Reading the Indicators - Cival Manipulation & Pressure Toolkit

This page explains every element each indicator draws, every state it can be in, every panel line and how each number is computed. Defaults are the values a fresh instance loads with; the full list is in 02-Settings-Reference. Setting names appear in italics exactly as ATAS shows them.

Both indicators describe what already happened in candles and executed order flow. The vocabulary is deliberate: pattern, candidate, evidence points (a checklist, not a probability), reference prices (not orders or targets). Footprint volume shows executed aggression; it does not show who holds positions or where stops are.

Demo data used for the worked examples. The store demo was produced by the shipped plugins (Cival.Manipulation 1.1.1.0 and Cival.PressureToolkit 1.0.0.0, default settings) on Hyperliquid BTC-PERP five-minute bars built from trades recorded by the Cival order-flow recorder, 17 to 21 October 2026, 1,380 closed bars, tick size 1. Times are UTC bar open times. Examples illustrate mechanics on one sample; they say nothing about future results. The demo carries no individual-print or live-depth feed, so those two Pressure Toolkit layers are explained from their rules only.

Layout note (both indicators). Group 00 Chart clarity holds Compact display (full detail must be off) (on) and Full analysis detail (on). With the defaults you get the full layouts described below, and a small text layer spells out abbreviations (RVOL is shown as Relative volume) and rewrites a few labels, noted where it matters. Turning Full analysis detail off and Compact display on gives a reduced layout: Manipulation shows only its newest cycle, Pressure Toolkit shows at most 8 recent markers (roughly the last 20 bars) with plain-language labels and at most 3 levels, and both hand their summary to a small shared panel titled CIVAL / CURRENT CHART READ.

Cival Manipulation

1. What it measures

Cival Manipulation maps one sequence on closed candles: a compressed range, a one-sided sweep of one boundary, a closed-price reclaim back inside, and a later directional expansion through the opposite boundary. "Manipulation" is the name of the pattern, not a claim about anyone's intent, hidden stops or conduct. It is described in the product as research: it draws reference prices, it does not place orders.

Only one cycle is active at a time. When a cycle ends (for any reason) there is a cooldown of 4 bars before a new range can be recognised. The indicator keeps at most 64 cycles and 512 events.

2. The phases

Other events recorded in the report: retest (first later bar within 0.1 ATR of the broken boundary that closes back on the expansion side) and target1 (1R reference observed). They are not drawn as markers.

Reference prices. At expansion the indicator sets: Reference = the expansion bar's close; risk = the distance from Reference to the invalidation; 1R and 2R = Reference moved in the cycle's direction by one and two times that risk. These are analytical distances: not entries, orders, targets or profit and loss. If a later bar crosses both the invalidation and a target, the invalidation takes priority.

3. Fixed rules (not settings)

ATR here is the simple mean of the previous 14 true ranges (ATR baseline bars), excluding the current bar (a plain average, not Wilder smoothing), and is frozen into each cycle when its range is recognised.

4. Evidence points (a checklist, not a probability)

Evidence is recomputed at each phase event on the bar that produced it, and frozen after expansion. The panel shows Evidence N/100 · M observable: N is the points met, M is the points available (items that could not be evaluated, such as flow items without footprint, are not scored).

5. Recorded order flow

A candle has recorded flow only if its bid + ask volume adds up to its candle volume within 1%, with consistent totals. Anything else is treated as unavailable, never as zero. Orderflow confirmation chooses how flow is used:

6. What is drawn

The newest Maximum visible cycles (8) that overlap the visible bars are drawn. Labels move to a free spot nearby or are dropped if crowded. There are no markers for retests or targets; read those in the exported report.

7. The panel, line by line

Hover a candle to select the cycle spanning it; otherwise the newest cycle is shown, even if it has finished. The panel sits in Dashboard corner (TopRight) with Dashboard width 370 (240 to 520).

8. Timing, history and gaps

• Closed bars only. The newest candle is always treated as forming and excluded, even on a static chart. A phase appears when the candle that completes it has closed. There is no live preview layer. Alerts (if on) fire when the next candle begins.
• Ranges are drawn back, known late. A range is recognised at the end of its 12-bar window. The early part of the box is context, not an earlier signal.
• Revisions. If ATAS delivers a corrected candle or you change a setting, the study rebuilds from the loaded history, and earlier cycles can change.
• History window. Maximum bars back (3000, 100 to 10000) bounds the analysed bars; each warm-up needs about 15 bars.
• Coverage breaks. On a time-based chart, a gap between consecutive candle times of more than 1.5 x the chart's timeframe (a session break, a weekend, or an interval with no candle) invalidates any active cycle, clears the baselines and restarts the warm-up. A malformed candle does the same.

9. Status messages

10. Alerts and export

• Alert. Local alert on NEW confirmed expansion (off) notifies once for each new expansion on the candle that just closed: Cival Manipulation: long expansion confirmed, 100/100 evidence points. (or short, with the actual score). Nothing fires while history loads, during recalculation or after a rebuild. Alert sound defaults to alert1.
• Export. Export current diagnostic report (an action) writes manipulation-report.json to %APPDATA%\ATAS\CivalManipulation, overwriting the previous file. It holds the instrument, provider, a note that timestamps are the ATAS chart clock (milliseconds since 1970, no timezone inferred), a flag that reference guides are not trades, and the full report: settings used, every retained cycle with its evidence items, and the retained events (range, swept, reclaimed, expansion, retest, target1, complete, invalidated, expired).

11. Where every setting acts (32)

Out-of-range numbers are clamped to the range shown.

12. Worked example from the demo data

Range (known at bar 246, 20:30). Bars 235 to 246 (19:35 to 20:30) formed a window with high 81,622 and low 81,514: width 108, against an ATR of 36.36 (limit 3.2 x 36.36 = 116.3), and a price efficiency of 0.195 (limit 0.30). Evidence at this point: 10 points (compression).

Sweep (bar 252, 21:00). The bar traded up to 81,639, past the high by more than max(1, 0.15 x 36.36 = 5.45). A buy-side sweep, so the cycle leans short. Evidence 30.

Reclaim (bar 253, 21:05). The next bar pushed the extreme to 81,662 (40 beyond the high, 1.10 ATR, within the 1.75 limit) and then closed at 81,604, below 81,622 - 0.05 x 36.36 = 81,620.2. Invalidation: 81,662 + max(1, 5.45) = 81,667.45, rounded outward to 81,668. Evidence 85.

Expansion (bar 256, 21:20). Bars 254 and 255 did not qualify. Bar 256 opened at 81,592 and closed at 81,480: body 112 = 3.08 ATR (needs 0.8), body / range 0.97 (needs 0.55), closing location 0.97 (needs 0.68), and a close below 81,514 - 0.1 x 36.36 = 81,510.4. Reference 81,480; risk 81,668 - 81,480 = 188 (5.2 times the frozen ATR); 1R 81,292; 2R 81,104. Evidence 100/100: relative volume 5.76x, direction-adjusted delta +95.4%, five-bar CVD response, counterflow +23.7%, a recorded stack.

Afterwards. Bar 257 (21:25) traded down to 81,120, observing the 1R reference. Bar 258 (21:30) traded down to 80,966, observing 2R: complete.

The same run, in aggregate. 56 ranges were recognised; 50 were swept; 34 were reclaimed; 5 reached expansion; 2 of those ran on to the 2R reference; 3 were invalidated by a close back through the midpoint, one of them with 100/100 evidence. The other cycles ended before expansion: 17 invalidations (touch of the invalidation), 12 expiries without qualified displacement, 8 expiries without a reclaim, 8 sweeps that went too deep before reclaim, 5 sweeps deeper than 1.75 ATR, and 1 range that aged out. These are lifecycle counts on one short sample, with no entries, costs or profit and loss modelled. They show how few recognised ranges run the whole sequence, and that a full evidence score is not a guarantee of anything.

13. If a reading looks wrong or stays empty

Do not loosen thresholds merely to force a pattern. Source corrections, a different amount of loaded history, a different timeframe or bar type, and settings changes can legitimately recalculate results; cursor movement only changes which cycle the panel shows.

14. What it cannot know

• Whether anything was "manipulated", who acted, or where anyone's stops were. A sweep is a price excursion beyond a boundary, and the evidence points are checklist items for a hypothesis, not exposure of hidden positions.
• Why a candle closed where it did, or what happens after it.
• Resting orders: it uses candles and executed volume only.
• Anything on a bar with no recorded flow: such evidence is unavailable, not zero. Flow coverage is an internal consistency check, not an audit of the feed.
• Session context: it resets on time gaps on time-based charts but has no concept of session open, news or liquidity.
• Anything before Maximum bars back or before a restarted warm-up.

Cival Pressure Toolkit

1. What it draws

Five independent layers, each with its own switch and data needs:

A row is one price level of a candle's footprint: bid volume (trades at the bid, aggressive selling), ask volume (trades at the ask, aggressive buying), and row volume = bid + ask. Volumes are in the feed's units, not dollars. The newest candle is always treated as forming; a candle counts as closed once the next one exists.

2. Pressure bubbles

For each closed candle and each side, a row qualifies when, at the same price, the aggressive side's volume is at least Minimum aggressive row volume (10, in feed units) and at least Pressure ratio (3) times the opposite side's volume. Buying qualifies on ask volume against bid; selling on bid volume against ask. If any row qualifies:

• Position. The bubble sits at the qualifying row with the largest row volume.
• Number in the centre. The aggressive-side volume summed across all qualifying rows of that side in the candle. It is not the candle's volume and not a single order, and it counts only the aggressive side even when the bid-plus-ask volume in those rows is larger.
• Label. Pressure or Strong pressure, then B: (sum of ask volume over the qualifying rows), S: (sum of bid volume) and Δ: (sum of ask minus bid), in compact form (K / M).
• Strong. The label says Strong pressure (and the fill is brighter) when the biggest qualifying row's row volume is at least max(the floor, mean + 2 standard deviations of row volumes over the previous 20 bars). Both signals need the previous Row statistics lookback (20) bars to have footprint and at least 30 rows in total, so the first reading comes on the 21st footprint bar. This statistic does not decide whether a row qualifies; it only sets "Strong" and the absorption threshold.
• Colour. Green (#3CB371, Bull color) = aggressive buying. Tomato (#FF6347, Bear color) = aggressive selling.
• Size. Radius 5 to 20 pixels: 5 + 15 x the square root of (this bubble's number / the largest bubble of the same kind in the previous 100 bars), so bubbles are relative to recent history and do not resize when you pan.
• One of each side per candle, so a candle can show both a buying and a selling bubble.

Numbers are drawn with a dark outline in Bubble number color (white), largest bubble first. Text stays anchored to its price and crowding hides text instead of moving it: with Avoid overlapping labels on, a number or label that would overlap one already drawn is skipped, and labels are also hidden when bars are too close together (Minimum bar spacing for signal labels (pixels), 6). The bubbles themselves are always drawn.

3. Wick absorption candidates and lines

A candidate needs a row whose row volume is at least the same "Strong" threshold (the greater of Minimum aggressive row volume and mean + 2 standard deviations over the previous 20 bars) and that lies outside the candle's body:

• Lower wick (row price below the lower of open and close) where bid volume exceeds ask volume: aggressive selling into a lower wick, which may mean buyers absorbed it. Drawn green.
• Upper wick (above the higher of open and close) where ask volume exceeds bid volume: aggressive buying into an upper wick, which may mean sellers absorbed it. Drawn tomato.

The largest such row per side is marked with an Absorption? bubble (number = that row's bid + ask) and creates a line: a one-tick band centred on the row price (price +/- half a tick), beginning on the next bar and running to the right. The question mark is deliberate: footprint rows cannot show which passive participant took the other side. A candidate is not confirmed absorption: this toolkit applies no price-response confirmation to it.

A line ends at its first touch: the first bar, including the still-forming one, whose high-low range overlaps the band. Touched lines are hidden by default (Show touched levels, off). The Maximum visible zones (12) untouched lines nearest to the last price are drawn, each labelled Wick absorption candidate New (or Tested when shown after a touch). Fill and centre line are the side's colour; a touched line is grey. Because touches are checked on the forming candle, a line can disappear mid-candle.

4. Big individual prints

Each closed candle can show one Print <volume> bubble per side, at the price of that side's largest exceptional print, with a number equal to the total exceptional print volume for that side.

• Source. Individual trade ticks only. ATAS supplies historical trades for up to 7 days (the request starts about Print statistics lookback (bars) + 2 bars back), and live trades are added as they arrive. Trades with no classified direction are ignored; aggregate cumulative volume is never used in place of individual prints.
• Threshold, computed from prints in the previous Print statistics lookback (bars) (100) bars, never the current one, and never below Minimum individual print volume (0):

• Readiness. At least Minimum populated lookback coverage (%) (80) of the previous 100 bars must contain prints; at least 30 prior prints must exist; the bar itself must have prints; and, unless Allow incomplete history is on, the print history must not have been flagged partial (trade groups without ticks, prints dropped by the retention limit, or ambiguous bar time stamps). The panel's tape line shows why a bar produced none.
• What this means on a fresh load. Only about the last 100 bars of prints are loaded, so the 80% coverage rule is met only for roughly the newest 20 closed bars right after loading. Older bars will not show print bubbles. After that, each new closed bar is evaluated as it closes.

5. Live resting depth

With a live depth feed, the 20 price levels nearest the last price in the indicator's depth book are drawn as short horizontal bars at the right edge of the price pane (up to 80 pixels wide, scaled to the largest displayed), green for bids and tomato for asks, under the label LIVE resting depth (updated levels only). The book is built from the platform's depth updates and re-seeded from a snapshot of the platform's market depth about every 2 seconds while updates are flowing. Rows not refreshed for Depth row expiry (seconds) (60) are dropped. If no depth update has arrived for Depth stale timeout (seconds) (10), the bars disappear and the label reads Depth stale/awaiting fresh feed. Depth is never reconstructed for history, and a feed may not show every resting order.

6. Round numbers

Nine faint grey horizontal lines, labelled Round <price>, centred on the last price. The step is 1, 2 or 5 times a power of ten, the smallest such step at or above 0.5% of the price, and never below one tick; for a price near 82,000, 0.5% is about 410 so the step is 500. Round-number step (0 = automatic) overrides it. They are arithmetic guides, not claims about orders. Alert on round-number cross (off) sends Cival round-number cross at <price>; step <step> when the live price moves from one step interval to another; this alert has its own switch and is not gated by the master alert switch.

7. The information panel and statuses

A panel at the top-left (Show readiness/status panel, on) shows the title, a readiness line and a tape line. Automatic panel placement stacks it with other Cival panels (a gold N panels need space: hide unused panels message appears when there is not enough room); Panel font size (13, 10 to 22) and Compact panel (title + two lines) (on) adjust it.

8. Alerts

Master switch Alerts on closed signal / level touch (off) must be on for the first four rows. Alerts are silent while history loads or recalculates.

9. CSV export

If CSV export folder (blank disables) names a folder, the indicator writes two files there and updates them as markers arrive and after each recalculation: PressureToolkit-signals.csv (time,available_time,bar,price,side,kind,label,volume) and PressureToolkit-zones.csv (available_bar,low,high,poc,side,label,state,touched_bar,invalidated_bar). available_time is when the marker could first be known (the next candle). Use a different folder per chart or instrument so files do not overwrite each other. It keeps the latest 500 markers and 1,000 lines. A failure shows in the readiness line as CSV export failed: <message>.

10. What updates when

11. Where every setting acts (44)

12. Worked example from the demo data

Pressure bubbles (bar 867, 20 October 00:15). The five-minute bar opened at 81,766, reached 81,786, fell to 81,562 and closed at 81,585 on 501.8 BTC with delta -228.08. The indicator drew a selling bubble at 81,631 labelled Strong pressure B:10.6935 S:130.52075 Δ:-119.82725: across the qualifying sell rows (bid volume at least 10 and at least 3 times the ask at the same price), 130.52 BTC sold into the bid against 10.69 BTC bought, and the number in the centre is 130.52. The largest pressure bubble (either side) in the previous 100 bars was 236.59, so its radius was 5 + 15 x square root(130.52 / 236.59), about 16 pixels. The same bar also drew a buying bubble at 81,657 (B:35.29434, radius about 10). Bar 868 followed with a buying bubble of 119.34 at 81,586.

Wick absorption candidate (bar 831, 19 October 21:15). The bar opened at 82,164, reached 82,208 and closed at 82,064. At 82,208, in the upper wick, ask volume exceeded bid volume and the row's total of 16.192 BTC exceeded the "Strong" threshold, so an Absorption? marker was drawn there and a line (82,207.5 to 82,208.5) started on bar 832. Bar 843 (22:15) traded up to 82,275 and touched it: 11 bars, 55 minutes, after it began. The line then stopped being drawn.

In aggregate. The run created 285 absorption lines; 277 were touched later, and 158 of those by the very next candle. Being touched is the ordinary fate of a line, and says nothing about whether the level mattered. The indicator keeps the latest 500 markers; in the demo's retained set, 352 were pressure bubbles and 148 absorption markers. Big prints and live depth are not in the demo data.

13. If a reading looks wrong or stays empty

Do not loosen thresholds merely to force output. Source corrections, a different amount of loaded history, a different timeframe or bar type, and settings changes can legitimately recalculate results; cursor movement does not change the closed-bar sequence.

14. What it cannot know

• Absorption. A large row in a wick with the opposite side dominant is a candidate. It cannot show that a specific passive participant absorbed anything.
• Resting orders in history. Depth is live-only and is not a record of the book over time; a different tool is needed for resting-liquidity history. No complete initial depth snapshot and no dollar liquidation amounts are implied.
• Who or why. Large executions are observed aggressive prints, not a trader's identity or a verified position opening. Executed volume does not identify participants, their size, or whether trades opened or closed positions.
• Print quality. Big prints depend on the provider's trade history, its caps and its classification of direction.
• Units. Row, print and bubble volumes are in the feed's units, not dollars; thresholds need setting per instrument.
• No invalidation, no lifecycle. Pressure Toolkit does not produce stops, targets or invalidation prices. Everything it draws is an observation on a closed candle (or live context).
• Anything about the future. None of the layers has been validated to predict price.