LevelLogic Research

Range odds: where does price go next?

17 years of Nasdaq futures, minute by minute. Pick the range that just finished, say where price is, and read the odds of each target level — drawn on the range.

Get the free indicator ↓
Unpacking the events…

The best usable odds

Click a row to open it with every filter set.
#SetupAll years 2020+Cases

When a session finishes, where does the next one go?

Your situation

Tap a level or a time.

A real day like this

Details: every number and note

The previous candle as a target map

Your situation

Tap a level or a time.

A real day like this

Details: every number and note

How to read this

The words

  • High first / low first: price trades through that side of the range before the other side (a cross: a trade beyond the level).
  • Midpoint first / nearer 25/75 first: that level is touched strictly before either side breaks.
  • After a side is taken: back to the midpoint, the other side too, or a further half of the range beyond the side that was taken.
  • 3/4-range run / full-range run: price runs a further three quarters of the range, or a whole range, beyond the side that was taken. Same rule as the half-range run: a touch of the level, counted from the minute after the cross; cases that had already reached it are left out. Each step further out is less likely than the one before; the random-walk column shows how much of that is just distance.
  • Already touched (after a cross): two more choices say which levels have traded since the cross — back inside (the mid, or the other side, which means the mid too) and beyond (the half-, 3/4- or full-range run; each further run means the nearer ones too). Those levels stop being targets and the rest show their odds given that touch: only cases where the level traded strictly before the target count, and 1h/2h/4h are counted from the touch. Touching a level on the far side of the taken level usually lowers the odds of the runs (they are now a range or more away); touching a level on the way raises the odds of the next one. All of it tracks the random walk from the touch: distance and time. The dip lines are hidden while a touched level is set (they are measured from the decision).
  • Range size: the last session’s or candle’s high-low range as a fixed % of price, in three groups (Small / Normal / Large) whose cut points are printed on the control. Each session and each candle timeframe and start time has its own cut points, set from the 2020+ ranges (a third in each group). Because the sizes are fixed % of price, most older years fall in Small, so while a size group is on every odds number is read on 2020+ (all years in small print) and the random-walk comparison is not shown (a range’s size carries over into the next one). Size changes the odds through distance: a bigger range puts every level further away. There is no direction filter: a down session closes low, so pick where price is; trend direction adds nothing once you do.
  • Which comes first = your directional lean: the target versus the nearest level on the other side. (Details also gives the share where neither traded in the window, and the race against the levels further out.)
  • Window: from the decision to 1h / 2h / 4h later, the end of the next session or candle, or 17:00 ET. Pick it on the diagram.
  • Random walk: what a directionless market with normal volatility would do from the same spot. When the real number sits close to it, the odds come from distance and time.

Where the rest of the numbers are

The panes show only the odds, the levels, the dips and one timing fact. Each pane’s collapsed Details holds the rest: the 2020+ figure with its 95% range, all years, the random-walk reference, the three eras, cases that had already reached the level, neither / both sides, how far price got when it missed, the full dip sentence (with the minute the dip usually bottoms), the notes and the build-up table.

  • Dips (the dashed lines): how far price went against the target before touching it, on winning cases since 2020, as % of price. After a cross they are measured from the taken level, starting the minute after the first cross; otherwise from the decision price (the candle / session open).
  • The cross is a trigger to look, not a signal: after a cross the odds match a random walk from the same spot.
  • Across the 17:00-18:00 halt the odds are measured from the 18:00 open print; nights that open beyond the range have already decided every target and are left out.

How a leaderboard row gets there

At least 200 cases (80 since 2020); 65%+ over all years and since 2020; the three eras within 8 points; not a near-certainty on a tiny move. Sorted by usefulness: how far the lower end of the 2020+ range sits above a coin flip, scaled down when the target is under 50 points away. Every row’s odds come mostly from distance and time left (see What the odds mean). Click a row to see its random-walk reference and eras in Details.

What the odds mean

These odds are real. They come mostly from how far away each level is and how much time is left, so a level’s odds rise as price gets closer and fall as it moves away or time runs out. Use them to pick targets and set expectations; they don’t mean the level itself pulls price.

Which filters survived (v2 audit)

A filter stays only if some of its choices lift a main target by 5 points or more over the same situation without it, with at least 80 cases since 2020 and in every era.

FilterVerdictWhy
Where price is (quarter of the range)keptThe main driver: e.g. the 6h low goes first 45% from anywhere, 78% from the bottom quarter.
Candle start timekeptTime of day changes how much the market moves: 139 cells gain 5pp or more over “any time” (e.g. 12h top quarter, high first within 4h: 62% → 80% for the 06:00 candle).
Sunday evening (18:00 candle only)keptLeaving Sunday out lifts the 1h/2h 18:00 candle’s midpoint and 25/75 odds by about 8pp, in every era.
What has happened since (side taken / closed back inside / closed beyond)kept (narrowly)Counting only cases where the target was still untouched at the decision: closed back inside → midpoint by 17:00 is 7 to 10pp above “just taken” on 4h/6h/8h candles (71-76%) and after Asia closes back inside NY PM (76%); closed beyond → half-range run +8pp on 6h (74%). Elsewhere it adds under 5pp. No inside bar needed; not offered for the full-day range.
What the range did vs the one before (broke high / low / both / inside), 2-step chains, last 1-3 candlesdroppedOnce position is known they add nothing usable: 4 of ~1,200 session cells and 48 of ~20,500 candle cells (holding start time too) pass, scattered and at the sample floor — what chance alone produces.
Where it closed vs where the next one opensmergedBoth measure where price is; the one at the decision is used.
Wick vs 1-minute close; time-window menuremovedTargets are crosses (a trade through the level). The window is picked on the diagram.
Mother candles tabremovedNo better odds than the plain ranges; its one useful result (the closed back inside / closed beyond trigger) now lives in both tabs.

Data

NQ 1-minute bars, 2008 to 2025, back-adjusted continuous contract (% of price is only meaningful on recent years, so distances use 2020+). Trading day 18:00-17:00 ET. Standard days only. Examples never show dates.

Every number reproduces from the event tables

Study headlineWhereStudyThis dashboard
Asia -> London: nearer side first (test, next session)R2 s169.2% (n 1,363)69.2% (n 1,361)
London -> NY AM: nearer side firstR2 s173.8% (n 1,363)73.8% (n 1,360)
NY AM -> NY PM: nearer side firstR2 s168.1% (n 1,363)68.1% (n 1,359)
NY PM -> next Asia (18:00 open): nearer firstR2 s155.9% (n 1,316)55.9% (n 1,315)
18:00 open in top quarter of NY PM: high firstR2 s569.2% (n 435)69.2% (n 435)
Previous 4h candle: nearer side first (during the candle)R3 T_headline65.7% (n 7,994)65.7% (n 7,973)
Previous D candle: nearer side firstR3 T_headline73.3% (n 1,188)73.3% (n 1,187)
6h opens in bottom quarter: 6h low firstR3 T_quarter77.7% (n 1,120)77.7% (n 1,120)
D opens in top quarter: D high firstR3 T_quarter86.9% (n 467)86.9% (n 467)
4h opens lower-middle: Q25 before a breakR3 CSV / VERDICT76.7% (n 1,770)76.7% (n 1,770)
6h reference: nearer side first within 1hR3 T_horizon33.0%33.3% (n 5,258)
LevelLogic research · NQ 1-minute data · ET clock · plain odds, not advice.

Educational research, not financial advice. These are historical measurements from futures data; they describe what happened, not what will happen.

EXPLORE THE LEVELLOGIC INDICATORS

Every LevelLogic tool is built from stats like these.

Related indicator

  • SEIZ SEIZ shows where a candle sits against its own measured history, on the chart.

Related research