Market Profile Myth check Intraday futures

The 80% rule and the initial balance: 2,154 sessions of Nasdaq futures

A viral clip says a candle established inside yesterday’s value area fills value 80% of the time, and that the initial balance rejects its first test nine times in ten. We tested both on every Nasdaq futures session since 2018, one minute at a time. Value filled about half the time, the initial balance was a coin flip, and one 1-minute reentry rule survived every out-of-sample test we had: small, recent, and Nasdaq only.

Atty Labs · 22 October 2026 · 14 min read

48% of reentries filled the far side of the previous day’s value area after his 5-minute confirmation. The claim is 80%.

51% of first tests of the initial balance moved 10 points the trader’s way before 12.5 against him. A random entry: 54%.

72% winners across 728 trades for the one reentry rule that survived, on 1-minute data, Nasdaq only

1.02 profit factor of the same rule on a two-day-old value area, against 1.35 on the real one: the level does the work

Where this came from

This started with a 55-second clip from the Titans Of Tomorrow podcast on TikTok, in which a prop-firm trader named Vince describes the two trades he watches every day. He marks the previous day’s value area, the range that held about 70% of the day’s trading, and the initial balance, the first hour’s high and low. His words on the first trade: “If price opens outside of that value area and it moves into that value area on a five minute chart, I wait until a candle establishes in that value. And if it establishes in that value, there’s an 80% rule that it’s gonna fill value.” He credits the rule to Jim Dalton. In the full episode he puts the initial-balance play at about a 90% rejection on the first test of either boundary.

Credit where it is due: both setups are described precisely enough to test, which is rare, and the 80% rule has a forty-year paper trail in Market Profile books^1,2. We wrote the tests and the pass bars down before anything ran, measured the behaviour first and built trading rules second, and the checks that came after the first results are labelled as such in the Method section.

What was tested

Everything runs on Nasdaq 100 futures (the NQ contract, priced as its micro, MNQ) at one-minute resolution from 2 January 2018 to 27 August 2026: 2,154 full sessions after removing holidays, early closes and the first session after each contract roll. The previous day’s value area was built two ways, from the time-price profile (TPO, 30-minute periods, exact from the minute bars) and from volume (approximate, since the data carries volume per minute and not per price), and for two sessions, the 09:30 to 16:00 day session and the full overnight-to-close electronic session. The initial balance is the high and low of 09:30 to 10:30 Eastern.

The 80% rule. Sessions that opened outside the previous value area and came back in. Seven confirmation rules, from a touch of the boundary through his first 5-minute close to Dalton’s original two consecutive 30-minute periods, and for each one the share that reached the far side of value before the close.
The initial-balance first test. The first touch of the IB high or low from inside the range, the retest from outside after a breakout, and the failed breakout that comes back inside. For each event, how often the favourable move arrived before the adverse one, on his own numbers (10 points for, 12.5 against) and on symmetric thresholds.
The same tests on other markets, then execution. The event studies were repeated on S&P 500, Russell 2000, crude oil and gold futures, and the two setups were turned into 1,164 mechanical rule variants and run as trades in AmiBroker with commissions, a development window, a validation window and a sealed holdout.

The 80% rule: about half, however you confirm it

The previous day’s TPO value area was 85 points wide on the median session. Nasdaq futures opened outside it on 1,424 of 2,108 usable sessions, and on about 60% of those the market came back inside at some point. After his confirmation, the first 5-minute close inside value, the far side of value was reached before the close on 48% of the sessions that had opened above it (467 reentries, 95% interval 44% to 53%) and 42% of those that had opened below (336 reentries). Waiting for a whole 5-minute candle inside lifts it to 53% and 45%; a simple touch of the boundary gives 43% and 38%. The point of control, the single price with the most trade, is reached far more often, about 76% of the time, which is probably where the folk number comes from.

Dalton’s own wording asks for two consecutive 30-minute periods inside value^1,3, not one 5-minute candle. That version does better, 62% for sessions that opened above value and 53% for those that opened below, and it is still well short of 80%. It also fires on only 38% of the outside opens, because an hour of acceptance is a lot to ask before lunch, and by the time it fires there is less session left to travel.

Two checks say the value area itself is not doing much here. A value area taken from two sessions back, which has no business describing today, filled at 49.5% after the same confirmation. And the overnight-session value area, which is wider, filled less (38% and 33%). Whatever is being measured, it is the general tendency of a market that has come back into yesterday’s range to keep going about half the time.

Wide value areas fill less, not more

The clip prefers wide value areas, with more distance between entry and target. The data runs the other way. Splitting the reentries into ten equal groups by the width of the value area, the narrowest tenth (about 23 points) filled 67% of the time and the widest tenth (about 235 points) 30%; for sessions that opened below value the widest tenth filled 9%. The adverse move before a fill grows with width too, from 16 points to 115. A wider value area pays more per fill and fills far less often, and with a stop of half the width the point expectancy of the widest groups is deeply negative.

The initial balance: a coin flip at the first test

The initial-balance claim is that the first test of the first hour’s high or low rejects about 90% of the time, with a 50-tick (12.5-point) stop and a reaction of 10 points being enough. We required the market to have been at least a quarter of the range away from the boundary first, took the first bar that touched it from inside, and measured what happened next, adverse move first whenever a bar held both. On 1,165 first touches of the high, the favourable 10 points arrived before the adverse 12.5 on 50.6% of them; on 1,060 first touches of the low, 50.4%. On symmetric thresholds (10 basis points each way) the fades score 45% to 48%: a touch of the boundary is followed by continuation slightly more often than by rejection. Second and third tests look the same.

Two controls settle whether the boundary matters. A level drawn a quarter of the range above or below the real IB high, run through the same first-touch machine, scores 53.5%. A random minute after 10:30 with a random side scores 53% to 55% on the same 10-before-12.5 measure, because the measure itself is lopsided. The real boundary is not special.

One thing in the first-test data does lean: the retest of a boundary in the direction of a breakout, that is, the IB high tested from above after the market has already closed above it. Those go the trade’s way 62% to 68% of the time. But the placebo levels show the same 66% to 69%, so it is the ordinary persistence of a market that has just made a directional move, not the initial balance. Crude oil shows it most strongly, and its placebo shows it just as strongly.

The same picture holds on S&P 500 futures from 2015, Russell 2000 micros for the one year available, crude oil at five different session starts from 08:00 to 10:00 Eastern, and gold at the same five starts back to 2010. No instrument, no session start and no year puts the first-test fade above 55%.

What the 1-minute data did show

Measuring behaviour is one thing; trading it is another, so we turned both setups into rules and ran them as trades in AmiBroker: one micro contract, $0.50 per side, a stop before a target whenever a bar held both, out at the close. The development window was 2018 to March 2023, the validation window April 2023 to December 2024, and 2025 to August 2026 was sealed until the end. Every initial-balance rule lost. The limit fade at the first touch was negative in all 78 of its stop and target settings; the penetration fade, the 5-minute rejection candle and the failed breakout were negative in almost all of theirs, and the two that cleared the development bar fell over in validation or the holdout.

The value-area rules were different, on one profile. Buying or selling the first 1-minute close back inside the overnight session’s volume value area, with the previous point of control as the target and a stop one full width away, won 72% of 728 trades across the three samples: profit factor 1.35 in development, 1.42 in validation and 1.36 in the sealed holdout, for $13 to $25 per contract per trade. Four ticks of slippage per side leaves it at 1.27. Run the same rule on a value area from two sessions back and the profit factor drops to 1.02, so for this rule, unlike the 80% fill, the level carries the information. A sibling rule that waits for a 5-minute close and then a retest of the boundary behaves the same (1.34, 1.32, 1.53).

Before anyone trades it, the caveats. The curve is flat through 2018 and 2019 and earns everything from 2020 on. The profile that works is the approximate one, built by spreading each minute’s volume across its range, and the exact TPO version of the same session is negative on the same target. The edge is about seven trades a month and small in dollars. Time stops and trailing stops destroy it, so it is a hold-to-target trade with a wide stop, which is a poor fit for a prop-firm trailing drawdown: in a bootstrap of a typical $50,000 evaluation, one micro passes slowly and two or more blow the funded account in most paths. And it does not travel. On S&P 500 micros the rule is positive but a two-day-old value area beats the real one; on crude and gold it loses.

A rolling walk-forward, re-optimising the rule’s stop and target on two years and trading the next six months, was positive in 9 of 13 folds and made $8 per trade against $16 for the fixed setting, with the picked setting drifting between the point of control and the far boundary as the target. That is the honest shape of it: a plateau of similar rules that has worked for six years on one contract, not a law.

What to take from this

The 80% rule is closer to 50%. After a 5-minute close back inside the previous value area, the far side was reached 48% of the time; Dalton’s stricter two-period version reaches 62% on fewer days. The point of control is reached about three quarters of the time, which is a different and much nearer target.
Wide value areas are the wrong ones to prefer for a full-traverse target. Fill rate falls from 67% in the narrowest tenth to 30% in the widest, and the adverse move before a fill grows with width.
The initial-balance first test does not reject. It is a coin flip on his own numbers and a slight continuation on symmetric ones, on five markets, and a level drawn anywhere nearby behaves the same.
There is something in the 1-minute reentry on the Nasdaq. Small, recent, one contract, one profile, and it passed every out-of-sample test we had. It is worth watching, with an exact volume profile and real fills, before anyone trades it.

What we’d test next

Three cheap tests decide whether the reentry rule is a value-area effect or an artefact of how we built the profile. First, rebuild the overnight volume profile from trade-level data, so the point of control is exact, and rerun the rule; if it needs the approximation, it is not a value-area edge. Second, replay the 728 trades on 1-second bars to check the next-open entries and the limit fills at the point of control. Third, split the result by year on basis-point rather than dollar thresholds, to see whether 2018 and 2019 were a regime or a scale effect. Only then a forward observation.

Method

Data: Databento continuous front-month futures, one-minute bars, Eastern time, stamped at bar start: NQ 2018–2026 (the research series; MNQ pricing of $2 a point for trades), ES 2015–2026, M2K 2025–2026, CL 2018–2026, GC 2010–2026. Contract rolls fall at 00:00 UTC in expiry week; the first session after each roll is excluded from every previous-day calculation. Sessions with a NYSE holiday, an early close or fewer than 350 of the 390 day-session minutes are excluded, leaving 2,154 Nasdaq sessions.
Value area: Previous session only, never touched by the current session. TPO profile: 30-minute periods, one count per 0.25-point bin each period touches; volume profile: each minute’s volume spread evenly across the bins its range covers (an approximation, since the bars carry no volume at price). The point of control is the bin with the most counts, nearest the range centre on a tie; the value area grows from it two bins at a time on the heavier side until it holds 70% (median share 70.1%). Both a day-session (09:30–16:00) and an overnight-to-close profile were built.
Events: Fill study: sessions whose 09:30 open was outside the previous value area; confirmations A1 touch, A2 first 1-minute close inside, A3 first 5-minute close inside, A4 whole 5-minute candle inside, A5 two 5-minute closes, A6 reentry then a boundary retest, A7 two consecutive 30-minute periods closing inside (Dalton), A8 two 30-minute periods entirely inside. Targets: 25%, 50%, point of control, 75%, far boundary, far boundary plus 25%, each measured to the 16:00 close with the adverse move before the target recorded. Initial balance: first touch from inside after the market had been at least 0.25 of the IB width away (0.10, 0.50, fixed ticks and ATR units reported as sensitivity), retest from outside after a 1-minute close beyond the boundary by the same distance, failed breakout on a close back inside; excursions measured from the boundary from the next bar on, adverse first within a bar, for favourable thresholds of 2 to 60 points, ticks, basis points and IB-width units.
Controls: Initial balance: the same machine run on a level 0.25 of the IB width outside and inside the real boundary, and five trials of random post-10:30 minutes with random sides, matched in count per session. Value area: the previous session’s profile replaced by the one from two sessions back. Both were decided before the first result.
Trading rules: AmiBroker 6.93, custom formula, one micro contract, $0.50 commission per side, slippage 0 / 1 / 2 / 4 ticks reported. Level fills at max(level, open) or min(level, open) with a one-tick trade-through; targets as resting limits with a one-tick trade-through; stop before target in any bar that held both; session exit at the 15:59 close; one trade per session. Development 2 January 2018 to 31 March 2023, validation 1 April 2023 to 31 December 2024, holdout 1 January 2025 to 27 August 2026, opened once per pick after validation. 1,164 settings in 26 projects; a pick per family was the best development profit factor with at least 150 trades, carried with its nearest neighbours to validation; the bar was PF above 1.05 and a positive net. Bootstrap intervals use 10-day blocks, 5,000 resamples. The walk-forward re-runs each family’s 24-setting grid on two years and tests the next six months, 13 folds.
Added after the first results: The Dalton two-period confirmation (A7, A8), after a search showed the original rule asks for two 30-minute periods rather than one 5-minute candle. The cross-market execution runs on MES, MCL and MGC with the Nasdaq-chosen setting. The prop-firm and cash-account bootstrap, because the mandate asks for it once a rule survives cost testing.
Defects found and fixed: Three, all before any number was kept: a swapped high/low in the short-side excursion code, caught by a hand audit of ten events; the AmiBroker importer raising the level symbol’s high, caught by reading the imported bars back; and a case-insensitive name collision in the formula that froze the levels at the first session’s values, caught because the first value-area trading result was too good and identical across different targets. Every trade list was audited against the raw bars after the fixes.

References

Dalton, James F., Eric T. Jones and Robert B. Dalton. Mind Over Markets: Power Trading with Market Generated Information, updated edition. Hoboken, NJ: Wiley, 2013. The 80% rule and the value-area definitions.
Steidlmayer, J. Peter, and Kevin Koy. Markets and Market Logic. Chicago: Porcupine Press, 1986. The original Market Profile text: value area, point of control, initial balance.
NexusFi. “80 Percent Rule.” Trading-concepts entry summarising the rule as two consecutive 30-minute periods inside the previous value area after an open outside it. Link
CQG. “Market Profile Value Areas (MPVA).” Help documentation describing the 70% value-area expansion from the point of control. Link
Dalton, James F., Robert B. Dalton and Eric T. Jones. Markets in Profile: Profiting from the Auction Process. Hoboken, NJ: Wiley, 2007. Value, acceptance and the initial balance as the authors use them.
Titans Of Tomorrow (@titansoftomorrow). TikTok video, 29 August 2026. The prompt for this study; every figure here is computed independently. Link

Historical behaviour of Nasdaq 100 futures around the previous day’s value area and the initial balance, with hypothetical trading rules run in AmiBroker. Not a strategy or a recommendation, and past results do not predict future ones. Full disclaimer.