Victor's sweep and IFVG model

XAUUSD · 5m chart · strategy s33_victor_ifvg · 2 variations

After a strong close breaks a swing, wait for price to sweep the nearest highs, then sell when a 5-minute fair value gap is inverted (buys mirrored).

What it is about

A strong candle body closes through a recent swing: that is the setup. Price then often runs back up to take the nearest highs, where late buyers sit, before the real move continues.

The entry comes when that sweep fails: the 5-minute fair value gap left by the push up is closed through. The trade only goes with the Daily and Weekly structure.

Everything is mirrored for buys. Once a set of highs has been swept it is used up, and the model waits for a fresh pattern.

How it trades

Chart & market

Gold. 5-minute candles for the entry; 1-hour or 15-minute candles for the setup, depending on the variation.

Trend

The last closed Daily and Weekly breaks of structure must agree with the trade.

Setup

A setup candle's body closes below the latest swing low (made within the swing age limit). Its low is the reference for the move.

Sweep

Price takes the first or second nearest high above. Taking a third high cancels the setup. Highs up to the sweep are then used up.

Entry

At the close of the 5-minute candle that closes below the first bullish fair value gap above the 50% level of the move, within the sweep's setup candle or the next one. Three or more gaps in a row in the push are not valid. With no gap, a 5-minute break of structure does the same job.

Second chance

If the first sweep fails, a later new high (no more than 2 ATR higher) allows an entry within the next 3 candles.

Exit

Stop 0.30 beyond the sweep. Target 5R, with the stop moved to entry after +2R. Out after 3 days of 5-minute candles at the latest. One trade at a time.

Days skipped

No entries on the day before an FOMC release.

s33-A: any time except 12:00-17:00 New York, about 3-4 trades a month. s33-B: 08:00-12:00 New York, about 1 trade a month.

What to expect

About 1 trade in 4 or 5 wins. Winners are about 5R, losers -1R, and many trades end at breakeven.
Backtest 2018-today (zero cost): s33-A 321 trades, +111.7R, largest drop 26R; s33-B 140 trades, +89.2R, largest drop 10R.
Losing streaks of 9-11 trades happened in the backtest.

Risks and caveats

Results are mostly a gold result: the same rules on 15 other markets were no better than chance, apart from NAS100 and USDCAD candidates.
The settings were chosen with 2024-26 visible, so the live accounts are the real test.
On s33-A, the direction filter did not beat random direction in 2018-23. On s33-B, 2020 alone made about half of the 2018-23 result and sells lost in 2024-26.

Where it comes from

Victor's own model, from 20 chart screenshots of his trades (2026-10-29). The code reproduces his marked trades and his marked no-trade. The Pine indicator on TradingView uses the same rules.

Main variation: s33-B ★ Main

Live result

0.0R

0 closed · 0 open · zero cost

Backtest total

+89.2R

140 trades · +0.64R a trade · win 26%

Backtest max drawdown

−10.0R

total ÷ drawdown 8.9

Years won

8/9

1.3 trades / month

Backtest R per year (zero cost), s33-B

The variations

Each variation is the same strategy with a different approach, trading its own paper account. Click one to see exactly what it changes.

s33-BGold, 15-minute setups, New York morning★ Main

15-minute setups with 5-minute entries, only 08:00-12:00 New York. Swings stay valid for 8 hours. Target 5R, breakeven at +2R, no entries the day before FOMC.

s33-AGold, 1-hour setups

1-hour setups with 5-minute entries, any time except 12:00-17:00 New York. Target 5R, breakeven at +2R, no entries the day before FOMC.

Variations side by side

Live result of all 2 together: 0.0R (zero cost).

			Live paper account (R, zero cost)				Backtest 2018 – today (R, zero cost)

s33-B★ MainVictor's sweep and IFVG model	XAUUSD	5m	–	0	–	–	+89.2R	+0.64R	8/9	1.3	26%	−10.0R
s33-AVictor's sweep and IFVG model	XAUUSD	5m	–	0	–	–	+111.7R	+0.35R	6/9	3	20%	−26.0R

Main = the variation with the most backtest R per R of max drawdown (2018 – today, zero cost). It is picked on that history, so the live accounts are the real test.