PHPMem v2.0.1

Version
1.6.45
Uptime
15 days 18 hours 12 minutes 56 seconds

Memory

Total
512MB
Used
9,38MB (1.83%)
Free
502,62MB

Keys

Current
11 436
Total (since start)
35 066
Evictions
0
Reclaimed
738
Expired Unfetched
0
Evicted Unfetched
0

Connections

Current
6 / 1 024 max
Total
178 047
Rejected
0
llm:67e62a5a7faa9f8bfed24f99a043d4aa59f075de9b53efed5904788c5ecf6333
TTL 6 days 10 hours 1 minute 3 seconds Size 2,32KB Export
Edit
Here are three questions this dataset can answer, based on what the profiling queries showed. The data covers 152 semiconductor companies, one row per stock per trading day, from 2012-01-02 to 2025-08-04, in 422,729 rows. **1. Did the AI surge change the price and risk profile of semiconductor stocks, and for which companies?** - About 99,984 rows, roughly 24% of the data, fall in the post-ChatGPT window. Every company therefore has a before and after to compare. - Across the full history the average daily range (high minus low, as a share of price) is about 3.7%, which gives a baseline for volatility. - A per-company comparison of price growth and range before and after the window would show which stocks re-rated and which stayed flat. - This serves the overweight/underweight decision and the question of whether the surge is a sector-wide shift or a few names. **2. Is the AI rally broad, or concentrated in a few companies?** - The median close across all rows is about $35.57, so a typical semiconductor stock is mid-priced. - Ranking companies by cumulative return since 2012 and since the surge would show how much of the sector's gain comes from a handful of leaders versus the rest. - This matters for building a sector basket or benchmark, because an equal-weighted basket and a cap-weighted one could behave very differently. **3. When does trading volume signal unusual interest, and does it lead price moves?** - Volume is highly skewed: the average is about 8.47 million shares a day, but the median is only 898,500. A few very heavily traded stocks dominate the average, so volume should be compared within each company rather than across companies. - Flagging days when a company's volume is far above its own rolling norm, then checking the following days' returns, would test whether volume spikes are useful timing signals. - The 4,888 zero-volume rows (about 1.2% of the data) should be excluded or investigated first, since they probably reflect halts, illiquid names or missing data. They would distort any volume-based signal. Every question here uses the existing date, price and volume columns. There are no sector, market-cap or exchange fields, so comparisons by exchange or company size would need outside data.