Sun for a weekend-only search (the default); tick one day (e.g.
only Monday) to study that specific day; tick all 7 for the same result as ordinary Daily Consistency.
⚠ Gold/forex markets are normally CLOSED on weekends — weekend results are mainly meaningful for BTCUSDm/ETHUSDm.
⚙️ Settings
/
Load the database first
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
ℹ️ How this tab works
🧠 Smart Scan = Daily Consistency + anti-whipsaw entry filters. A cross only becomes a trade if:
(1) EMA gap — |fast−slow| at the cross is at least your ×ATR threshold (chop crosses have near-zero separation, real trends open a gap);
(2) Confirmation candle (optional) — enter one candle later, only if that candle closes beyond the fast EMA in the signal direction (whipsaws reverse immediately, real breaks follow through);
(3) Cooldown — a cross within N candles of the previous cross is ignored as flip-flop noise;
(4) Trend filter (optional) — the cross candle must close on the EMA200 side of the signal (BUY above / SELL below), optionally with the EMA200 sloping the same way — the classic "trade with the big trend" rule, one binary switch so overfit risk is low;
(5) HTF filter (optional) — the last closed higher-timeframe candle must have EMA20 > EMA50 for BUY (reverse for SELL) — multi-timeframe agreement with zero lookahead;
(6) Entry style: Pullback (optional) — instead of entering on the cross, wait up to N candles for price to wick-touch the fast EMA and close back in the signal direction, entering at that close (this replaces the Confirm setting when selected).
All gates are evaluated at the cross candle, exactly the way the SmartScan EA reads them.
The RR×ATR grid is deliberately limited to the sane zone (RR 1:2–1:5 × ATR 1–3×) to reduce overfit lottery winners.
Ranked by Smart Score = net P/L + win-rate margin above the RR breakeven + green-day % + sample size − losing-streak penalty. Only net-positive combos rank (Custom EMA shows the honest result either way).
With Session = All, every row also shows its single BEST session — and clicking a row prints the full per-session split in the insight box, so you can compare Tokyo / London / NY / laps vs all-sessions for the same cross.
⚠ PORTABILITY: your Tomorrow's-Plan EA has NO gap / confirm / cooldown / trend / HTF / pullback inputs — Smart Scan results are traded with the dedicated SmartScanEA.mq5, which exposes every filter on this tab as a 1:1 input (Fast/Slow EMA, RR, ATR×, gap, confirm, cooldown, trend filter, HTF, entry style + wait, exit mode, trailing, one-trade, session window). Nightly workflow: run the scan → copy the winning row's settings into the EA inputs → deploy on VPS.
⚠ HONESTY: unlike Tomorrow's Plan, NONE of these filters are walk-forward validated — the same in-sample-looks-better trap that caught ADX/RSI applies to every one of them, and each extra filter you switch ON is another chance to curve-fit. Re-test any winner on a second, non-overlapping date range before trusting it, and prefer configs that stay profitable with filters OFF too. One more caveat: with a specific Session selected in Stop & Reverse mode, the backtest chain trades 24h and only counts in-session entries — an EA restricted to that session will diverge; for session-restricted live trading prefer Hold mode, where trades are independent.
⚙️ Settings
/
🔬 Advanced Filters (all off by default — each is independent)
3. ATR ExpansionTrade only when volatility is expanding
4. Bollinger SqueezeEnter only after squeeze breakout
5. ADX StrengthFilter weak/choppy markets
6. RSI MomentumBUY 50–80 · SELL 20–50 momentum zones
7. Volume ConfirmationTrade only on above-average volume
Load the database first
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
📋 SmartScanEA input text — copy these into the EA
ℹ️ How this tab works
🌍 Session Strategy Finder — pick only Market + Timeframe and run. One pass scans every EMA pair (Fast 2–100 × Slow 5–200), then evaluates the SAME trade list against all five sessions separately (entry time decides the session), so all 5 Top-20 rankings come from a single scan.
Trend engine (fixed on purpose): Stop & Reverse · far TP (TP = SL×10 default, so it almost never caps a trend) · no trailing — each trade rides until the opposite cross ends it: "BUY ———— Exit", one long move per signal.
Quality Score (0–100), weighted: 30% Trend Quality (how much of each move is captured + time-in-trend + few insignificant trades) · 20% Profit Factor · 15% Expectancy (in R) · 10% Net Profit · 10% Low Whipsaw · 5% Drawdown (recovery factor) · 5% Trade Count · 5% Session Consistency (stable across the window's quarters AND across high/low-volatility entries — one lucky stretch can't dominate).
Whipsaw Score (0–10, lower = better): instant flips (≤3 candles) + quick-reversal losses (≤5 candles) + over-frequent crossing. Confidence % blends sample size with stability: ~300 stable trades ≈ high 90s, ~40 good trades ≈ low 70s, under 20 trades = flagged NOT statistically reliable regardless of results. Hard filters: net-positive · PF ≥ 1.2 · whipsaw rate ≤ 50% · minimum trades.
Every row has a 💡 Why button: score drivers, volatility & direction conditions it excels in, best/worst hours, per-session comparison, consistency quarters, failure conditions, and recommended use.
⚙️ Settings
Load the database first
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row in any session table to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
ℹ️ How this tab works
Mode 1 — One cross at a time: enter on EMA cross, ignore ALL subsequent crosses while in trade. After TP/SL, take the next cross. Prevents overlapping entries during a trending run. Mode 2 — One cross + whipsaw filter: only enters when the cross direction is OPPOSITE to the last closed trade (genuine reversal). Consolidation crosses that alternate back and forth are ignored — only the first real direction change after a trade resolves triggers the next entry. Whipsaw window (Mode 2): min candles between two crosses to count as a genuine new reversal rather than chop. Fewer trades is expected and is the point — quality over quantity. Ranked like Daily Consistency: fewest losing days → day % → net P/L.
⚙️ Settings
/
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
ℹ️ How this tab works
A "day" is active if at least one trade was entered that UTC day; PERFECT ★ = zero losing days. Ranked by fewest losing days → day % → net P/L. Caution: this ranking rewards sparse combos — 3 traded days, all green, outranks 20 days with one red day. Always read the Trades and Green Days counts before trusting a top row, and re-test winners on a second date range.
⚙️ Settings
/
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor
XY
ℹ️ How this tab works
Scans the RR×ATR grid (RR 2–15 × ATR 1–8) over all EMA pairs, ranked by your objective. HONESTY RULE: a combo must be net-positive to rank — a high-WR loser can never appear. But note: at RR up to 1:15 the breakeven win-rate is ~6%, so a couple of lucky far-TP hits can crown a combo — treat extreme-RR winners as suspect until they survive a different date range. ALMOST EQUAL: scores your objective on EVERY CALENDAR DAY of the window separately (a day with no trade is a valid $0/0% data point, not excluded). Ranks ZERO-LOSS-DAY combos first, always — then, among the rest, by the lowest total distance of each day from the cross's own average day (a steadiness score: lower means more even, higher means more erratic).
⚙️ Settings
📊 Performance charts
🕯️ Candlestick chart
🕯️ Candlestick chart (Dhaka time) — pick a row above to overlay its EMA cross and signals · click ▲▼ entry or ●✕ exit for TP/SL + trade details · Ctrl+scroll to zoom at cursor