Fair Value Gap Retest vs Fill Trade: Betting on the Fill Made Money Before the Spread on 3 of 4 Charts, and Lost After It on All 4
Key takeaways
- There are two opposite ways to trade a fair value gap. The retest waits for price to come back into the gap and trades with the original move. The fill trade bets that price will go back and close the gap.
- I ran both, plus a momentum entry on the gap bar, on EURUSD and gold, M5 and M15, with the same stop, target and cost.
- The fill trade made money before the spread on 3 of 4 charts and lost after it on all 4. On EURUSD M5 it made 0.60 pips a trade and paid 1 pip to get in.
- The retest lost on 3 of 4 charts, and on EURUSD M5 it lost more than random entries: -1.14 pips a trade against -1.02 for random.
- The two trades split by market. EURUSD favoured the fill, gold M15 favoured the retest, and neither beat the random 95th percentile on the chart where it did best.
The fair value gap is sold two ways at once. One camp says price comes back to “rebalance” the gap, so you sell a bullish gap and buy it back once it fills. The other says the gap is where smart money will defend, so you wait for the pullback into the gap and buy. Both cannot be right on the same chart.
The gap rule and the data are the same as in fair value gaps tested. Here the three entries go head to head.
What exactly are the three trades?
| Trade | Entry | Direction | The idea behind it |
|---|---|---|---|
| Retest | Close of the first bar that touches the gap and does not close through it, within 48 bars | With the gap | The gap is support or resistance |
| Momentum | Close of the bar that completes the gap | With the gap | A gap shows a strong move |
| Fill | Close of the bar that completes the gap | Against the gap | Price comes back to fill it |
All three exit the same way: stop 1.5 ATR(14), target 3 ATR, close after 48 bars, one trade at a time. Cost 1 pip on EURUSD and $0.30 an ounce on gold, as in spread and swap costs in MT4. Data: 16,926 EURUSD M5 bars from June 3 to August 26, 2026 and 13,736 gold M5 bars from June 16, with M15 built from them.
Which trade made money before the spread?
| Before cost, per trade | EURUSD M5 | EURUSD M15 | Gold M5 | Gold M15 |
|---|---|---|---|---|
| Fill, against the gap | +0.60 | +0.16 | +0.19 | -1.46 |
| Retest, with the gap | -0.14 | +0.07 | -0.42 | +1.42 |
| Momentum, with the gap | -0.26 | -0.65 | -0.56 | +2.39 |
Pips on EURUSD, dollars an ounce on gold.
Before cost the picture is clean. On three charts, betting on the fill beat both ways of trading with the gap. A fast three-candle move on EURUSD M5 gave back enough in the next bars to pay 0.60 pips a trade. Gold M15 went the other way: there the gap pointed to more of the same move, and the momentum entry made $2.39 an ounce.
And after the spread?
| After cost, per trade | EURUSD M5 | EURUSD M15 | Gold M5 | Gold M15 |
|---|---|---|---|---|
| Fill | -0.40 | -0.84 | -0.11 | -1.76 |
| Retest | -1.14 | -0.93 | -0.72 | +1.12 |
| Momentum | -1.26 | -1.65 | -0.86 | +2.09 |
| Random entries, median | -1.02 | -1.27 | -0.58 | -0.36 |
| Random entries, 95th percentile | -0.76 | -0.41 | +0.04 | +2.37 |
| Trades (fill / retest / momentum) | 1,318 / 1,332 / 1,378 | 381 / 349 / 375 | 820 / 727 / 852 | 247 / 217 / 241 |
The spread decides it. The fill trade lost on all four charts once 1 pip or 30 cents was paid. On EURUSD M5 it was the only one of 16 runs that beat the random 95th percentile, -0.40 against -0.76, which says the pull back into the gap is a real effect. It is just smaller than the cost of trading it.
The retest, the entry most courses teach, lost on EURUSD, M5 and M15, and on gold M5. On EURUSD M5 it lost more than the median random entry. It made money only on gold M15, and there random entries with the same exit reached +$2.37 at the 95th percentile, so +$1.12 is inside what luck can do.
Why does the retest lose where the fill wins?
They are two halves of the same move. On EURUSD M5, 67% of gaps were touched again by the very next bar. That touch is the fill trade going right and the retest trade getting its entry. From there, the retest needs the original move to resume, and on EURUSD it usually did not: price kept drifting back.
| Winners after cost | EURUSD M5 | EURUSD M15 | Gold M5 | Gold M15 |
|---|---|---|---|---|
| Fill | 40.4% | 35.2% | 34.4% | 30.0% |
| Retest | 32.7% | 34.7% | 34.8% | 37.8% |
| Momentum | 31.8% | 32.0% | 33.5% | 40.7% |
With a target twice the stop, a system needs about 33% winners plus cost to break even. The fill trade on EURUSD M5 cleared that by 7 points and still lost, because the 1 pip spread is half of a typical EURUSD M5 ATR of 2 pips.
Does a trend filter rescue the retest?
Partly, on gold. Keeping only gaps of 0.5 ATR or more and only retests on the EMA 200 side moved gold M5 from -0.72 to -0.18 and gold M15 from +1.12 to +3.05 an ounce. On EURUSD it made the retest slightly worse, -1.19 on M5 and -1.09 on M15. The same filter question for other entries is in multi-timeframe confluence tested, and the gold M15 result again sat under its random 95th percentile of +$3.87.
Which would I use?
- Neither as a system. 0 of 16 runs passed all three checks.
- On EURUSD, do not chase the gap candle. Momentum entries on the gap bar were the worst line on both EURUSD charts.
- If you trade retests, test them on your own market. The same rule flipped sign between EURUSD and gold, the way the same session filter did in gold scalping sessions.
- Cost first. An edge of 0.60 pips before the spread is a reason to look at your broker’s spread, not at your chart.
How a plain breakout of a recent range did against named patterns is in chart patterns vs range breakout. How to mark the gaps and which settings were tested is in fair value gap settings for MT5.
Frequently asked questions
Should you trade the fair value gap fill or the retest?
Before cost the fill did better on 3 of 4 charts. After cost both lost on EURUSD and gold M5, and only the retest made money on gold M15, within random luck.
Why do fair value gaps get filled so quickly?
Most are small. On EURUSD M5 the median gap was 1.3 pips and two thirds were touched by the next bar.
Is the FVG retest entry better than entering on the gap candle?
On EURUSD M5 and M15 and gold M5 yes, it lost less. On gold M15 the gap-candle entry made more, $2.09 against $1.12 an ounce.
Does an EMA 200 filter help FVG trades?
On gold it cut losses or raised profit; on EURUSD it made the retest slightly worse.
How much do FVG trades lose to the spread?
1 pip a trade on EURUSD, which turned a +0.60 pip fill trade into -0.40 on M5.
Looking for a cleaner way to read the trend?
It is my own enhanced DeMARK Trend Line indicator for MetaTrader 4 and 5. Non repaint, clean, and free.
Forex educator and indicator developer. I build and trade my own MetaTrader tools, and share the ones that genuinely help.