**The most surprising finding is that the widest price range in the dataset is not a recent AI-era winner, and that range is not the same as return.** - **NVIDIA has the largest range of any stock.** Its close goes from about $0.26 to about $179.27, a max/min ratio of roughly 687x over 3,415 trading days (2012-01-03 to 2025-08-01). That fits the AI-surge story (step-3/step-7). - **The next three are not obviously "AI" stories.** - Lasertec, a Japanese chip-inspection equipment maker, is at about 443x (100.6 to 44,520 JPY). - Wolfspeed, a silicon-carbide and power-chip company, is at about 355x (0.40 to 141.87). - HANMI Semiconductor is at about 284x. - **This is a range, not a return.** The max/min ratio does not say whether the peak came after the trough. A stock that rose 300x and then collapsed looks identical to one that is still climbing. I did not check the order of highs and lows, so I can't say which of these names are still near their peaks. - **The largest names are the tamest.** Samsung ranges only 5.6x (15,300 to 85,681 KRW), and Kokusai Electric 3.3x, though it only has 434 days of history (step-2). - **Prices are in local currencies.** The data mixes KRW, JPY, USD and others, so raw price levels such as SK Hynix at 300,000 are not comparable across companies. Only ratios and returns are (step-1). - **Some extremes may be data artifacts.** - BrainChip's minimum close is $0.009265. - SK Hynix's maximum close is exactly 300,000.0, a suspiciously round number that I have not verified against the source. - 461 rows have a close outside the day's high/low range and 10 have an open outside it (step-0). - 4,888 rows have zero volume and 6,840 are "flat" days, so some quotes may be stale or filled in. The "AI surge" is probably real, but rankings built on max/min price can be distorted by currency, penny-stock effects and bad rows. I would clean the price rows and compute dated, currency-neutral returns before sizing any position on this data.