See the results.Understand the numbers.
Published XAUUSD signals on M5–H1. Wins, stops and early exits, with the counting rules beside the results.
Updated 2 Nov, 13:33 ICT · 444 scored signals · 14 Aug 2026 → 2 Nov 2026
The win rate above is measured the way every backtest on this site was measured: the trade is held until price reaches the published first target or the published stop. The flip exits are counted separately, below, with what they cost.
Headline: XAUUSD on M5, M15, M30, H1 — the market and timeframes the site's signal claims are about. Closed rows outside it: 234 — 223 on the UNSCORED markets (BTCUSD, ETHUSD) and 11 on gold's H4/D1 context charts. They are listed, labelled, in the tables and the ledger below, and are not counted in these numbers.
Data source: outcomes are graded on market prices, not broker fills — gold on the MT5 bridge when it is fresh, otherwise Binance PAXG or Twelve Data; crypto on Binance.
The flip exits, and what they cost
When the engine flips direction, the open trade is closed at the price the reversal printed instead of being left to run to TP or SL. No backtest counts those trades — this ledger does. They are the whole difference between the two win rates on this page, so here is the arithmetic:
- TP and stop rows+15.2R
- 17 flip exits · 5 green, 12 red-6.3R
- Net+8.9R
Counted as wins and losses, the flip exits take the win rate to 52.7% over 129 closed results. Both numbers are true; they answer different questions.
Median flip exit -0.49R, worst -1.38R — these are real costs, not rounding.
Closed-signal ledger
Every closed trade in this cohort, newest first, with the prices it was published at. Open trades are listed once they close.
| Closed | Market | Timeframe | Side | Confidence | Entry | Stop | Target | Result | R |
|---|---|---|---|---|---|---|---|---|---|
| 27 Oct, 00:00 | XAUUSD | M30 | SELL | 68% | 3,992.88 | 4,028.21 | 3,939.88 | Stop | -1.00 |
| 21 Oct, 20:35 | XAUUSD | M5 | BUY | 68% | 4,138.78 | 4,132.85 | 4,144.71 | Stop | -1.00 |
| 16 Oct, 20:50 | XAUUSD | M5 | BUY | 46% | 4,172.99 | 4,165.57 | 4,180.81 | Final target hit | +1.05 |
| 7 Oct, 17:00 | XAUUSD | H1 | SELL | 68% | 4,128.51 | 4,140.80 | 4,116.21 | Final target hit | +1.00 |
| 6 Oct, 07:30 | XAUUSD | M30 | SELL | 68% | 4,127.75 | 4,151.40 | 4,092.27 | Stop | -1.00 |
| 2 Oct, 21:25 | XAUUSD | M5 | SELL | 68% | 4,190.71 | 4,204.69 | 4,176.58 | Final target hit | +1.01 |
| 1 Oct, 20:35 | XAUUSD | M5 | SELL | 68% | 4,171.06 | 4,177.81 | 4,164.31 | Final target hit | +1.00 |
| 1 Oct, 10:00 | XAUUSD | H1 | SELL | 68% | 4,151.64 | 4,167.79 | 4,135.49 | Stop | -1.00 |
| 30 Sep, 17:50 | XAUUSD | M5 | SELL | 68% | 4,188.59 | 4,194.39 | 4,182.54 | TP2 hit | +1.04 |
| 30 Sep, 13:50 | XAUUSD | M5 | BUY | 68% | 4,181.85 | 4,175.46 | 4,188.24 | Final target hit | +1.00 |
| 30 Sep, 12:15 | XAUUSD | M5 | SELL | 68% | 4,172.65 | 4,177.78 | 4,167.53 | TP2 hit | +1.00 |
| 30 Sep, 10:30 | XAUUSD | M5 | BUY | 68% | 4,182.47 | 4,176.57 | 4,188.37 | Stop | -1.00 |
| 30 Sep, 08:10 | XAUUSD | M5 | SELL | 68% | 4,177.64 | 4,182.23 | 4,173.05 | Final target hit | +1.00 |
| 30 Sep, 01:15 | XAUUSD | M5 | BUY | 68% | 4,162.96 | 4,155.83 | 4,170.08 | TP1 hit | +1.00 |
| 29 Sep, 22:45 | XAUUSD | M5 | SELL | 68% | 4,152.55 | 4,161.13 | 4,143.97 | Stop | -1.00 |
Confidence is the engine's agreement score at the flip, not a win chance. It can only take a few values: 68% means all three checks (SuperTrend, RSI, MACD) agreed on a fresh flip, which is what most published signals look like — so a long run of 68% is expected.
Compare periods, markets and timeframes
Published by the desk
56.3% · 112 closed
The signals that passed the desk's own rules — the record of the service. The headline above counts these on XAUUSD, M5 to H1.
Every scored flip
49.5% · 420 closed
Every closed-candle flip the engine scored on XAUUSD, M5 to H1, including the ones its own gates refused. The record of the engine, and always the weaker number.
Sent on Telegram
53.3% · 30 closed
The XAUUSD M5–H1 rows that actually left the building as a message. The smallest sample on the page.
Month by month
| Month | Closed | Wins | Stops | Win rate | With flip exits | Net R |
|---|---|---|---|---|---|---|
| Aug 2026 | 53 | 30 | 23 | 56.6% | 56.1% · 57 | +6.7R |
| Sep 2026 | 51 | 29 | 22 | 56.9% | 50.0% · 64 | +2.2R |
| Oct 2026 | 8 | 4 | 4 | 8 closed — under the 20-trade floor | — · 8 | +0.1R |
By market
| Market | Closed | Wins | Stops | Win rate | With flip exits | Net R |
|---|---|---|---|---|---|---|
| XAUUSD | 112 | 63 | 49 | 56.3% | 52.7% · 129 | +8.9R |
| BTCUSDUNSCORED | 61 | 43 | 18 | 70.5% | 43.4% · 106 | +13.9R |
| ETHUSDUNSCORED | 70 | 48 | 22 | 68.6% | 45.5% · 112 | +12.9R |
Only XAUUSD is scored: its rules were measured on five years of broker candles. BTCUSD and ETHUSD are read by the same engine and listed here and in the ledger, but are UNSCORED and not counted in the headline — their numbers are arithmetic, not a measured edge.
By timeframe
| Timeframe | Closed | Wins | Stops | Win rate | With flip exits | Net R |
|---|---|---|---|---|---|---|
| M5 | 62 | 33 | 29 | 53.2% | 50.8% · 65 | +1.4R |
| M15 | 15 | 10 | 5 | 15 closed — under the 20-trade floor | — · 16 | +4.7R |
| M30 | 19 | 12 | 7 | 19 closed — under the 20-trade floor | 54.8% · 31 | +2.2R |
| H1 | 16 | 8 | 8 | 16 closed — under the 20-trade floor | — · 17 | +0.7R |
| H4CONTEXT | 7 | 2 | 5 | 7 closed — under the 20-trade floor | — · 9 | -3.3R |
| D1CONTEXT | 1 | 1 | 0 | 1 closed — under the 20-trade floor | — · 2 | +1.0R |
H4 and D1 are context timeframes. The engine scores them and its rules passed these rows, so they sit in the desk lane — but no H4 or D1 signal is sent to members or traded by the bot, so they are not in the headline.
How these results are counted
A signal is written down when the verdict changes — entry, stop and targets exactly as the panel and the alert showed them — before anybody knows the outcome. Nothing can be added afterwards, because scoring only ever reads candles stamped after the signal.
It is then scored against those candles: whichever of the published stop or the published first target the market touched first. A candle that covers both is recorded as undecided, never as a win — a candle carries no path, and guessing would inflate this page. A signal that reaches neither within 96 bars expires and stays in the count as such.
Undecided and expired rows are shown beside every rate and inside none of them. No row is ever removed: there is no code path on this page that can drop a trade for losing, and the cohorts are decided by how a signal was published, before the market answered.
Your own Terminal reads are not in these figures. This page counts the signals the desk published, scored by the engine that published them; a read you ran for yourself is yours, and it never enters the public record in either direction.
Gold is graded on the MT5 bridge when it is fresh, otherwise Binance PAXG or Twelve Data; crypto on Binance. The ledger keeps the most recent 400 signals per chart, so "all-time" means every signal still in it. Past results do not predict future performance.