---
title: "Which AI Tool Can Check Whether Two IDX Stocks Move Together?"
description: "An actual MEDC–PGAS chat test shows why date alignment matters: recover the full window, reproduce 23 paired returns, and distinguish correlation from diversification."
author: "Trade Inc Research"
published_at: "2026-10-05T04:07:19Z"
updated_at: "2026-10-05T04:11:16Z"
canonical: "https://tradeinc.app/en/blog/ai-chat-idx-stock-return-correlation"
tickers: ["MEDC","PGAS"]
citations: ["https://www.itl.nist.gov/div898/software/dataplot/refman2/ch2/correlat.pdf"]
---

# Which AI Tool Can Check Whether Two IDX Stocks Move Together?

An actual MEDC–PGAS chat test shows why date alignment matters: recover the full window, reproduce 23 paired returns, and distinguish correlation from diversification.

## Summary

Trade Inc chat can calculate how closely two stocks’ daily returns moved together and explain the inputs. In our actual Pro-account test on 5 October 2026, MEDC and PGAS had a **Pearson return correlation of 0.6518** across **23 paired observations**, using stored closes from 1 September through 2 October 2026.

Getting the date window right required a follow-up. The initial answer used a shorter sample and returned 0.6427; a longer-history retrieval produced the requested window and matched our separate calculation. This guide shows that recovery, not a claim of flawless first-pass accuracy.

*Disclosure: Trade Inc Research publishes this guide about its own product. This is a bounded product demonstration, not an independent ranking, investment recommendation or forecast. Data cutoff: 2 October 2026.*

## Key takeaways

- Ask for correlation of **returns**, not a correlation between two price-level columns.
- Require matched dates, the earliest prices and the number of return pairs.
- Here, 24 aligned closing-price observations produce 23 returns because each return needs two endpoints.
- A positive historical correlation is neither a probability of moving together nor proof that one company causes the other’s moves.

## How to ask the question in Trade Inc

Open **Percakapan / Chat**, start a conversation, enter both tickers and specify an inclusive date range. We used **Cepat** response mode in an account displaying **Pro**, then pressed **Kirim / Send**. This observation does not establish minimum plan eligibility or current usage limits.

Our request asked for MEDC–PGAS daily simple-return correlation, a complete paired table and any uncertainty about price adjustments. It excluded creating a Screener, backtest, alert or monitoring task.

The first answer started from 3 September prices, giving 21 return pairs. Its 0.6427 coefficient was correct for that shorter sample, but it did not cover the requested starting dates. We asked for 60 calendar days of history, followed by a filter to the exact original window. Two earlier follow-up attempts failed to return a verified answer; the subsequent retry succeeded.

![Actual Trade Inc chat follow-ups requesting a longer history and filtering it to the original MEDC–PGAS date window.](https://cdn.sanity.io/images/eybaj6wm/production/29ed7d879074a30532c7c5c8048d554a31e8fff2-1024x356.png)

*Figure 1. The actual follow-ups. The repeated requests reflect recovery from failed attempts, not a requirement to submit the question multiple times. Screenshots are cropped to the conversation panel.*

A returned subset is not necessarily the entire available history. Check whether a short lookback has clipped the requested start before concluding that source data are missing.

## Inspect the recovered result

The successful response showed MEDC closes of Rp1,460 and Rp1,445, and PGAS closes of Rp1,555 and Rp1,535, for 1 and 2 September respectively. It returned **23 pairs and correlation 0.6518**.

![Trade Inc chat showing the recovered initial closes, 23 paired returns and Pearson correlation of 0.6518.](https://cdn.sanity.io/images/eybaj6wm/production/f4a4dec66b46f3c473f5e601dcbcf71e0f5b3df9-1024x356.png)

*Figure 2. Actual corrected-window result. The UI displays its formula as raw notation; the readable calculation below explains the same operation. Adjustment status remains unverified.*

We independently recalculated the coefficient from another retrieval of Trade Inc’s stored daily candles: **0.6518428249**, which rounds to 0.6518. This checks arithmetic and date alignment within the same data system; it is not an independent exchange-data audit.

## What the coefficient measures

First calculate each daily simple return:

`return = current close / previous stored-session close − 1`

For 2 September, that gives MEDC **−1.0274%** and PGAS **−1.2862%**. Pair the two returns for each matching interval. Subtract each series’ mean, sum the products of those deviations, and divide by the square root of the product of the two sums of squared deviations.

The [NIST DATAPLOT correlation reference](https://www.itl.nist.gov/div898/software/dataplot/refman2/ch2/correlat.pdf) describes Pearson correlation as a measure of linear relationship, ranging from −1 to +1, with equal-length inputs. Here the inputs are paired returns, not rupiah prices.

A coefficient of 0.6518 indicates positive linear co-movement in this sample. It does **not** mean a 65.18% chance of both stocks rising, identical percentage changes, or a profitable pair trade.

Correlation also does not measure each stock’s own volatility. Even with the same coefficient, portfolio risk changes with position weights and the size of each stock’s fluctuations. This article neither constructs nor recommends a portfolio.

## Who benefits from this workflow?

It suits a researcher who wants to investigate whether two holdings share historical movements and ask follow-up questions without leaving the conversation. The useful output is an inspectable pair of series, not just a coefficient.

If you already have clean, aligned data in a spreadsheet, a correlation function may be sufficient. This demonstration does not verify a dedicated correlation matrix, automated portfolio optimization or unrestricted code execution.

Before relying on the result, test other preselected periods and inspect unusual returns. Stability across longer and different windows would support using it as a descriptive input; large changes would weaken that use. Neither outcome guarantees future diversification.

## Method and limitations

We used 24 matching stored closes, with 23 consecutive-observation return intervals whose endpoints both fall inside 1 September–2 October 2026. No previous-month return was imported and no missing value was forward-filled. The appendix exposes the full series. We did not independently reconcile the exchange calendar or corporate-action adjustment treatment.

Twenty-three returns are a short sample. Outliers and common market movements can affect the coefficient; it does not establish an economic cause. Unknown adjustments and stale prices can distort return calculations. No separate dividend reinvestment or transaction costs were modeled, and these are not verified total-return series.

The pair was a demonstration choice, not a screen of all IDX stocks or a ranking. This is descriptive statistics, not a strategy backtest.

## Conclusion

Choose a tool that can expose the dates and inputs behind a correlation claim. Trade Inc chat completed this calculation after a longer-history retry, and the result reproduced. The practical habit is to validate coverage before interpreting the coefficient—especially when a polished answer uses fewer observations than you requested.

## Appendix: aligned inputs

Trade Inc stored closes retrieved on 5 October 2026; prices in rupiah per share. Returns below are rounded for display; calculation used unrounded ratios.

| Date | MEDC (Rp) | PGAS (Rp) | MEDC return | PGAS return |
|---|---:|---:|---:|---:|
| 2026-09-01 | 1460 | 1555 | — | — |
| 2026-09-02 | 1445 | 1535 | -1.0274% | -1.2862% |
| 2026-09-03 | 1485 | 1540 | 2.7682% | 0.3257% |
| 2026-09-04 | 1495 | 1520 | 0.6734% | -1.2987% |
| 2026-09-07 | 1485 | 1520 | -0.6689% | 0.0000% |
| 2026-09-08 | 1535 | 1540 | 3.3670% | 1.3158% |
| 2026-09-09 | 1575 | 1545 | 2.6059% | 0.3247% |
| 2026-09-10 | 1570 | 1515 | -0.3175% | -1.9417% |
| 2026-09-11 | 1555 | 1520 | -0.9554% | 0.3300% |
| 2026-09-14 | 1575 | 1550 | 1.2862% | 1.9737% |
| 2026-09-15 | 1565 | 1530 | -0.6349% | -1.2903% |
| 2026-09-16 | 1565 | 1520 | 0.0000% | -0.6536% |
| 2026-09-17 | 1525 | 1495 | -2.5559% | -1.6447% |
| 2026-09-18 | 1460 | 1475 | -4.2623% | -1.3378% |
| 2026-09-21 | 1475 | 1500 | 1.0274% | 1.6949% |
| 2026-09-22 | 1465 | 1455 | -0.6780% | -3.0000% |
| 2026-09-23 | 1465 | 1440 | 0.0000% | -1.0309% |
| 2026-09-24 | 1490 | 1440 | 1.7065% | 0.0000% |
| 2026-09-25 | 1450 | 1430 | -2.6846% | -0.6944% |
| 2026-09-28 | 1495 | 1440 | 3.1034% | 0.6993% |
| 2026-09-29 | 1460 | 1435 | -2.3411% | -0.3472% |
| 2026-09-30 | 1390 | 1395 | -4.7945% | -2.7875% |
| 2026-10-01 | 1380 | 1380 | -0.7194% | -1.0753% |
| 2026-10-02 | 1375 | 1375 | -0.3623% | -0.3623% |

## Related guides

- [Calculate maximum drawdown through chat](https://tradeinc.app/en/blog/ai-chat-idx-maximum-drawdown)
- [Evaluate custom portfolio simulations](https://tradeinc.app/en/blog/ai-chat-portfolio-simulations)


[1]: https://www.itl.nist.gov/div898/software/dataplot/refman2/ch2/correlat.pdf "NIST DATAPLOT — Pearson correlation definition and formula"