GDI · GOLD DATA INTELLIGENCE

How GDI works

GDI looks for the times of day when gold keeps doing the same thing after it breaks a key level. It tests every candle time on 7 timeframes, proves each idea on old AND new data and against luck, then watches the winners live and learns again every 5 minutes. It only shows setups — it never places a trade.

Open GDISee the 6 steps
the whole process

From years of candles to one clear setup

Tap a step to see exactly what happens in it.

what it analyses

7 timeframes · 5 levels · every candle time

GDI reads XAUUSD from your broker's MT5 (the agent on the VPS) on seven timeframes. Faster timeframes use recent years, slower ones go back decades — as far as the broker's history goes.

TimeframeStyleAnalysedCandle times / dayTrade can run
the 5 levels
1Previous candle high / lowthe candle just before — on every timeframe
2Yesterday's high / lowM1 → H4
3Asian session rangehigh / low of 04:30 – 11:30 IST · M1 → H1
4London opening rangehigh / low of 13:30 – 14:30 IST · M1 → H1
5New York opening rangehigh / low of 20:00 – 21:00 IST · M1 → M30
2,183candle times / day
all timeframes
× 5levels× 2breaks
up · down
× 2ways to trade
follow · reverse
× 9stop / target pairs
0.5 · 0.8 · 1.2 ATR × 1 · 1.5 · 2 R
= 0ideas scanned
every analysis
one strategy, step by step

What a GDI strategy looks like

An illustration (not real prices) of a strategy like "03:30 IST · M5 · close above the previous candle high → SELL". Press play.

step 0 / 5
1 · THE TIMEOnly the M5 candle that opens at 03:30 IST is checked. Other candles are ignored.
2 · THE LEVELThe level is the previous candle's high (yellow line).
3 · THE BREAKThe 03:30 candle CLOSES above that high. A close is needed — a wick alone is not a break.
4 · THE ENTRYSELL at the open of the next candle (a market order). Stop 0.8 × ATR above, TP1 1 R and TP2 3.1 R below.
5 · THE RESULTFollowed until TP1 / TP2, the stop, or the time limit. A candle that touches both stop and target counts as a LOSS.
how a strategy is proven

Passing twice — then the luck check

Scanning hundreds of thousands of ideas will always find some that look good by chance. GDI throws almost all of them away.

Every idea scannedcandle time × level × break × trade way × stop/target—
Enough past tradesat least 40 trades, and at least 15 in EACH half of the historyreal tests
Good in the older half AND the newer halfwin rate ≥ 55 % in both halves · average result above +0.05 R in both · spread includedstrong
WATCHINGgood in both halves, but not yet proven against luck · no setups — tracked as paper entriespaper only
READYalso passes the false-discovery check (Benjamini–Hochberg, 10 %): unlikely to be luck even after so many testssends setups
older half of the historyLearnIs the win rate clearly above break-even? (break-even for TP1 = R × stop is 1 ÷ (1 + R))
newer half of the historyConfirmDoes it STILL work on the newer data? A pattern that only worked long ago is dropped.
every day

Live all day · learning every 5 minutes

every 5 secondsLive price

The agent sends the gold price and the forming candles to the GDI page.

at each strategy's timeThe check

When that candle closes, GDI checks the level: READY + break → setup · WATCHING + break → paper entry · no break → logged.

fundamental filterNews

No setup while high-impact USD news is out (the agent's calendar).

every 5 minutesQuick analysis

The newest candles are added and every idea is tested again: strategies are added or retired as the data changes. The page shows “ANALYSING NOW” while it runs.

02:00 ISTFull re-check

Once a night the whole history is checked again (missing years are asked from the broker again). The live record since a strategy was found is its real test.

HONEST LIMITS · GDI is research, not a promise. Win rates come from past data and spread is included, but slippage, news spikes and changing markets can make live results different. How far back GDI can look depends on your broker's MT5 history — the GDI page shows the exact years it really has. GDI shows setups only and never places a trade. Not financial advice.