THE 764 METHOD
764
THE 764 METHOD
19 Aug 2026
SAST · chart time
The day tape
conflictedevent risk elevatedwindow 07:00-09:00
⚠ Euro area CPI y/y 08:00. …⛔ FOMC Meeting Minutes 20:0…

Gold went up 1747 pips today and this desk did not have a single setup. Both of those sentences are true and neither is an excuse.

The day, in order:

  • Overnight low 4324.73 at three in the morning, ten dollars under yesterday's close.
  • A box between 4324.73 and 4363.22 that held all morning, with 4363.2 tapped five times inside four pips.
  • The box broke at 12:15. Price retested 4363.2 twice, at 13:30 and 14:00, and held both.
  • At 14:30 one fifteen minute bar ran 521 pips.
  • By 15:23 it was up 1021 pips from the two o'clock low. By the close it had reached 4499.41.
  • Day open 4334.33, high 4499.41, low 4324.73, last 4485.

The morning call was that the draw was up before it was down, that the unfilled gap at 4390.14 to 4392.39 was where price was headed, and that no short existed until price got there and turned. The first two were right. The third never happened, because price did not turn. It went through the gap and kept going.

That is a trend day, and a trend day produces nothing for a method built on fibbing a leg and selling the retracement into it. Zero on a day like this is the design working, not the design failing. It is worth being blunt about that rather than dressing it up, and equally worth not pretending a thousand pip move was somehow available.

The rule that earned its keep was the one about unfilled gaps overhead. It says do not sell when there is an untraded gap between your entry and your stop, because price is drawn to that gap and your stop is in the way. This morning that rule forbade every short on the board.

One comparison book took exactly the forbidden trade. It sold around 4355 just before ten, with its stop at 4364.68, four dollars above a shelf that had been tapped five times and was plainly visible. That stop went at quarter past twelve, fifty five dollars before the move even started. The other book asked for 4390 to 4404, which was the right neighbourhood, but the candle that arrived there did not stop, and price finished a hundred dollars above the top of its zone.

Neither book has posted since twenty past ten. Nine and a half hours of silence through the largest move of the week, including the run through one book's zones and the other's stop out. That silence is part of the comparison record.

What moved it was rates. The dollar index fell through the afternoon, ten year yields dropped to 4.65, and silver flipped from down 1.2 percent to up almost 2. Oil fell and the volatility index fell, so there was no fear bid anywhere. Falling yields and a falling dollar with no flight to safety is a repricing of what the Fed will do, and the July minutes landed at eight tonight from a meeting that split 9 to 3 on hiking. Four searches through the day found no single headline behind the two thirty timing, and it is more useful to record that than to invent a cause.

Nothing was ordered today and no alert was ever armed. The desk is on observation for the rest of the month.

Levels8 levels · no spot

PWH4449.81prev week high
PDH4436.31prev day high
AsH4362.07asian high
OPEN4334.33daily open
PDL4329.31prev day low
AsL4324.73asian low
SL4324.73session low
PWL4311.10prev week low

Trade decision log

Flat — nothing mapped

No call on the board. The desk is watching, not chasing — most sessions end here, and that is the method working rather than a gap in the record.

Before & after

The morning chart, left un-redrawn

Morning — planned zones, scenarios, invalidation
Morning — planned zones, scenarios, invalidation
End of day — what triggered, what failed, the real move
End of day — what triggered, what failed, the real move

The striking feature

Morning thesis vs market reality

Morning expectation
Actual result
Called the draw UP before it was down at 10:10, naming the unfilled 5m gap at 4390.14 to 4392.39 as the target and the only place a short could begin to exist.
Correct, and then some. Price reached the gap at 14:30, four and a half hours later, having spent the intervening time exactly where the map said it would be. It filled the gap and kept going another 107 dollars.
Refused every short below 4390 on the overhead-imbalance test, on the grounds that a sell with an untraded gap between it and its stop is a sell waiting to be run over.
The rule paid. Benchmark 2 took that exact trade at 09:51, short around 4355 with its stop at 4364.68, and was stopped at 12:15. The desk was flat through a move that would have been a full loss and then some.
Named 4407.20 as the level where the bearish case ends, and said a 30 minute close above it would finish the thesis.
The 14:30 30 minute candle closed at 4438.95, thirty dollars clear. No ambiguity, no argument, and the invalidation was recorded within twenty minutes of it happening.
Flagged the equal highs at 4363.2 as resting liquidity at 10:55, after four taps inside a four pip band, and said compression under a visible level usually resolves through it rather than away from it.
A fifth tap held at 10:45, then it broke at 12:15 and never traded back below on a closing basis. The read of what those equal highs meant was right and it was written down an hour and twenty minutes before the break.
Said macro leaned gently the wrong way for a short, with yields, the dollar and equities all a little lower, and that the two sides roughly cancelled.
That lean was the story and the brief undersold it. All three moved much further in the same direction through the afternoon and the rates repricing is what drove the entire move. The macro read pointed the right way and was given too little weight against the structural read.
Logged a near miss at 13:47: the 12:00 to 13:00 up leg fibs to a long at 4363.13, stop 4358.99, target 4372.46, 2.25R, with the pullback low at 4363.32.
Missed by 1.9 pips, and it would not have qualified anyway because the leg had already traded through its own target before the pullback began. A spent draw is not a setup. Right shape, wrong moment, correctly declined.

End-of-day verdict

What we learned

Morning bias correct?No
Setup valid?No
Driver that mattered most
Rates, and it was not close. The dollar index went from down 0.18 percent at the open to down 0.64, ten year yields from down 0.38 to 4.65, and silver flipped from down 1.21 percent to up 1.88. Oil and the volatility index both finished lower, so there was no risk-off bid anywhere in it. A falling dollar and falling yields with no flight to safety is a repricing of Fed expectations, and the July minutes landed at 20:00 from a meeting that split 9 to 3 on whether to hike. Four searches across the day found no specific headline for the 14:30 timing. Recording that rather than inventing a cause.
Did price respect the zones?
Every level held until the one that mattered stopped holding. 4363.2 was tapped five times inside four pips between 09:35 and 10:45, rejected every time, then broke at 12:15 and was retested twice from above at 13:30 (low 4363.32) and 14:00 (wick to 4362.19, closed back at 4368.38). Both held. The unfilled gap at 4390.14 to 4392.39 was the stated draw from 10:10 and price reached it at 14:30 exactly as mapped. 4407.20, the named invalidation, went in the same candle. After that nothing held: PDH 4436.31, prev week high 4449.81 and 4500 all gave way inside two hours.
Carry forward into tomorrow
Three things. First, the overhead-imbalance rule was the whole day. It forbade every short on the board this morning and it was right. Benchmark 2 took precisely the trade the rule exists to prevent, short 4355 with its stop four dollars above a five-times-tested shelf, and was stopped 55 dollars before the move began. A rule that costs nothing on a quiet day and saves a large loss on a violent one is a rule worth having. Second, calling the direction correctly and having no way to participate is now the third time this has happened in a fortnight, after 08-13 and 08-18. The method fibs a leg and sells the retracement into it, so a leg that never retraces yields nothing, and on a trend day zero is the correct answer rather than a miss. That is settled and should not be relitigated. Third, both benchmark books went silent from 10:20 to the close, nine and a half hours, straight through the largest move of the week including a run through one book's zones and the other's stop out. Neither acknowledged anything. How a book behaves when it is wrong belongs in the comparison record alongside where it puts its zones.

The method

Two anchor prices on the displacement leg, and a fixed fib grid derives the whole order: entry floored at the 0.764 that names the method, stop at 1.05, target a fixed span below. R is a consequence of the grid, never a number chosen after the fact. Every read is timestamped and left un-redrawn.

← all sessions