Research · a claim, tested

Does yesterday's high or low matter?

It is sold as one of the two or three levels that matter on any chart. We found 68,688 occasions across 19 markets where price traded through yesterday's high or low, and asked whether the day then closed beyond it. Answer: 50.4% — against 51% for a level with no story attached.

50.4%of breaks closed beyond yesterday's extreme
51%for a matched arbitrary level — the control
68,688breaks measured
0.6 ptsthe entire effect

The claim

Yesterday's high and low are meant to be decision points: break above and the day belongs to the buyers, break below and it belongs to the sellers. The competing story — the one that has taken over in the last few years — says the opposite: those levels exist to be swept, so a break through them is where the move ends rather than where it begins. Both stories cannot be right, and both are told with charts rather than counts.

What we measured it against

For every day we take the first moment price trades through yesterday's high or low, and record whether the day closed beyond it. Then we do exactly the same thing for a synthetic level the same distance from the same day's open — a level with the same geometry and no history. If yesterday's extreme is special, breaking it should behave differently from breaking a line drawn at the same distance for no reason.

The control is the test. Roughly half of all breaks hold, at any level, because that is what a nearly-random walk does once it has moved. The number that could be evidence is the difference, and it is the number these reports never print.

The result

Pooled across 19 markets: 50.4% of breaks through yesterday's extreme closed beyond it, against 51% for the matched level. That is 0.6 points in the direction the sweep story predicts — the famous level is very slightly less likely to hold than an arbitrary one. No spread survives 0.6 points.

MarketBreaksClosed beyondMatched levelΔMedian distance from open
EUR/USD5,538 51.2%51.5%-0.3 0.181%
GBP/USD4,902 50.8%52.0%-1.2 0.193%
USD/CHF3,037 48.2%48.7%-0.5 0.179%
USD/JPY4,124 50.5%52.1%-1.6 0.184%
AUD/USD5,325 50.2%51.6%-1.4 0.242%
USD/CAD5,631 49.3%50.2%-0.9 0.175%
NZD/USD4,799 51.4%53.0%-1.6 0.268%
EUR/JPY4,779 51.8%51.5%+0.3 0.208%
GBP/JPY3,811 50.4%50.8%-0.4 0.228%
EUR/GBP2,297 48.0%49.0%-1.0 0.158%
Gold (XAU/USD)5,418 50.6%51.0%-0.4 0.361%
Silver (XAG/USD)4,796 39.0%41.6%-2.6 0.895%
US30 (Dow)2,613 50.0%50.4%-0.4 0.256%
NAS1001,645 52.9%51.0%+1.9 0.399%
SPX5001,658 51.9%51.0%+0.9 0.308%
GER40 (DAX)2,366 50.6%51.8%-1.2 0.315%
UK100 (FTSE)897 45.4%48.1%-2.7 0.287%
BTC/USD2,603 48.9%51.0%-2.1 0.986%
ETH/USD2,449 49.6%49.5%+0.1 1.335%

15 of 19 markets agree — and that is weaker than it looks

15 of the 19 markets lean the same way. That reads like 15 independent confirmations and it is not: these markets move together. We measure US30 and SPX500 at 0.94 correlation of hourly returns, NAS100 and SPX500 at 0.93, BTC and ETH at 0.84. Several of the "agreements" in that table are the same observation counted more than once, which is why we lead with the size of the effect rather than with the head-count.

What we would actually say

Full method and the raw counts are in section 6 of the evidence page.

Other reports this category sells, tested

Do weekend gaps fill?"Gaps fill 90% of the time" is true and empty — a matched midweek move fills just as often. The indices do the opposite.Does the opening range set the day?The range breaks on 99.8–100.0% of days. A signal that fires every session has not selected anything.When does the day's high and low actually form?"The high forms at the open" is the boundary you chose, not the market. Move it and every single market changes its answer.Does the first hour set the day?A green first hour tells you the first hour was green. Not one of 17 markets reaches even a weak significance bar.Does the trading day of the month matter?The chart this whole category is built on. 285 slots tested, 0 clear the bar, and the strongest anywhere reaches z = 2.99.The best time to trade forexThe overlap is real — as a movement and cost argument, not a direction one. The half that survives a holdout is WHEN, not which way.Does Bitcoin move on the weekend?The weekend is an ordinary trading period whose timetable is a different timetable — the biggest hour moves from 14:00 to 00:00 UTC.

Every claim we have tested, including the ones that held, is on the reports index.

We publish the tests that failed as loudly as the ones that held. Every number on this site carries its sample size, and every claim carries what it was measured against.

Read the whole evidence page →