What is SMT in trading? Compare prices without inferring orders

The name is not a standardized exchange signal

Community authors use SMT for related but different detectors. For example, the linked script author describes cycle extremes and intraday comparisons. That publication supports the use of the term, not a universal formula or a verified trading edge. A session-extreme detector and a confirmed-pivot detector can legitimately mark different events.

A numerical example and two controls

These are invented matched observations in two positively related markets. A has a prior low of 100 and B a prior low of 200. At a later eligible comparison, A reaches 99 while B reaches 201. Under the strict rule “lower low versus higher low,” that is a bullish candidate.

Bearish comparisons and reversed roles

For prior highs A = 110 and B = 220, new highs A = 111 and B = 219 meet the corresponding bearish-candidate rule. Both-new-highs and equality cases do not. Apply the same rules with A and B exchanged.

• Absolute prices need not be similar. Comparing A = 111 directly with B = 219 has no meaning for this event definition.
• This page assumes a positively related pair. An inverse comparison requires different high-versus-low rules, specified separately.
• A relationship can change. Define any correlation eligibility test in advance, including whether it uses prices or returns and its lookback. Do not choose a pair only because its latest divergence preceded a reversal.

Why SMT can have confirmation delay

Suppose a strict pivot requires three lower highs on each side of a candidate high. The right-hand bars must exist before the pivot can be confirmed. On uninterrupted one-minute bars, the pivot bar opening at 10:00 has three subsequent bars opening at 10:01, 10:02 and 10:03; the earliest confirmation is their final close at 10:04.

• If comparing two pivots, both markets must satisfy the confirmation and matching rules. A missing peer bar can make the event unavailable.
• A historical line drawn back to 10:00 does not establish that an alert was possible then. Log detection time separately from swing time.
• A rolling high can be observed without waiting for a pivot, but it is a different model. It should not be promoted as the same pivot signal without delay.
• RSI divergence compares price with an oscillator; SMT compares instruments. Both still depend on their selected references and confirmation rules.

Turn the observation into a research specification

Before generating code, specify ticker IDs, interval, shared session, level definition, timestamp matching, equality policy and missing-data behavior. Keep an event log before adding trading assumptions.

• Use aligned data from both markets. Changing the chart timezone alone does not synchronize prices or feeds.
• In Pine, request.security can retrieve peer data, but lookahead_off is not a blanket non-repainting guarantee. Developing higher-timeframe values require separate treatment.
• For a trading test, add the traded instrument, entry time and price, stop, exit, size and costs. A marker alone has no defined win rate.
• Adding a fair value gap or structure-break condition creates a new combined rule. Compare it on an untouched period rather than assuming extra labels add independent confirmation.