# Three questions this dataset can answer The data has daily open, high, low, close and volume for 152 semiconductor-related stocks. It has 422,729 rows from 2012-01-02 to 2025-08-04. About 23.7% of rows (99,984) fall after ChatGPT's launch, which gives a sizeable "AI era" window to compare against the earlier period. ## 1. Did the AI surge change the price and risk profile of semiconductor stocks, and for which companies? - **Why it's interesting:** This is the core theme of the dataset. The pre-ChatGPT history from 2012 is long enough to act as a baseline, and the post-ChatGPT window holds about 100k company-days. - **How to answer it:** Compare each company's pre and post-2023 returns and daily high-low range (the overall average is 3.69% of close). The yearly per-company gold table `semi_conductor_se_by_company_name_date__yyyy` supports this directly. - **What it would show:** Whether the gains were concentrated in a few AI names or spread across the sector. That matters for over- or underweighting decisions. ## 2. Has trading activity shifted, and when does volume signal unusual interest? - **Why it's interesting:** Volume is extremely skewed. The median daily volume is about 898,500 shares, the mean is about 8.47M, and the maximum is about 3.69B. A few heavily traded names or spike days dominate the total. - **How to answer it:** Track volume by year and by company, then flag days that are far above a company's own rolling median. This turns raw volume into an entry and exit timing signal. It would also show whether the AI era brought a structural increase in volume or just a few bursts. - **Caveat:** 4,888 rows have zero volume. These are probably halted days, illiquid tickers or data gaps, and they should be filtered or checked first. ## 3. Which companies behave like a "semiconductor basket", and which are outliers? - **Why it's interesting:** With 152 tickers, the set is broader than the usual handful of big names. It invites questions about which stocks move together, which ones diversify, and which are volatile relative to their price level. The median close across all rows is about $35.57, but price levels vary widely by company. - **How to answer it:** Compute return correlations between companies, and compare each company's volatility and volume against the sector median. The per-company gold table `semi_conductor_se_by_company_name` gives a quick first ranking. Correlations would need the daily data. - **What it would show:** Candidate members for a sector benchmark, clusters of similar stocks, and outliers to treat separately when sizing risk. ## Caveats before any of these - The card lists about 149 companies, but the data shows 152. Check for duplicate or renamed tickers before ranking companies. - The data is observational, with no exchange or sector fields. Comparisons across exchanges are not possible here, and the "AI effect" can only be shown as a before-and-after pattern, not proven as a cause. I can run any of these next, starting with the pre- and post-ChatGPT comparison by company.