A historical frequency can be useful and still say nothing certain about the next bar. The distinction is easy to lose when an indicator prints a percentage beside a live chart. This guide explains how to read the chart-local counts in Trend Integrity Oscillator without converting a descriptive statistic into a trade promise.
Start with the event definition
TIO does not count every dip or rally that looks like a pullback. Its confirmed-bar state machine first establishes agreement between trend efficiency and pivot structure. A pullback becomes armed only when efficiency turns against that intact structure while the integrity threshold still holds. The episode then ends as resumed or broken under rules defined in the Pine source.
That definition is more important than the percentage. Change the arming rule, the frozen structural anchor or the resume rule and the denominator changes. The displayed rate belongs to this particular event definition; it is not a universal property of a symbol. An armed episode does not always close as a clean resume or a clean break, either — if the structural direction itself flips while the episode is still armed, the count updates silently without printing either outcome, which is one more reason to read the count as bookkeeping, not a forecast.
Never separate the rate from its sample
The panel maintains separate counts for bullish and bearish pullback episodes. It withholds a rate and shows collecting until the minimum sample required by the code exists. That guard is useful, but it does not turn the first visible percentage into certainty. A count can describe the loaded chart history accurately while remaining sensitive to the symbol, timeframe and available bars.
Read the rate and sample together, then ask what the denominator contains. It is a count of completed, rule-defined episodes on this chart—not trades, not account results and not a test of a complete trading plan.
Why the number moves
Each completed episode updates the counters. A marker printed on an earlier bar can therefore preserve the rate that existed when that event closed while the current panel shows a later rate after more episodes have been added. Both readings can be internally correct because they summarize different endpoints of the same growing history.
This is why copying a percentage into a static claim is misleading. The durable statement is the method: the indicator counts its own confirmed episodes and displays the current chart’s historical frequency with its sample.
Confirmation does not mean prediction
TIO commits its states, markers and counters inside a confirmed-bar block. The still-forming oscillator can move until the chart bar closes, while completed event states are not rewritten afterward. That is a non-repainting design choice, not evidence that the next confirmed event is predictable. See our guide to non-repainting indicators for the difference.
A practical reading checklist
- Confirm the symbol, timeframe and loaded history before comparing two panel readings.
- Keep bullish and bearish samples separate; the script does.
- Treat
collectingas a request for more history, not as missing evidence to fill with intuition. - Use the state label to understand which event is being measured before reading the frequency.
- Do not translate a historical continuation count into a win rate, expected return or entry instruction.
The TIO product guide covers the chart layers and alerts. For why two legitimate trend models can disagree, read Why Two Trend Indicators Can Disagree.
FAQ
Is the pullback rate an accuracy score?
No. It is the share of the script’s completed, rule-defined pullback episodes that resumed on the loaded chart history.
Why does an old marker differ from the current panel?
The marker freezes the rate at that event; the panel continues to update as later episodes complete.
Does a confirmed-bar count predict the next event?
No. Confirmation makes event bookkeeping stable after the close; it does not make the future known.