Entry Points Now
Cycle entry model: after each trading day’s close, it scans the largest 100 Taiwan stocks with a rules pre-filter (pullback, MA support, oversold reversal, volume contraction, momentum flip), then adds AI notes and fitted post-entry backtest statistics.
Today's entry points overview
AI今日共 4 檔標的通過進場門檻，分別為廣達、英業達、嘉澤及東元。觀察規則覆蓋率，R1 達 4/4，是今日所有標的的共同特徵；R4 覆蓋 3/4，顯示多數標的符合該條件。個別來看，廣達（2382）觸發 R1+R3+R4，距高 -18%；英業達（2356）觸發 R1+R3+R5，距高 -26%。嘉澤（3533）與東元（1504）則均觸發 R1+R2+R4，跌幅分別為 -35% 與 -39%。整體分布顯示，除了 R1 全數命中外，R2、R3、R5 的觸發情況較為分散，各標的進場邏輯存在差異。
Updated：10/21/2026 01:24 PM
Entry Points Now
| Rank | Stock | Cycle stage | Signals | Price | Score | Historical signal backtest5-year multi-rule resonance backtest. Differs from the Backtest Lab bias grid search methodology. | |
|---|---|---|---|---|---|---|---|
| #3 | Pullback | R1·Pullback in rangeR2·Above MAsR4·Volume contraction | 1,820 | 3 | Hist. signals 20 Avg. days to target 10.2 | ||
| #1 | Pullback | R1·Pullback in rangeR3·Oversold reversalR4·Volume contraction | 328.5 | 3 | Hist. signals 21 Avg. days to target 11.1 | ||
| #4 | Pullback | R1·Pullback in rangeR2·Above MAsR4·Volume contraction | 70.6 | 3 | Hist. signals 17 Avg. days to target 21.7 | ||
| #2 | Pullback | R1·Pullback in rangeR3·Oversold reversalR5·Momentum flip | 61.3 | 3 | Hist. signals 21 Avg. days to target 6 |
Methodology
The model pre-filters with the rules below (3+ rules hit, including R1 or R2, max 10 per day):
- Pullback in range: 15%–40% off the 52-week highest close, with close at least 1% (inclusive) above the 52-week lowest close (highs/lows measured on closes; intraday extremes excluded).
- Above MAs: close above both the 20-day and 60-day moving averages (state check; no pullback-touch test).
- Oversold reversal: RSI dipped below 35 within 10 days and has recovered above 40.
- Volume contraction: 5-day average volume below 20-day average.
- Momentum flip: MACD histogram turned positive within 5 days.
Passed candidates (score≥3 with R1 or R2) enter at the next trading day's open; 5-year backtest uses +8% target / −5% stop / 40 days: expiry without a trigger settles at the day-40 close (excluded from win/loss); same-bar double-trigger counts as a stop (conservative).
• This page (Historical Backtest): Evaluates strictly matching R1–R5 multi-rule resonance over the past 5 years (selective signals), reflecting multi-rule historical behavior. • Backtest Lab (/backtest): Runs a grid search over 5–15 years on single quarterly-MA bias thresholds to identify mathematically optimal entry levels.
Explore the Backtest Lab for custom parameters and interactive charts →
Everything on this page (including AI notes and historical backtest statistics) is for information only and does not constitute investment advice. AI and quantitative models can be wrong or delayed; investment decisions are yours.