MU, 1주 뒤 더 높을까?

Indicators as of 2026-08-20 close 974.33: price above all six moving averages (MA5 +0.75%, MA10 +4.72%, MA20 +8.99%, MA50 +1.17%, MA100 +21.50%, MA200 +71.73%), with 10-day slopes MA5 +11.94%, MA10 +8.50%, MA100 +6.90% and MA200 +6.73% positive, MA20 +0.07% flat and MA50 -0.91% negative. RSI(14) 55.16, neutral. Returns 1w +2.58% / 1m +1.55% / 3m +33.11%; 19.71% below the 52-week high. 20d realized volatility 100.72%; volume ratio 5d/20d 0.75, so the advance came on fading volume. 1w base rate up 0.5405, sample_size 988, basis individual; overlap-adjusted effective_n 197.6 and z_adj +1.14, which is below 2, so this base rate is not distinguishable from 0.5, is not used as directional evidence, and no base-rate-deviation label is claimed. The base-rate window is about 988 trading days (~3.9y) and is up-regime skewed (1y bucket 0.873, 6m 0.7973), so it is not treated as unconditional, especially with the smoothed US recession probability macro series at 0.60 versus 0.32 prior (regime tag, not used as a direction input). Public news context: HBM3E reported sold out through most of 2026, data-centre revenue reported up more than 150% y/y, FY2026 revenue guidance reported at 31.5-33.5B USD, a capital expenditure plan reported raised to about 20B USD, and a sell-side upgrade to Buy with a 1,250 USD target reported. Counter-evidence considered and treated as material: the same up thesis on this name resolved incorrect in the most recent weekly round (t0 2026-08-14, return -0.50%), the MA50 slope is negative, volume is fading, and realized volatility is above 100%. Confidence is therefore set below the 1w base rate itself. Not investment advice.