← FXAbsolute
Trading concepts, measured
Twenty-five lessons · every claim tested against 8,838,373 one-minute candles · gold, GBPUSD, EURUSD, US30, NAS100 · updated 3 September 2026
Order blocks, fair value gaps, liquidity sweeps — these ideas are everywhere on YouTube and almost nowhere is anyone counting. Each lesson here explains the concept in plain language, shows it on a diagram, and then reports what actually happened across five years of real candles. Including the concepts that come out worse than advertised.
Break of structure
The close beyond the last swing high that says the trend is still in charge — and why it is a direction reading, not an entry.
45.6%
of gold breaks followed through
Change of character
The first crack in a trend. Right more often than wrong, and wrong often enough that flipping direction on it is expensive.
42.4%
actually reversed; 28.8% resumed
Order blocks
The quiet candle before the violent move, and the strongest measured result in this whole series.
80.3%
of revisited gold blocks held
Fair value gaps
The band of price three candles skipped. The one popular claim here that survives being measured.
85.6%
of gold gaps filled within two days
Liquidity sweeps
Why your stop gets hit right before price goes your way — and where to put it instead.
52.1%
of sessions take the prior high
Equal highs and lows
Two swings at one price draw a line everyone can see. Treat it as a destination, not a wall.
66.3%
of equal highs taken within 48h
Support and resistance
The oldest claim in technical analysis, tested: levels get weaker with each test, not stronger.
69% → 46.9%
first test versus second test
Supply and demand zones
Quiet candles then a decisive move. The strongest measured result anywhere in this series.
71%
of revisited gold zones held
Flip zones
Price almost always returns to a broken level. Whether it then holds is close to a coin flip.
85.6% / 55.9%
returned, then held
Round numbers
Tested against a control at deliberately ordinary prices. No measurable edge, on any instrument.
-1.5%
difference versus the control
Premium and discount
Buying the cheap half of a swing means a better price and a lower chance the move resumes.
36.9%
of deep pullbacks made a new high
Displacement
A big decisive candle marks where the other concepts are. It does not predict continuation.
50.2% vs 48.9%
decisive versus ordinary candles
Killzones
New York holds the daily extreme at twice chance. London is at or below it on gold and the Dow.
23.2% vs 8.9%
New York versus London, against 12.5%
Judas swing
The first move after the open is supposed to be a fake. It continued more often than it reversed.
54.7%
continued at the London open
Power of three
Accumulation, manipulation, distribution — real in the data, and largely guaranteed by arithmetic.
63.2% vs 37.4%
up days versus all days, low formed early
The silver bullet
One hour, and it doubles its baseline everywhere. It also misses nine days in ten.
8.2%
holds the daily high, against 4.2%
Buy-side and sell-side liquidity
Take yesterday’s low first and the session usually keeps falling. Sweep-and-reverse has it backwards.
31.4%
closed up after taking the prior low
Inducement
The minor level in front of the real one. A stop-placement lesson far more than an entry.
61.6%
of minor levels were taken
Breaker blocks
An order block with a sweep in its past, taught as the premium version. Measured against the plain one, it is not.
65.5% vs 73.7%
breaker versus mitigation block
Mitigation blocks
Where early buyers get out at breakeven. The plainer pattern, and it beat the breaker on all five instruments.
73.7%
held when revisited
Market structure shift
A structural break on a decisive candle. Worth about ten points on four instruments and nothing at all on gold.
69.8% vs 70%
with displacement versus without, on gold
Balanced price range
Two opposing gaps overlapping. One of the few composite ideas that genuinely beat its own simpler version.
+5.4 pts
over a single fair value gap
Trendlines and channels
Tested against a flat line through the same swing. The slope is worth a few points — and far more touches.
26.8% vs 21.3%
sloped versus horizontal
Multi-timeframe analysis
The most repeated advice in trading. It removes about half your trades for a benefit four of five instruments could not show.
+2.5 pts
for trading with the daily trend
Confluence
The assumption under every checklist: more reasons, better odds. Scored zero to four, the line goes down.
-10.3 pts
from no factors to three, size-matched
Why these are different from the usual explanations
Most writing about smart-money concepts has the same shape: here is the pattern, here is a chart where it worked, now go and trade it. The chart always works, because it was chosen afterwards.
These lessons do something different. Each concept is given a precise, written-down definition, and then every occurrence of it in five years of hourly candles is found by rule and followed forward. The definition is printed under every table, because a number without its definition is a claim about the whole concept and would not be true.
That approach produces results in both directions, and they are published unchanged:
Where the folklore holds up
- Supply and demand zones held 71% of the time when price returned, on 1,102 gold zones — the strongest result anywhere here.
- Fair value gaps really do fill, 85.6% within two days, and all five instruments agree closely.
- Order blocks held 80.3% when revisited.
Where it does not
- Levels get weaker, not stronger, with testing. The first test of a gold level held 69%; the second held 46.9%, and it stays near a coin flip after that. This is the opposite of what is taught almost everywhere.
- Round numbers show no measurable edge — reactions at them are within 1.5% of reactions at deliberately ordinary prices. That comparison against a control is the part usually missing.
- Displacement does not predict continuation. Decisive candles followed through 50.2% of the time against 48.9% for ordinary ones.
- A break of structure followed through only 45.6% of the time — informative about direction, but not the confirmation signal it is sold as.
- Broken resistance became support only 55.9% of the time on gold, and less than half the time on GBPUSD.
- The London killzone holds the daily high 8.9% of the time on gold — below the 12.5% a random three hours gets. New York, measured identically, manages 23.2%.
- The Judas swing runs the wrong way. The first move after the open continued 54.7% of the time rather than reversing, at both opens, on all five instruments.
- Sweep-and-reverse is backwards. Sessions taking yesterday's low first closed up only 31.4% of the time — this is not a missing edge, it is an edge pointing the other way.
- The breaker block loses to the plain version. A zone with a liquidity sweep in its history held 65.5% on gold; the identical structure without one held 73.7%, and won on all five instruments.
- Trading with the higher timeframe removes about half your setups for 2.5 points on gold, and two of the five instruments came out slightly negative.
- Stacking reasons made things worse, not better. Matched for zone width, gold zones with no extra factors held 85.7% and zones with three held 75.4%.
Where the answer depends on what you trade
- Displacement on a structural break was worth around ten points on GBPUSD, EURUSD, US30 and NAS100 — and -0.2 points on gold. Same rule, same code, one instrument disagreeing with four.
- The balanced price range did beat a single gap, by 5.4 points on gold, and it is one of the few composite ideas here that beat its own simpler form.
- Trendlines beat a flat line through the same swing on every instrument — 26.8% against 21.3% on gold — and got tested far more often, which matters as much as the hold rate.
Every number here describes the stated definition on hourly candles over 2021-01-03 to 2026-08-30. Your definitions and your timeframe will differ, and so will your results. That is the point of the drill at the end of each lesson.
Reading about a pattern is not learning it
Every lesson ends with the same drill, because it is the only part that actually works: replay real candles one at a time, mark the pattern before you know the outcome, and write down what happened. Thirty repetitions will teach you more than thirty videos.
Open the free backtester →
How to work through them
In this order, if you are starting out.
- Levels first: support and resistance, then supply and demand zones. Everything else is a refinement of "this price mattered", and the difference between a line and a zone shows up clearly in the numbers.
- Then structure: break of structure and change of character. Without swing highs and lows, none of the smart-money concepts can be located on a chart at all.
- Then the things structure creates: order blocks, fair value gaps and displacement. All three are found relative to a break, so they only make sense once structure is readable.
- Then liquidity: sweeps, equal highs and lows and flip zones. These explain why the others work, and they are the ones that most directly change where you put a stop.
- Then the clock: killzones, the silver bullet, the Judas swing and the power of three. Time-based claims are the easiest to check, because a window of a given length has a baseline you can work out on paper. Two of these four beat it and two do not.
- Then the rest of liquidity: buy-side and sell-side and inducement. The first carries the strongest negative result in the series; the second is where most people's stops are sitting without them realising.
- Then the filters: premium and discount and round numbers. One improves the price you pay. The other turns out not to do anything, and finding that out properly is the most transferable lesson here.
- Then the combinations: breakers and mitigation blocks, the market structure shift, the balanced price range. Each of these claims that adding a condition to a simpler idea improves it, and each is measured against that simpler idea rather than against a baseline. Two of the four earned their extra condition.
- Last, the two habits: multi-timeframe analysis and confluence. Read these after the rest, because they are about how to combine everything above — and they are the two results most likely to change how you actually trade. Trendlines sit alongside them as the oldest version of the same question.
The measurement covers 149,382 hourly candles built from 8,838,373 one-minute candles — the same archive the backtester replays, so anything you read here you can go and watch happen.