Cival Auction Structure — Reading the Indicator

Cival Auction Structure builds a TPO (time-price opportunity) profile for each session. A TPO profile answers "how long did the market spend at each price?", not "how much traded?". The unit is the time bracket (default 30 minutes): a price row's count is the number of different brackets in which that row was visited. Two visits inside the same bracket count once; a huge volume in one bracket still counts once. For volume at price use a volume profile; the two are complementary.

Everything below comes from the shipped plugin (v1.0.0). Defaults are the ones the indicator starts with.

1. What is drawn on the chart

Colors: lines and arrows use the Bull color (default green #3CB371) or Bear color (default red #FF6347). TVAL is always green and TVAH always red. TPOC is green if the prior session's TPOC is below the new session's first open and red if it is above. The line is a 1-tick-thick band with a faint fill (the TPOC fill is twice as opaque) and a 1-pixel line at the exact price. A level that has been tested turns grey.

Labels are drawn above the line in the form <name> <MM-dd> <state>; by default ATAS shows the names spelled out: TPOC as Time value center, TVAL as Time value low, TVAH as Time value high. The same expansion applies to the panel text. Nothing is drawn for the developing session's TPOC or value area except the histogram and the panel; its lines only appear after the session ends.

2. The session

The session is a daily window: it starts Session start (minutes after midnight) (default 0) after midnight and lasts 24 hours.

• Session timezone blank: the cut is made on the clock the chart shows.
• Session timezone filled with a Windows time-zone name (for example UTC, Eastern Standard Time): the indicator treats the chart's candle times as UTC, converts them to that zone, cuts the day there and converts back. An unknown name is rejected when you type it. If you use this, display your chart in UTC.
• The date in a level label (MM-dd) is the date of the session's start. Crypto and 24-hour markets have no universal session; the daily cut is your analytical convention.

Initial balance minutes appears in the settings list but this indicator does not use it.

3. How the profile is computed

1. Brackets. For each closed bar, bracket = floor((bar open time − session start) ÷ TPO bracket minutes). Brackets are measured from the session start, not from the clock hour.
2. Rows. The price grid is tick size × TPO row size (ticks) (default 4 ticks). A price belongs to the row floor(price ÷ row) × row; the row is identified by its lower edge. If a session's range would need more than 2,000 rows, the row size is doubled (repeatedly) and the existing visit sets are merged by union, so no bracket is ever counted twice. The panel's Effective row line shows the row size actually in use.
3. Visited prices. With footprint data, every price that traded volume in the bar (grouped into rows) records the bar's bracket. If a bar has no footprint rows, every row between its low and its high records it, and the profile is marked APPROXIMATE.
4. Count of a row = number of distinct brackets that visited it. Ten candles that all visit a row inside one bracket, plus one candle that visits it in a second bracket, give that row a count of 2, not 11. Footprint volume only decides which rows a bar visited; it is grouped into rows, never added to the count.
5. TPOC (time value center) = the row with the highest count; ties go to the lowest price.
6. Value area (TVAL–TVAH). 70% of the total of all row counts. Starting at the TPOC row, the area grows one row at a time towards the neighbour with the larger count (both when they are equal) until it holds at least 70%. - The edges shown are placed half a row from the row prices: TVAH is at the middle of the highest value-area row and TVAL half a row below the bottom of the lowest value-area row. With 4-dollar rows that is a 2-dollar offset. Treat the two lines as placed within about one row of the true value-area edge.
7. Single prints = rows with a count of exactly 1 (anywhere in the profile, not just at the edges).
8. Shape candidate, tested in this order (later rules override earlier ones):

9. Auction edges (low and high separately): if the extreme row has a count of 2 or more, Poor low candidate / Poor high candidate. If the extreme row has a count of 1 and the next row inward is adjacent and also has a count of 1 (a tail of at least two single rows), Excess low candidate / Excess high candidate. Otherwise Edge unresolved.

All shape and edge labels are rule-based candidates that change as the session develops; they are not statements about who is buying or selling.

4. Lifecycle of a session level

There is no "invalidated" state for these levels: a Tested level simply stops mattering to the indicator.

Live versus recalculated. When history is calculated (on loading or after any recalculation) a level is tested against its own first bar as well. In live trading the first bar of the new session is not tested at the moment the level is created, only the bars after it. A level that the first bar's range covered can therefore show New live and Tested after a reload.

Which levels are drawn: the nearest Maximum visible zones (default 12), ranked by distance from the last price divided by strength (TPOC counts as strength 2, the others 1), with a mild penalty for age (up to double at 1,000 bars old). Three levels per session means about four sessions of levels at the default. Levels that are not drawn still exist: they are still tracked (tested) and still written to the CSV zones file. Maximum visible zones only limits what is drawn.

5. Data-quality flags

If none applies the last two panel lines read Visited prices from real footprint rows and Bracket counts; not volume counts.

6. Failed-auction marks

For every closed bar of a session, the indicator compares it with the previous session's finished profile, if that profile was complete and exact.

• Let top be the lower edge of the previous profile's highest row and bottom the lower edge of its lowest row.
• If the bar's high is above top and its close is below top, a ▼ FA prior TPO high mark is placed at the bar's high.
• If the bar's low is below bottom and its close is above bottom, a ▲ FA prior TPO low mark is placed at the bar's low.

The mark is created when the bar closes. Several marks can appear in one session. Because top is the bottom edge of the highest row, a bar can be marked while its high is still below the previous session's exact high, by less than one row. A mark records that one bar probed beyond the prior extreme and closed back inside; it says nothing about what happens next.

7. The panel, line by line

The panel has eight lines. By default only the first three are visible; untick Compact panel (title + two lines) (group Information panel) to see all of them.

If no profile can be built, the panel shows No eligible TPO profile; check rows and tick size (or Tick size unavailable). If a CSV problem occurs, an extra line with the message is inserted as line 2.

8. Compact display (optional)

Under group 00 Chart clarity, Full analysis detail (default on) and Compact display (full detail must be off) (default on) work together. With the defaults you get everything described above. If you turn Full analysis detail off, the chart switches to a short style: at most 3 levels and 8 recent signals (last 21 bars), short names (Time value center, Value area low, Value area high, High rejected (time profile)), the large panel is replaced by a compact board in the top-right corner (CIVAL / CURRENT CHART READ) with one line such as Time profile: Value 82694–83906; center 82768, and Show touched levels is ignored.

9. Alerts and CSV export

• Alerts on closed signal / level touch (default off) plays Alert sound (default alert1) when a failed-auction mark closes (Cival FA prior TPO high: <price>) or when a level is first touched (Cival level touched: <label> <low>–<high>). These are observations, not instructions. Recalculating the history does not fire alerts for old marks.
• CSV export folder (blank disables) writes AuctionStructure-signals.csv (columns time, available_time, bar, price, side, kind, label, volume; kind is failed-auction, volume is always 1; signals older than the last 1,000 retained bars are omitted) and AuctionStructure-zones.csv (columns available_bar, low, high, poc, side, label, state, touched_bar, invalidated_bar). side is 1 (green) or −1 (red). The files are rewritten when history finishes calculating and when a new signal appears; a level's later Tested state shows up in the file only at the next write. Use a separate folder per chart.

10. Settings that matter, and settings that do nothing here

Changing a calculation setting (row size, bracket minutes, session settings, Maximum bars back) recalculates the whole history. The indicator also recalculates when ATAS revises an already-closed bar. Source corrections, a different history prefix (loading more or fewer bars, which changes which sessions are complete), changes to any calculation setting and a change of timeframe can all legitimately change the results. Panning, zooming or moving the cursor changes only what is drawn, never the profile or the closed-bar marks. Hiding the panel, the histogram or the labels does not remove the underlying levels, marks or CSV rows.

11. Worked example — real numbers from the demo

Source. The store demo: Hyperliquid BTC perpetual, 15-minute bars (tick size 1.0) built from trades recorded by the Cival order-flow recorder, 17–21 October 2026, 459 closed bars. Settings: TPO row size (ticks) 4 (so 4-dollar rows), TPO bracket minutes 30, session start at midnight in chart time. Raw output: demo\out\auction-structure.json; profile data: demo\web\cival-auction-structure.json.

One session. The 18 October session (96 bars, 48 half-hour brackets) built this profile:

Levels carried forward. Because that profile was complete and used real footprint rows, three lines were published on the first bar of 19 October: TPOC 10-18 (81,700), TVAL 10-18 (81,194) and TVAH 10-18 (82,222). The TPOC line is green because the new session opened above it. In the recalculated history TVAH 10-18 is already Tested on that same first bar, because the bar's range reached 82,222, so it is not drawn by default.

Failed-auction marks. The 18 October profile's highest row is 82,268. During 19 October six bars (one per mark, at their highs 82,389; 82,288; 82,419; 82,310; 82,378; 82,275) exceeded 82,268 and closed back below it. Each got a FA prior TPO high mark. The whole demo sample produced 30 failed-auction marks, 12 session levels (4 completed sessions × 3) and 5 profiled sessions (the last still developing).

What happened next, as an example of the limits. The 19 October profile itself came out as b candidate: TPOC 82,008, value area 81,858–82,606, 24 single rows, with rows reaching 83,140. Value had moved up, and price traded well above the level where the six marks were placed. The marks did not call the top. They recorded that individual bars closed back under the prior high.

The last panel of the sample (the still-developing 21 October session) is quoted in section 7.

12. What the indicator can and cannot know