Which AI Tool Lets You Check Whether a Weekly IDX Candle Is Complete?
A BBCA read-only test returns different weekly bars for the same date. Daily rows explain the gap—and provide a practical acceptance test for an AI research tool.
By Trade Inc Research
Trade Inc MCP can return daily and weekly price bars for inspection, but our BBCA test found that a weekly date label alone does not prove a complete week. Changing the requested lookback from 14 to 21 days changed the bar labelled 2 October 2026. For buyers of AI research tools, the useful test is whether the underlying daily rows let you detect that difference.
This guide assesses that narrow audit workflow, not whether Trade Inc automatically guarantees complete weekly bars. The public connection guide documents daily and weekly retrieval through an authenticated MCP connection. Access depends on the account and compatible AI client; this is not a claim about every website control or subscription.
Key takeaways
- Request daily data as well as weekly data, with an earlier starting window.
- Compare the same labelled week across requests, not just the latest closing price.
- Exclude a boundary week from full-week comparisons until its constituent sessions are established.
- A reproducible aggregate is not an independent audit of the underlying exchange data.
The same date, two different bars
On 14 October 2026, we made read-only BBCA weekly requests with 14-day and 21-day lookbacks. The following values were returned for the same date label. Prices are rupiah per share; volume is the numeric volume field returned by the service.
| Request | Open | High | Low | Close | Returned volume |
|---|---|---|---|---|---|
| 14-day lookback, 2 October label | 6,100 | 6,150 | 5,975 | 6,100 | 471,300,900 |
| 21-day lookback, 2 October label | 6,225 | 6,250 | 5,975 | 6,100 | 733,191,900 |
The close and low match, but the open, high and volume do not. A check that looked only at closing prices would miss the difference. Repeating both weekly requests produced the same discrepancy.
We then examined the daily responses. The shorter window began on 30 September; the longer one included 28 and 29 September as well.
Reconcile the missing portion
The daily volume fields for 28 and 29 September were 112,107,600 and 149,783,400. Their sum is 261,891,000—exactly the difference between the two weekly volume fields:
733,191,900 − 471,300,900 = 261,891,000.
The longer window’s opening value of 6,225 matches 28 September’s open. Its high of 6,250 also appears on that date. The shorter window’s open of 6,100 matches 30 September. These observations are consistent with the first weekly aggregate being restricted by the request boundary. They do not establish the implementation cause; we did not inspect server code.
As a second check, the 9 October weekly bar was identical in both requests: open 6,100, high 6,200, low 5,975, close 6,050 and volume 442,318,000. Summing the returned daily volumes for 5–9 October also gives 442,318,000.
A practical acceptance test before subscribing
Ask the connected assistant: “Retrieve daily and weekly BBCA bars using two lookbacks. For the same weekly date, show which daily dates contribute to open, high, low, close and volume. Flag a first week that starts inside the shorter window.”
Then inspect the actual returned dates. To reconstruct an aggregate from a specified set of daily rows, take the first open, maximum high, minimum low, last close and sum of volumes. This checks internal consistency, not market-data completeness.
If the first week changes, do not silently compare it with full weeks. Request enough earlier data to cover the intended period, establish the relevant sessions, or omit that boundary bar. Asking for more data without rechecking merely moves the boundary farther back.
What this test does not settle
The daily response contained a flat 6,050 bar with zero volume on 7 October. We have not established why. A zero is not proof of a holiday, suspension or missing record. We also have not independently verified corporate-action adjustments, volume units, exchange coverage or latency. For that reason, this example reconciles the returned volume fields without converting them to lots or turnover.
No Screener was created and no backtest was run. Candle shapes and this consistency check are not trading signals.
Who benefits from this workflow?
It fits researchers who need inspectable price data inside an AI workspace and are willing to check period boundaries. It is less useful if an existing data service already supplies verified full-period exports and you do not need a conversational interface.
Trade Inc supplied enough daily detail to expose this particular discrepancy, but the test does not justify an unconditional endorsement of weekly completeness. Explicit constituent dates and a partial-period flag would strengthen the result; unexplained changes after matching the input dates would weaken it.
For navigation rather than data auditing, see our daily-versus-weekly chart walkthrough.
Trade Inc Research is affiliated with Trade Inc. This is a bounded product audit, not an independent ranking or investment recommendation. The cover is a conceptual illustration, not a chart of BBCA data.
This article is for research and information only. It is not personalized financial advice or an instruction to trade.