Three-bar reversals: what follows the turn?

The question

One bar carries a move. A second pushes to a new extreme. The third turns and closes beyond the far side of both earlier bars. That is the three-bar reversal studied here: a failed push confirmed by a close.

This study follows the pattern in Nasdaq (NQ), S&P 500 (ES), Russell 2000 (RTY), gold (GC), crude oil (CL), and natural gas (NG) futures, across five timeframes. It measures how much room price offered in the reversal direction, then compares that movement with eligible baseline bars from the same market and timeframe.

The question is whether the pattern adds information beyond that market’s background behaviour—and how that answer changes with direction and the size of the move being measured.

A second experiment asks what happens when the pattern becomes a complete trading rule: enter on confirmation, then exit after five hourly bars. That test covers NQ, ES, GC, and Dow (YM) futures in 2020–2024. Its different markets, sessions, and exit policy are kept separate from the pattern statistics.

The findings in brief

The pooled pattern rate exceeded its matched baseline at 1R in all thirty market/timeframe combinations. At 2R, only daily natural gas remained above baseline. At 3R, none of the thirty pooled samples did.

The comparisons use different sessions: NQ, ES, and RTY cover regular trading hours; GC, CL, and NG cover their full exchange sessions. Each result has its own matched baseline, but this is not a controlled comparison of six markets over identical hours.

The five-bar strategy produced a mixed result after modeled commissions: NQ and ES gained, YM finished slightly positive, and GC lost money. All four experienced substantial drawdowns while positions were open. The cost scenarios also show how little room ES and YM had for worse execution. Explore the strategy experiment.

Recognising the pattern

The formation contains exactly three bars, including the reversal bar. Both the candle direction and the confirming close matter.

For a bearish reversal:

1. Bar 1 closes above its open.
2. Bar 2 makes a higher high than bar 1.
3. Bar 3 closes below its own open and below the lows of both earlier bars. Its high must not exceed bar 2’s high.

For a bullish reversal, mirror those rules: bar 1 closes down; bar 2 makes a lower low; bar 3 closes up and above both earlier highs, without undercutting bar 2’s low. Bar 2’s candle colour is unrestricted. A wick through the earlier bars without the required close does not qualify.

These NQ daily examples illustrate the three-bar formation. They do not show the full 20-bar observation window or represent typical outcomes.

The reference price is bar 3’s close. 1R is the distance from that close to bar 2’s low for a bullish reversal, or bar 2’s high for a bearish reversal. The calculation takes the extreme across all three pattern bars; the trigger rules ensure that bar 2 holds that extreme.

“Offered at least 2R” means that price reached twice that distance in the reversal direction before touching the invalidation level, within the observation window. It does not mean a trade captured that movement. No preceding trend, volume threshold, or minimum candle-body size is required for the scan; those would be separate conditions to investigate.

Explore the results

Compare how often price reached 1R, 2R, or 3R after a reversal, then explore the differences by market, timeframe, and direction.

Across the study

Compare markets and timeframes

Compare the pattern with its own market baseline. Select a cell to inspect the full result below.

Both directionsBullish reversalsBearish reversals

Compare room offeredAt least 1RAt least 2RAt least 3R

Pattern rate − baseline rate
Percentage points (pp)

Scroll across to see all five timeframes →

Both directions · at least 1R · n = measured reversals

Market	5 min	15 min	30 min	1 hour	Daily
NQRegular session	+3.0 ppn = 20,529	+3.9 ppn = 7,167	+3.4 ppn = 3,562	+3.7 ppn = 1,646	+7.5 ppn = 353
ESRegular session	+3.7 ppn = 16,129	+4.6 ppn = 6,342	+5.1 ppn = 3,358	+4.9 ppn = 1,594	+4.0 ppn = 344
RTYRegular session	+3.2 ppn = 19,708	+2.9 ppn = 6,838	+3.3 ppn = 3,450	+2.3 ppn = 1,549	+4.6 ppn = 333
GCFull exchange session	+2.5 ppn = 55,831	+2.8 ppn = 20,381	+1.3 ppn = 10,778	+2.0 ppn = 5,554	+8.1 ppn = 313
CLFull exchange session	+2.5 ppn = 52,805	+2.9 ppn = 20,704	+2.8 ppn = 11,069	+2.4 ppn = 5,967	+6.1 ppn = 296
NGFull exchange session	+1.6 ppn = 36,912	+2.8 ppn = 14,844	+2.6 ppn = 8,086	+2.6 ppn = 4,546	+9.7 ppn = 352

NQ, ES, and RTY use regular-session bars; GC, CL, and NG use full exchange-session bars. Each cell has its own matched baseline. These differences are descriptive, not a ranking of trading returns.

Inside the selected sample

InstrumentNQ — Nasdaq futuresES — S&P 500 futuresRTY — Russell 2000 futuresGC — Gold futuresCL — Crude oil futuresNG — Natural gas futuresTimeframe5 min15 min30 min1 hourDaily

NQ · 1 hour · Both directions

2 January 2008 – 30 April 2026
Regular session

Showing NQ · 1 hour · Both directions: 1,646 measured reversals. Overview compares at least 1R.

Measured reversals

1,646

Offered at least 1R

46.5%

1R versus baseline

+3.7 pp

Room offered before invalidationShare of observations

Pattern Baseline

At least 1R

Pattern: 46.5%

Baseline: 42.8%

At least 2R

Pattern: 26.1%

Baseline: 29.4%

At least 3R

Pattern: 14.3%

Baseline: 21.6%

NQ · 1 hour · Both directions · pattern and matched baseline

Room offered	Pattern	Baseline	Difference
At least 1R	46.5%	42.8%	+3.7 pp
At least 2R	26.1%	29.4%	−3.3 pp
At least 3R	14.3%	21.6%	−7.2 pp

The baseline combines 55,447 direction-specific observations, weighted to match the pattern’s bullish/bearish mix. Observation windows can overlap. The daily samples are much smaller than the intraday samples.

A positive difference means the pattern reached that excursion more often than its matched baseline. Percentage points are differences between rates, not percentage investment returns.

What changes across the study

Index futures: direction matters, but larger excursions weaken

At 1R, bearish reversals had a larger difference from baseline than bullish reversals across all five NQ and RTY timeframes. The NQ hourly result illustrates the split: the bullish difference was +2.9 pp, while the bearish difference was +4.4 pp. ES was less uniform: its 30-minute bullish difference was slightly larger than its bearish difference.

Every pooled index-futures sample was below baseline at both 2R and 3R. A few bearish samples exceeded baseline at 2R—ES on 30-minute and hourly bars, and RTY on daily bars—but none of the bullish or bearish index samples exceeded baseline at 3R. The stronger 1R comparison therefore does not establish a short-selling strategy or a general advantage at larger targets.

Gold: a daily direction split within a weaker 2R comparison

Both GC directions exceeded baseline at 1R on all five timeframes. Every pooled GC sample was below baseline at 2R and 3R.

The daily 2R result shows what pooling can hide. The bullish difference was +1.4 pp, while the bearish difference was −1.7 pp. The combined rate was slightly below baseline. The bullish daily comparison rests on only 143 measured reversals.

Energy: crude oil follows the broad result; daily natural gas differs

CL exceeded its pooled baseline at 1R on all five timeframes and fell below it at both 2R and 3R. Both individual directions followed that same positive-at-1R, negative-at-2R-and-3R pattern throughout the crude-oil samples.

NG’s intraday pooled results followed the same pattern. Daily natural gas was the only pooled sample in the whole study above baseline at 2R, with a difference of +1.1 pp. The bullish daily sample was above baseline at 2R; the bearish daily sample was below it. Neither direction exceeded baseline at 3R.

That exception is a research question, not a reason to rank NG as the best market for the pattern. It comes from 352 daily reversals, selected from a much larger set of comparisons.

Daily observations are a smaller sample

Daily results contain hundreds of measured reversals, compared with thousands or tens of thousands in the intraday samples. Splitting by direction reduces those counts further: the smallest directional sample is 127 bearish CL daily reversals. The same 20-bar horizon also covers a much longer elapsed period on a daily chart than on a 5-minute chart.

GC and the energy contracts include more hours per session than the index futures. Their larger intraday counts should not be read as a higher pattern frequency under equal exposure. Small differences and apparent rankings need further investigation before being treated as stable properties of a market or timeframe.

Pattern methodology and coverage

• Observation period: the history runs from early 2008 through April 2026; start dates differ, with NG beginning in February. Exact dates and counts accompany each result. Analysis completed on 22 September 2026 (UTC).
• Sessions: NQ, ES, and RTY are filtered to 09:30–16:00 New York time. GC, CL, and NG use available full-session data grouped at the exchange’s 17:00 Chicago session boundary. Their daily bars therefore represent a different session from the equity-index daily bars.
• Bar construction: intraday bars group 5, 15, 30, or 60 available minute bars within each session. Trailing partial blocks are dropped and missing minutes are not filled. A block can span more clock time when minutes are missing. Daily bars aggregate the available data in each session. Pattern sequences and forward windows can continue into later sessions.
• Eligible observations: a positive risk distance and a full 20-bar forward window are required. The reported sample is the number measured under those rules, which can differ from the raw number of detected formations.
• Measurement: maximum favourable movement from bar 3’s close before the invalidation level is touched, observed for up to 20 subsequent bars. Movement is capped at 3R in this study. The confirmation bar itself is not part of the forward window.
• Bar ordering: if a bar touches the invalidation level, measurement stops before counting that bar’s favourable extreme. This is a conservative convention where intrabar ordering is unavailable.
• Control: eligible bars from the same series, with the same direction, three-bar extreme risk definition, and observation horizon. Baseline bars do not have to form the reversal, so their risk extreme need not be on the middle bar. The pooled baseline is weighted to match the pattern sample’s bullish/bearish mix. Its count combines direction-specific observations; these are not independent trades.

Trading the pattern: a five-bar hold

Offering room for a move and capturing a trade are different questions. This experiment uses the same three-bar trigger, adds a fixed exit, and measures the resulting profit and loss. It is one specified rule, with no claim that five bars is the best holding period.

Entry

Buy a bullish confirmation; sell a bearish confirmation. One position at a time.

Exit

Close after five subsequent hourly bars. Opposite signals are ignored while holding. No protective stop or profit target.

Position size

One standard contract throughout, starting from $100,000 per market. No compounding or volatility adjustment.

Coverage

NQ, ES, GC, and YM · 1 hour · 2020–2024. Available full exchange sessions, aligned to 17:00 Chicago. No regular-hours filter.

Simulated fills

Entry at the next available minute bar’s open after confirmation; timed exit at the latest minute close. Any remaining position is closed at the end of the run.

Costs

$3.90 round trip for NQ, ES, and YM; $4.34 for GC. These are model assumptions, not verified current broker fees. The base run has no added slippage; separate scenarios increase costs.

The strategy’s hourly bars use clock-aligned exchange sessions built from available minute data, including partial bars. Positions can span session breaks; five bars need not mean five uninterrupted clock hours. After a timed exit, the strategy waits for a later confirmation before entering again.

The pattern’s invalidation level is not a stop in this strategy. The earlier study observes up to 20 bars and ends measurement at invalidation. This test holds for five bars regardless of that level. NQ and ES also use broader sessions here, and YM appears only in the strategy test. The two sets of results therefore do not measure the same outcomes.

A separate execution experiment

Four markets. One exit rule.

One contract, long and short, held for five hourly bars. Each market is a separate simulation with $100,000 starting capital.

2020–2024 · after modeled commissions · USD rounded

Market	Net P&L	Profit factor	Max drawdown¹
NQ	$110,037	1.12	$45,798
ES	$52,640	1.10	$39,347
GC	−$49,440	0.89	$72,852
YM	$6,104	1.02	$44,047

¹ Mark-to-market drawdown measured at minute closes, including open positions. One standard contract has different exposure in each market.

Strategy instrumentNQ — Nasdaq futuresES — S&P 500 futuresGC — Gold futuresYM — Dow futures1 hour · Long and short
2020–2024

NQ · Five-bar hold

1 January 2020 – 31 December 2024
Full exchange sessions · One contract · No stop or target

Showing NQ strategy: net P&L $110,037, 1,488 trades, maximum minute-close drawdown $45,798. View: performance.

Net P&L

$110,037

Closed trades

1,488

Profit factor

1.12

Minute-close drawdown

$45,798

Minute-close drawdown · %

37.9%

Total commissions

$5,803

Dollar figures above are rounded. Profit factor divides total net winning-trade profit by total net losing-trade loss. Dollar and percentage drawdown maxima can occur at different points.

PerformanceMonthly P&LLong & shortCostsEntry control

Performance & risk

What an average trade looked like

Win rate

50.87%

757 of 1,488 trades

Average winning trade

$1,353.32

After modeled commissions

Average loss / breakeven

−$1,250.92

731 non-winning trades

Average net P&L / trade

$73.95

Historical expectancy

A win has positive net P&L; losses and exact breakevens form the non-winning group. These are historical averages after commissions, before any added slippage—not a forecast or a stop/target ratio.

Equity while positions are openUSD

NQ · two views of drawdown

Measurement	Maximum USD	Maximum %
At minute closes	$45,798	37.9%
At trade closes	$44,148	36.9%

The maximum drawdown is measured at minute closes; the chart summarizes each session. Neither measurement captures moves between minute closes.

Time below the peak

Peak to trough

310 days

Worst dollar drawdown

Peak to recovery

685 days

Includes the decline

Sessions below peak

92.81%

Of all recorded session closes

Durations follow the worst dollar drawdown at minute closes, in whole elapsed days. They do not measure the longest drawdown episode or necessarily the worst percentage decline.

The session percentage uses each session’s closing equity and its running closing peak, including starting capital. It is not the share of losing sessions or of clock time spent underwater.

Inspect the closed-trade curves

Equity at trade closeUSD

Closed-trade drawdownUSD

NQ · annual net P&L, USD · attributed at trade close

Year	Combined
2020	−$24,163.10
2021	$37,358.40
2022	$84,994.80
2023	$26,026.60
2024	−$14,179.90
Full period	$110,036.80

Annual P&L is attributed to trade closes. These are separate fixed-size runs, not a combined portfolio or a risk-adjusted market ranking. The base run has no added slippage allowance.

Historical backtest examples · NQ futures

These hypothetical trades use historical prices and were not executed live. The examples were selected with hindsight and do not represent typical results. P&L includes modeled commissions but no added slippage; live fills, liquidity and costs can differ. Backtests do not predict future performance. For education only, not trading recommendations. Read the risk disclosure.

Anatomy of a trade

From the signal to the exit

A winning trade and a losing trade from the NQ hourly 2020–2024 backtest. Follow the confirmation, entry and five-bar exit.

Choose a historical trade example

Winning exampleLosing example

NQ · 1 hour · Long · 1 contract
Jul 28, 2022

A long trade that follows through

Net trade P&L+$5,196.10After $3.90 commission

Hand-picked for a clearer signal

Scroll the chart to follow the trade →

1. 01
   Wait for confirmation

   Bar 3 closes above both earlier highs, completing the bullish signal at Jul 28, 2022 · 11:00 CT.

2. 02
   Enter on the next minute

   Buy at the next available minute open: 12,664.25 at Jul 28, 2022 · 11:00 CT.

3. 03
   Exit after five bars

   Sell to close at 12,924.25 at Jul 28, 2022 · 16:00 CT, using the fifth held bar’s final minute close. No stop or target.

What to notice

The long trade closes above its entry. The time-based rule exits after the fifth bar; it does not try to capture the highest price in the chart.

Entry → exit

12,664.25 → 12,924.25

Gross trade P&L

+$5,200.00

Commission

$3.90

Net trade P&L

+$5,196.10

Trade details

Each chart is scaled to its price range. Candles are equally spaced, with breaks in time marked //. Faded candles show price action before or after the trade.

The entry uses the minute’s opening price. Its execution record is stamped at that minute’s end: Jul 28, 2022 · 11:01 CT. The exit uses the last minute close and is recorded at Jul 28, 2022 · 16:00 CT. These fills have no additional modeled slippage. All times use America/Chicago.

Hourly prices · index points · candle opening times

Date / time (CT)	Role	Open	High	Low	Close	Minutes of data	Missing hours before
Jul 28, 2022 05:00	Context	12,528.00	12,544.25	12,520.00	12,542.00	60/60	0
Jul 28, 2022 06:00	Context	12,542.25	12,554.75	12,532.25	12,539.25	60/60	0
Jul 28, 2022 07:00	Context	12,539.50	12,626.25	12,496.25	12,579.75	60/60	0
Jul 28, 2022 08:00	Signal 1	12,579.50	12,654.00	12,453.75	12,490.00	60/60	0
Jul 28, 2022 09:00	Signal 2	12,490.00	12,577.50	12,449.00	12,576.25	60/60	0
Jul 28, 2022 10:00	Signal 3	12,576.25	12,695.25	12,572.50	12,664.00	60/60	0
Jul 28, 2022 11:00	Held 1	12,664.25	12,679.50	12,634.50	12,643.50	60/60	0
Jul 28, 2022 12:00	Held 2	12,644.00	12,738.50	12,624.00	12,729.25	60/60	0
Jul 28, 2022 13:00	Held 3	12,729.25	12,742.00	12,691.75	12,736.50	60/60	0
Jul 28, 2022 14:00	Held 4	12,737.25	12,764.50	12,710.50	12,738.75	60/60	0
Jul 28, 2022 15:00	Held 5	12,738.25	12,948.00	12,732.00	12,924.25	60/60	0
Jul 28, 2022 17:00	Context	12,920.25	12,920.25	12,872.50	12,880.00	60/60	0
Jul 28, 2022 18:00	Context	12,880.25	12,894.25	12,870.75	12,893.00	60/60	0
Jul 28, 2022 19:00	Context	12,892.50	12,893.50	12,879.50	12,884.75	60/60	0

NQ · 1 hour · Short · 1 contract
Jan 13, 2020

The same exit, a losing short trade

Net trade P&L−$988.90After $3.90 commission

First losing trade to close

Scroll the chart to follow the trade →

1. 01
   Wait for confirmation

   Bar 3 closes below both earlier lows, completing the bearish signal at Jan 13, 2020 · 06:00 CT.

2. 02
   Enter on the next minute

   Sell at the next available minute open: 9,008.00 at Jan 13, 2020 · 06:00 CT.

3. 03
   Exit after five bars

   Buy to close at 9,057.25 at Jan 13, 2020 · 11:00 CT, using the fifth held bar’s final minute close. No stop or target.

What to notice

Price finishes above the short entry. The five-bar rule closes the position at a loss; a confirmed pattern does not guarantee follow-through.

Entry → exit

9,008.00 → 9,057.25

Gross trade P&L

−$985.00

Commission

$3.90

Net trade P&L

−$988.90

Trade details

Each chart is scaled to its price range. Candles are equally spaced, with breaks in time marked //. Faded candles show price action before or after the trade.

The entry uses the minute’s opening price. Its execution record is stamped at that minute’s end: Jan 13, 2020 · 06:01 CT. The exit uses the last minute close and is recorded at Jan 13, 2020 · 11:00 CT. These fills have no additional modeled slippage. All times use America/Chicago.

Hourly prices · index points · candle opening times

Date / time (CT)	Role	Open	High	Low	Close	Minutes of data	Missing hours before
Jan 13, 2020 00:00	Context	9,013.50	9,017.25	9,007.75	9,008.75	60/60	0
Jan 13, 2020 01:00	Context	9,008.75	9,013.75	9,006.00	9,009.00	60/60	0
Jan 13, 2020 02:00	Context	9,009.00	9,022.50	9,008.75	9,017.25	60/60	0
Jan 13, 2020 03:00	Signal 1	9,017.25	9,025.75	9,012.00	9,025.50	60/60	0
Jan 13, 2020 04:00	Signal 2	9,025.75	9,027.25	9,020.00	9,025.25	60/60	0
Jan 13, 2020 05:00	Signal 3	9,025.75	9,025.75	9,007.25	9,007.75	60/60	0
Jan 13, 2020 06:00	Held 1	9,008.00	9,016.25	9,006.25	9,012.75	60/60	0
Jan 13, 2020 07:00	Held 2	9,012.75	9,024.50	9,012.00	9,015.75	60/60	0
Jan 13, 2020 08:00	Held 3	9,016.00	9,037.75	9,003.75	9,033.00	60/60	0
Jan 13, 2020 09:00	Held 4	9,033.25	9,038.50	9,019.75	9,035.25	60/60	0
Jan 13, 2020 10:00	Held 5	9,035.50	9,058.75	9,031.00	9,057.25	60/60	0
Jan 13, 2020 11:00	Context	9,057.25	9,068.75	9,057.25	9,064.00	60/60	0
Jan 13, 2020 12:00	Context	9,064.00	9,072.50	9,061.25	9,065.25	60/60	0
Jan 13, 2020 13:00	Context	9,065.25	9,077.25	9,061.50	9,072.00	60/60	0

What the trading results add

NQ had the largest net gain in this set, $110,037, alongside a maximum minute-close drawdown of $45,798. It lost money in both 2020 and 2024. A profitable five-year total did not mean a consistently profitable year-to-year experience.

ES also finished positive, but 2022 contributed $51,650 against a full-period gain of $52,640. Its losses in 2020, 2021, and 2024 matter when judging how dependent the overall result is on particular periods.

GC lost $49,440, with four losing calendar years before a positive 2024. Its largest minute-close percentage drawdown was 65.5%. This is a useful counterexample to treating the pattern’s stronger 1R excursion rate as evidence that any chosen exit will be profitable.

YM’s $6,104 net gain came with a $44,047 minute-close maximum drawdown and a profit factor of 1.02. That narrow margin deserves particular scrutiny under less favourable execution assumptions.

The side breakdown adds context: NQ’s long trades contributed more of its gain, while ES’s shorts contributed more. Both GC directions lost money. YM’s short contribution was positive and its long contribution negative. These are the trades taken inside the combined run; they do not establish what separately trading just one direction would have produced.

Wins, losses and time below the peak

The win rates sit close to half the trades, so the size of the wins and losses matters. NQ averaged $1,353.32 on its winning trades and −$1,250.92 on its non-winning trades. Its historical average across all trades was $73.95 after commissions. YM’s average was only $4.20; GC’s was negative. These averages describe this historical sample, not the profit to expect from the next trade.

The monthly view also exposes swings hidden by annual totals. NQ’s September 2024 result was −$26,632.50, followed by $22,671.40 in October. Despite that rebound, the year finished negative. Each month records net P&L at the trade’s closing date in UTC.

Recovery took time even in the two stronger markets. NQ’s worst dollar drawdown took 685 days from peak to recovery; ES took 804 days. GC and YM had not recovered their worst dollar drawdowns by the end of 2024. These durations include the decline and are measured in whole elapsed days. They are not necessarily each market’s longest drawdown episode. All four markets closed more than 90% of their recorded sessions below their running session-close equity peak, which is different from losing money in 90% of sessions.

Costs change the conclusion

NQ remained profitable in all six cost scenarios. ES turned negative with two extra ticks per side at the base commission, ending at −$18,960. YM turned negative with just one extra tick per side, at −$8,436. GC was already negative before adding slippage.

These scenarios charge more to the same executed trades. They show the margin available for costs, but do not recreate order execution, missed fills or margin constraints. The dollar comparisons also use one contract in each market, which does not equalise risk or establish a best market to trade.

A useful reference, with an imperfect match

Against 500 randomized-entry runs, NQ’s net P&L ranked at percentile 93.0, ES at 89.4, YM at 63.2, and GC at 9.8. All four actual results fell inside their reference’s 5th–95th percentile range.

That puts NQ and ES toward the stronger end of these timing experiments, while GC sat toward the weaker end. It does not establish statistical significance or a probability of future profit. The reference preserves direction, size, commission and elapsed holding time, matching local entry time where possible. However, its independently drawn trades can overlap and its exits use minute opens rather than the strategy’s minute closes. It is not an equivalent executable portfolio or an independent out-of-sample test.

Read the risk figures at their actual resolution

The main drawdown figures include open-position P&L marked at each minute close. The equity chart shows session-close equity and each session’s range of minute-close equity. That range cannot recover the exact sequence of intraday peaks and troughs, so the maximum drawdowns come directly from the minute-close measurements. The separate closed-trade curves remain available for comparison.

Neither view captures moves between minute closes. Dollar and percentage drawdown maxima can occur at different times. Percentage drawdowns depend on the $100,000 starting capital; changing that balance would change those percentages even with identical dollar trades.

There is no protective stop or demonstrated live track record here. The additional measurements make the historical result easier to examine, while leaving holding-period sensitivity, independent validation and more realistic execution as open questions.

Limitations and next questions

Observation windows overlap, and the thirty samples share markets and history. They should not be added together as independent evidence. This report makes no claim of an independently validated, out-of-sample edge, and selecting the largest difference from the grid introduces its own selection problem.

The pattern calculation does not deduct trading costs or model fill quality. The separate strategy test includes its stated commissions, with the execution limitations described above. Maximum favourable excursion cannot be captured automatically: choosing an exit changes the outcome being measured. A below-baseline result at 3R does not mean price never reached 3R; it means it did so less often than the matched control.

The backtest summary does not by itself establish data-vendor quality, rollover treatment, or stability across market regimes. Those questions require a separate audit of the source data and a reproducible rerun before drawing stronger conclusions. A session-matched comparison of the index, gold, and energy contracts is also outside this report.

The next useful investigations are whether the differences persist across separate historical periods, how session choice affects them, and how the five-bar rule behaves under more demanding execution assumptions. Those remain research questions rather than conclusions established by these backtests.

For the broader distinction between a chart setup and a strategy, see A setup is only part of a trade.