Does selling the first FVG retest on gold work? Backtest (last 30 days)
The rule, in plain words
A bearish fair value gap (FVG) is the empty space a strong drop leaves behind: the high of a 15-minute candle stays below the low of the candle two places before it. Between 07:00 and 21:00 UTC, the first time the price comes back up and touches a bearish gap formed in the last 20 candles, we sell. The stop goes just above the high of that candle (20 cents above), the target is twice the amount risked, and any trade still open is closed at 20:30 UTC.
Over the last 30 days this rule made money: 23 trades, 52% won, +11.0R.
R = units of risk: +2R means you won twice what you risked on each trade, −1R means you lost what you risked.
Cumulative result (R) trade by trade
Every trade
| Date (UTC) | Side | Exit | Result |
|---|---|---|---|
| Sep 23, 18:52 | Sell | Stop | −1.00R |
| Sep 24, 07:05 | Sell | Stop | −1.00R |
| Sep 25, 12:09 | Sell | Target | +2.00R |
| Sep 28, 07:58 | Sell | Target | +2.00R |
| Sep 29, 15:50 | Sell | Target | +2.00R |
| Sep 30, 11:39 | Sell | Target | +2.00R |
| Oct 01, 07:48 | Sell | Stop | −1.00R |
| Oct 02, 07:13 | Sell | Stop | −1.00R |
| Oct 05, 10:54 | Sell | Stop | −1.00R |
| Oct 06, 10:23 | Sell | Stop | −1.00R |
| Oct 07, 07:35 | Sell | Target | +2.00R |
| Oct 08, 08:03 | Sell | Stop | −1.00R |
| Oct 09, 09:25 | Sell | Target | +2.00R |
| Oct 12, 08:34 | Sell | Target | +2.00R |
| Oct 13, 07:03 | Sell | Target | +2.00R |
| Oct 14, 08:01 | Sell | Stop | −1.00R |
| Oct 15, 07:15 | Sell | Target | +2.00R |
| Oct 16, 10:18 | Sell | Stop | −1.00R |
| Oct 19, 08:16 | Sell | Target | +2.00R |
| Oct 20, 08:03 | Sell | Stop | −1.00R |
| Oct 21, 10:26 | Sell | Stop | −1.00R |
| Oct 22, 08:03 | Sell | Target | +2.00R |
| Oct 23, 14:46 | Sell | Still open | +0.03R |
Read this before trusting the numbers
Both periods we tested were a downtrend: from late August to October 23, 2026 gold fell from about 4,650 to about 4,000. This rule sells, so it worked because it traded with the trend, not against it. The buy version of the same rule did much worse: it lost 10R over the last 30 days and made only +2R over the previous 30. Figures as of October 23, 2026: last 30 days, 22 trades, 50% won, +11.0R; previous 30 days (August 24 to September 23), 22 trades, 41% won, +5.7R. Thirty days is a small sample: the numbers on this page are recalculated every day and can turn negative if the trend changes. They include no spread or commissions. Not financial advice.
Same rule, other direction: Does buying the first FVG retest on gold work? Backtest (last 30 days)
Test it yourself with your AI
Connect TraderWAI to Claude, ChatGPT or any AI and paste this. Change the market, the period or the rule and compare.
Backtest gold for the last 30 days: sell the first retest of a bearish M15 fair value gap between 07:00 and 21:00 UTC, stop just above the retest candle, target 2R, close at 20:30 UTC.Leave it running
If you like a rule, your AI can turn it into a conditional order: TraderWAI checks it every minute on your own broker account, with the stop and the risk per trade you choose, even with the chat closed. You can keep every order waiting for your approval.
Updated automatically every day · last update Oct 23, 2026 · period Sep 23, 2026 – Oct 23, 2026
How it is calculated
1-minute data from the TraderWAI market feed. Entry at the close of the 1-minute candle where the rule is met (the same engine that runs live conditional orders). If one candle touches both the stop and the target, the stop is counted (worst case). No spread, commissions or slippage.
Past results don't guarantee future ones. Not financial advice.
Questions
What is a backtest?
It applies a fixed rule to past prices to see what would have happened: which trades it would have taken and how each one ended. It tells you how the rule behaved, not how it will behave.
What does R mean?
R is the amount you risk on each trade (the distance to the stop). +2R is a win of twice that risk and −1R is a full loss. It lets you compare results without depending on account size.
Can I change the rule?
Yes. Ask your AI for the same backtest with another market, period, stop or target. On the free plan you can test the last 7 days; longer periods are available on paid plans.