REPORT / INITIAL BALANCE

Which side breaks first, or does it hold?

YM / New York / 5 yr
49%broke the initial balance high first, over 1,119 sessions
high first49%n 547
low first47%n 528
held the balance4%n 44
breaks at least one side (secondary)96%n 1,119
initial balance report / YM regular hours / 1,119 sessions

Built on a continuous index proxy, validated against CME data.

Filtered view. The headline above is the whole sample; the cuts and the rows below are the sessions matching this filter. Clear the filter

By weekday

share of sessions

By size

share of sessions

By move against

how much heat it took first

By extension

how far it ran past the level
The latest 60 sessions on file, not the full sample the headline counts
SessionOutcomeDirectionSize, ptsAgainst, ptsExtension, ptsMinutes
2026-09-02single-highup237.5219.0855.987
2026-08-26single-lowdown203.0190.4667.040
2026-08-19single-highup263.75247.850.58
2026-08-12single-lowdown233.48118.251.7553
2026-08-05single-lowdown366.5219.4936.98329
2026-07-29single-lowdown434.54327.01400.564
2026-07-22holdnone345.08---
2026-07-15single-lowdown321.51181.0490.01106
2026-07-08single-lowdown250.96163.51885
2026-07-01single-highup440.69148.35297.7227
2026-06-24single-highup302.91170.83346.612
2026-06-17single-lowdown308.62130.59578.14212
2026-06-10single-lowdown28723.51573.8916

Built from a Nasdaq 100 / S&P 500 / Dow 30 index tape, not from exchange futures ticks. Bar to bar return correlation against real MNQ is 0.988 over a 5,238 bar overlap, and 93 to 94 percent of swing pivots land on the same bar. The one minute pull is cross checked against a lossless local tick set over 2026-06-10 to 2026-08-25 and must correlate above 0.98 to ship. Levels are relative, so the rates hold; absolute prices carry a basis of roughly 160 points that drifts. The YM history starts 2021-09-01, eight months later than NQ and ES, and carries 124 unexplained weekday gaps in the underlying feed concentrated in 2021 to 2023 and tapering after 2024. Sample sizes on this page are smaller than NQ and ES for the same lookback because of this, not because of a filtering difference in the report.

Sample runs 2021-09-01 to 2026-09-04, 1,119 sessions. Skipped: 47 half days, 142 data holes, 0 sessions the report did not apply to.