Delta, CVD, footprints and open interest: check the source
Audit estimated delta, CVD resets, a chosen imbalance rule and OI versus volume with reproducible numbers.
Published: (UTC+3)
Русский · English · Deutsch · Español · Français · Italiano · 繁體中文 · 日本語 · 한국어
These are four related data checks, not one universal signal. Fictional trades and predefined rules illustrate measurement; they establish neither profitability nor new product features.
Delta: an estimate versus aggressor classification
Two lower bars have O 100→C 101 with volume 10 and O 101→C 100 with volume 16. The declared directional method estimates 10−16=−6 from total 26. A separate aggressor record for those 26 units has B=11, S=13, U=2 unknown. Classified delta is 11−13=−2; assigning U to either side bounds full delta between −4 and 0. Directional intrabar estimates are not directly observed aggressor sides.
Keep classification method, lower interval, venue, window and unknown trades with the research record. TradingView documents intrabar-direction estimates, including equal-O/C rules; do not relabel them exchange bid/ask delta. Aggressor volume also does not count more buyers than sellers: every matched trade has both sides. Here overall price 100→100 does not fall despite estimate −6. Neither its sign nor a method disagreement validates an Elliott count automatically.
| Research procedure | Volume | Δ |
|---|---|---|
| O 100→C 101 | 10 | +10 |
| O 101→C 100 | 16 | −16 |
| Estimated Δ | 26 | −6 |
| Buy B | 11 | +11 |
| Sell S | 13 | −13 |
| Unknown U | 2 | [−2, +2] |
| Classified Δ | 24 | −2 |
| Total | 26 | [−4, 0] |
Can −2 be called fully observed delta after silently dropping U=2?
No. −2 describes only the 24 classified units. Keep the unknown 2 separate; under the stated assumptions full delta lies in[−4, 0].
CVD: anchors and resets change the baseline
Given four deltas{6, −2, 4, −3}, accumulation from the start gives CVD{6, 4, 8, 5}. Resetting to zero before bar 3 gives{6, 4, 4, 1}. Final 1 does not mean the last delta changed: it remains −3. The accumulated interval changed.
Match source, delta method, timezone, session/reset and window; do not replace a missing delta with zero. Closes at bars 2/4 are 102/103. Comparing CVD 4 with 1 across the reset creates an apparent divergence; a common uninterrupted anchor compares 4 with 5, which has not fallen. This counters mechanical reversal “confirmation.” Even a valid divergence would be a separate observation, not proof of a wave or trading profitability.
| Bar | Close | Δ | No reset | Reset at 3 |
|---|---|---|---|---|
| 1 | 100 | 6 | 6 | 6 |
| 2 | 102 | -2 | 4 | 4 |
| 3 | 101 | 4 | 8 | 4 |
| 4 | 103 | -3 | 5 | 1 |
Can readings before and after a reset be treated as one continuous series?
Only after aligning their anchor. Reconstruct a common cumulative interval or explicitly compare separate sessions with their different baselines.
Footprints: cells, zeros and a chosen imbalance
At prices 100/101/102 assign(S, B, U)=(6, 2, 0), (3, 9, 2), (4, 0, 0). There are 24 classified units plus 2 unknown, total 26. Our teaching rule compares max(B, S)/min(B, S) horizontally on one row, requiring ratio≥3 and the smaller side≥3. At 101, classified 9/3=3 gives a provisional mark. At 100 the smaller side 2 fails; at 102 a zero denominator is ineligible.
Zeros here mean explicitly absent classified executions, not missing data; U is retained separately. If the unknown 2 at 101 are sells, 9/5=1.8 removes the mark. TradingView documents diagonal comparisons between adjacent prices; this horizontal rule is a different declared convention. Freeze classification, row size, threshold and zero handling before testing. Even a persistent mark establishes neither participant intentions, a wave nor the next price move.
| Price | Sell S | Buy B | Unknown U | Total | max/min |
|---|---|---|---|---|---|
| 100 | 6 | 2 | 0 | 8 | 6/2=3; min 2<3 |
| 101 | 3 | 9 | 2 | 14 | 9/3=3; min 3≥3 |
| 102 | 4 | 0 | 0 | 4 | 4/0: — |
| Total | 13 | 11 | 2 | 26 | Classified Δ=−2 |
Can 4/0 be called infinite proof of seller control?
No. The chosen rule excludes zero denominators and audits missing data separately. Cells are not counts of people and do not guarantee continuation.
OI: outstanding contracts versus traded volume
One fictional contract and expiry starts with OI=50. Both sides open 12 contracts: volume+12, OI 62. Both close 7: volume+7, OI 55. A transfer of 9 from closing to opening participants adds volume 9 while OI stays 55. Period volume 12+7+9=28 and OI change+5. Each contract is counted once despite having a long and a short side.
Match contract, expiry, venue and report time; expiry, delivery and corrections are excluded here. A predefined ΔOI>0 research flag is true in this example but does not mean a “net-long market” or cash inflow. Counterexample:7 opens, 7 closes and 14 transfers give the same volume 28 but unchanged OI 50. Neither volume nor OI recovers wave structure. This article does not establish field availability in a particular product.
| Research procedure | Volume | OI change | OI balance |
|---|---|---|---|
| Start | 0 | 0 | 50 |
| Both sides open | 12 | +12 | 62 |
| Both sides close | 7 | −7 | 55 |
| Position transfer | 9 | 0 | 55 |
| Total | 28 | +5 | 55 |
Can ending OI be calculated from volume 28 alone?
No. Opening/closing/transfer classifications and initial stock are needed. Volume 28 is compatible with ending OI 55 or 50 in the two examples.