Study · Cross-market · breakout

Does One More Candle Improve an Opening-Range Breakout?

Instrument
Cross-market
Theme
breakout
Data window
S&P 500, gold and EUR/USD CFDs · May 1, 2025–October 20, 2026 · 1,188 simulated records across 12 configurations · exploratory, before execution and financing costs
Published
Reading time
8 min

Question & scope

We compared a close beyond the New York opening range with waiting for the next M15 close to hold outside it. Across S&P 500, gold and EUR/USD, the extra candle cut filled trade records by 36% without demonstrating a general improvement in gross price returns.

Key finding · Historical sample

Waiting one more M15 candle reduced filled trade records from 724 to 464 across six market/direction pairs. Four gross mean-return comparisons worsened and two improved, but all six approximate uncertainty intervals included zero. EUR/USD shorts were positive with either entry rule; the extra confirmation added only 0.004R per trade.

An extra confirmation candle reduced filled trade records

Instrument and entry rule, both directions counted separately · Simulated trade records

An extra confirmation candle reduced filled trade records — numeric data
Instrument and entry rule, both directions counted separatelySimulated trade records
S&P 500: close-cross259
S&P 500: extra close182
Gold: close-cross213
Gold: extra close130
EUR/USD: close-cross252
EUR/USD: extra close152

Waiting for confirmation feels safer. A breakout closes beyond the opening range, then another candle closes outside it. Surely that second close removes enough false starts to improve the trade.

We tested that specific idea on S&P 500, gold and EUR/USD CFDs. The extra candle reduced activity substantially. It didn't demonstrate a general improvement in returns. On EUR/USD shorts, the strongest directional candidate in this comparison, the simpler entry did almost as well.

These are exploratory results from native strategy replay, measured before execution and financing costs. They support a narrower conclusion than “confirmation works” or “confirmation fails”: another close is a restriction that needs to earn its place.

The question our earlier studies left open

Our S&P 500 session comparison and gold opening-range study measured how breaks resolve. Those findings don't establish whether delaying an entry improves the resulting trade. A cleaner-looking signal can arrive at a worse price or leave less time for a target.

This study changes one entry rule while keeping the range, stop, target and exit instructions the same.

What we tested

The opening range contains the four M15 candles opening from 09:30 through 10:15 in America/New_York time. Its high and low are available only after that hour completes. The clock follows daylight saving time. The anchor is the US stock market's 09:30 local opening time; it is a common comparison window for the three CFDs, not a claim that gold or currencies open then.

The baseline requires an M15 close-cross above the range high for a long, or below its low for a short. The alternative requires the very next M15 close to remain beyond the same edge. If it fails to confirm on that next bar, the attempt expires; a later qualifying cross can start another attempt.

Both versions use market entries after the qualifying close, an unbuffered stop at the opposite opening-range edge, and a target one initial stop distance from entry. Entries use source candles opening from 10:30 until 12:00 local time. The latest eligible source candle closes at 12:15. Each direction can fill at most once per UTC day, with overlapping positions for the same strategy disallowed. Longs and shorts run in separate accounts.

A time-based exit watches the 15:45–16:00 local window. It needs an observed candle: when holiday or shortened-session data omits that window, the bracket remains active. Twenty-four of the 1,188 records closed after 16:00 New York time, including thirteen on a later New York calendar date. These remain in the results, including any realized price gap. The test does not guarantee every position is flat at the cash close.

We compared complete entry dates from May 1, 2025 through October 20, 2026. FMA's preliminary test starts at the newest half of each instrument's available history, so gold and EUR/USD initially covered longer periods. The tables below use the common-date subset of those native trade records. They are not newly replayed accounts starting on May 1, and the period is not an untouched validation sample.

The extra candle mostly reduced activity

One R is the distance between the actual entry and initial stop. The figures below measure signed entry-to-exit price R. Profit factor uses those R outcomes, giving each trade equal initial-risk weight; it isn't the platform's dollar profit factor or an account return.

Market and directionClose-cross recordsMean price RExtra-close recordsMean price R
S&P 500 long138-0.052103-0.104
S&P 500 short121+0.01279-0.072
Gold long117-0.00674-0.017
Gold short96-0.02156+0.064
EUR/USD long122-0.06970-0.082
EUR/USD short130+0.09182+0.094

The record count fell from 724 to 464, a 36% reduction. These totals count separate configurations, which can trade the same market day. They are not 1,188 independent opportunities or a combined portfolio. Entry-step completions exceeded filled trades, so these tables describe executed simulated records rather than the success rate of every breakout signal.

The extra close lowered the gross mean in four comparisons. Gold shorts improved most, from slightly negative to +0.064R, but the confirmation version contained only 56 trades. EUR/USD shorts improved by just 0.004R while producing 48 fewer records.

The apparent improvements are uncertain

We resampled trading days 2,000 times within each market/direction pair, preserving the shared days and each version's different trade selection. This estimates the difference between the two policies' mean price R per executed trade. It doesn't estimate the benefit of delaying the same individual entry.

Market and directionExtra-close minus close-cross mean RApproximate 95% interval
S&P 500 long-0.052-0.127 to +0.024
S&P 500 short-0.084-0.232 to +0.058
Gold long-0.011-0.107 to +0.091
Gold short+0.085-0.051 to +0.232
EUR/USD long-0.013-0.122 to +0.102
EUR/USD short+0.004-0.103 to +0.109

Every interval includes zero. That doesn't prove the rules are equivalent. It means this comparison doesn't separate the apparent differences clearly from sampling variation. The intervals are exploratory, assume days can be resampled independently, and aren't adjusted for choosing among six comparisons. They don't establish a validated edge.

EUR/USD shorts are the better follow-up question

The close-cross EUR/USD short averaged +0.098R in May–December 2025 and +0.085R in January–October 20, 2026. The extra-close version moved from +0.016R to +0.158R across those periods. Its overall mean was only slightly higher, with fewer trades and more variation between periods.

That makes the directional setup worth checking on fresh data. It offers little evidence that the extra confirmation is the valuable part. We haven't tested whether this reflects a broader short bias, the opening-range condition itself, or another feature of the period. A matched simple short-exposure control belongs in the next study.

Gold's confirmation shorts were positive in both subperiods, at +0.084R and +0.048R, but their small sample and uncertain advantage over the baseline limit the conclusion. S&P 500's close-cross shorts moved from +0.045R to -0.015R. A small positive full-period average hides that deterioration.

These splits are stability checks on examined data, not separate out-of-sample validations.

Costs can absorb a small gross advantage

The historical candle sample carried zero quoted spreads. Price R also excludes commission, slippage, swap and financing. We therefore charged a hypothetical additional round-trip cost to every record, rather than treating the screen as a live-execution estimate.

Positive gross candidateGross mean RAfter 0.02R per tradeAfter 0.05R per trade
S&P 500 short, close-cross+0.012-0.008-0.038
Gold short, extra close+0.064+0.044+0.014
EUR/USD short, close-cross+0.091+0.071+0.041
EUR/USD short, extra close+0.094+0.074+0.044

These are sensitivity scenarios, not calibrated broker costs. A 0.05R cost means 5% of that trade's initial stop risk. Actual cost in R varies with the spread, instrument and stop width. Positive rows under either scenario still need realistic execution assumptions and fresh validation.

What the test establishes

The saved strategy definitions were read back and replayed through FMA's native engine. All twelve reconstructed definitions reproduced the authored trade records exactly. A separate candle check verified opening-range stops and applicable next-bar timing on 46 records across all twelve configurations. Earlier screens with an incorrect reference-candle stop were rejected and excluded from every table above.

That verifies important mechanics. It doesn't resolve the unobserved path inside each M15 candle, establish complete historical data quality, or turn the common-date subset into an independent validation. The simulation uses the engine's synthetic intrabar path. Holiday exits and financing also need explicit treatment before an executable system is assessed.

For the next study, we'd preserve the simpler EUR/USD short rule, compare it with a matched short-exposure control, and test a genuinely unexamined period with measured costs. Adding another confirmation filter now would give us more parameters before we've established what the existing result depends on.