Sessions & Opening Range

Session signals give market events a clock

A session signal labels where an observation sits inside a trading calendar. It can identify the open, close, overnight period, a fixed time window, the initial balance, or levels inherited from the prior session.

These labels are not mere chart decoration. Many market behaviors depend on who is active, when scheduled information arrives, when positions are rolled or marked, and when liquidity concentrates. A price move at the open can have different context from the same move at midday.

The calendar is part of the signal. A “day” must specify exchange timezone, session template, holidays, shortened sessions, breaks, and whether overnight trading belongs to the coming or previous day.

Opening range and initial balance are declared intervals

An opening range is the high and low formed during a chosen interval after the session begins. Initial balance is a related auction-market term often defined over a longer opening interval. The labels are not universal durations.

For an illustrative market opening at 09:30:

• a 15-minute opening range uses events from 09:30 through the declared 09:45 boundary;
• a 60-minute initial balance uses the first hour;
• the high and low become final only after the interval closes.

A strategy cannot use the final range high at 09:35 unless it defines a developing range that may still change. The developing and completed signals are different.

The range develops during the shaded opening interval. Its final high and low become fixed only at the boundary. A later break and an even later retest are separate decision events. Plotting all four labels on the original opening bar would erase the timing distinction that makes the sequence usable.

Break, retest, and acceptance are separate events

A price trading one tick above the opening high is a break. A close above it is a stronger but later definition. A retest requires price to return to the level under a declared tolerance and then satisfy another condition.

The signal can expose these events. The strategy decides whether they trigger continuation, failure, or no trade.

Gaps compare two session references

An overnight gap is usually the difference between the current session’s open and a prior-session reference, often the previous close. The measurement must say:

• which session close;
• which current open;
• whether extended hours are included;
• price or percentage units;
• how holidays and contract rolls are handled.

A gap does not imply it will fill. It describes a discontinuity between two declared references. A gap-fill strategy and a gap-continuation strategy test opposite hypotheses.

Prior-session levels are historical references

Prior high, low, close, VWAP, and profile levels summarize a completed session. They are known at the next session start, which makes their timing relatively clean.

Their interpretation is still empirical. A prior high can act as a breakout boundary, a reversion anchor, or an auction reference. Calling it support or resistance does not specify what the strategy does when price approaches, crosses, or rejects it.

Time windows are eligibility signals

A Time Window output can declare that a strategy is active only between two times. That may reflect liquidity, data availability, scheduled events, or risk policy.

Time gating can improve a result simply by removing expensive or inactive periods. It can also overfit recurring historical accidents. Compare the time-gated strategy with the same rule across the full eligible session, and account for the number of windows considered.

A complete opening-range observation

Illustrative example. Assume a completed 30-minute range from 09:30 to 10:00 has a high of 101 and low of 99. At 10:05, price trades at 101.20.

The signal can state that an upside break occurred after the range completed. It cannot yet state:

• whether to enter immediately or wait for a close;
• whether a retest is required;
• how far beyond the range counts as meaningful;
• where the stop and target belong;
• how large the position should be;
• whether volume or volatility confirms the move.

Those are strategy decisions.

The shipped Opening-Range Breakout/Retest signal consumes explicit opening high and low values; it does not infer the session or opening policy. A break flag is emitted before a qualifying retest can occur, and retest_window limits how long that later event remains eligible. The Initial Balance Volatility Regime also consumes an explicit baseline_range, such as ATR or rolling range. Its compressed or expanded label changes when that upstream baseline changes.

Session data pitfalls

• Daylight-saving transitions can shift local and UTC cutoffs.
• Futures have exchange-defined sessions and maintenance breaks.
• Some equity data separates regular and extended hours.
• Crypto trades continuously, so a session is a researcher-defined partition.
• Holiday and early-close sessions can make ordinary windows incomplete.
• Continuous futures rolls can create apparent gaps unrelated to market movement.

The instrument context must resolve these before the signal is trusted.

Session Segments and Time Window are related but not interchangeable. Session Segments classifies bars relative to an explicit session start and several phase boundaries. Time Window reports membership in one local-time interval, uses an inclusive start and exclusive end, and can wrap overnight. Neither signal infers an exchange calendar on the reader's behalf.

False readings to avoid

• Assuming “opening range” has one universal duration.
• Using the final range before its interval completes.
• Calling any trade beyond a level a confirmed breakout.
• Treating a gap as a promise to fill.
• Mixing regular-session and overnight references.
• Using clock times without timezone and calendar.
• Testing many time windows and reporting only the winner.
• Treating prior levels as causal barriers rather than testable references.

Shipped signals in this family

Try it in Arizmic

Further reading

• Heston, Korajczyk, and Sadka, “Intraday Patterns in the Cross-Section of Stock Returns” (2010) — Documents return continuation at matching half-hour intervals across trading days. It supports treating clock time as a possible state variable while remaining specific to its equity sample and cross-sectional design.
• Andersen and Bollerslev, “Intraday Periodicity and Volatility Persistence in Financial Markets” (1997) — Models pronounced intraday volatility seasonality and shows why an opening-range move should be compared with the normal activity of that time of day.
• Sullivan, Timmermann, and White, “Dangers of Data-Driven Inference: The Case of Calendar Effects” (2001) — Reassesses calendar rules while accounting for data snooping. It is a direct warning that searching many sessions, weekdays, and seasonal windows can manufacture an apparently exceptional schedule.