skills/stock-technical-analysis/references/experience.md

# Stock Technical Analysis Experience

Purpose: keep reusable technical-analysis lessons without forcing the model to read a growing archive every time.

## Active Playbook

Read this section before using the stock technical analysis skill.

### Direct Sell/Hold Answers

- For direct stock questions such as `怎么看`, `该不该卖`, `能不能拿`, `要不要加`, or `合理估值如何`, do not answer from the individual chart alone. Start with a compact `个股-板块-大盘` read when market data is available: whether the stock is stronger or weaker than its sector, whether the sector is leading, rotating, or退潮, and whether broad-market breadth supports risk-taking.
- Separate `short-term execution` from `medium-term valuation`. A stock can be fundamentally reasonable but technically broken for short-term trading; it can also be technically weak without being fundamentally expensive. State which timeframe drives the answer and give different invalidation/confirmation levels for each when needed.
- When valuation is requested after a sell/hold discussion, provide conservative/base/bull ranges, compare current price with the base range, and explicitly say whether valuation gives enough margin of safety or only explains why a bottom may form later.

### Intraday Execution And Pressure Zones

- Separate `touch`, `break`, and `tradable hold`. A one-tick or one-minute wick above resistance may prove demand exists, but it is not actionable for most exits if price is immediately hit back. For practical judgment, prefer a 5-minute hold, repeated bids above the level, or a clean retest that turns resistance into support.
- For U.S.-listed stocks/ETFs only, ChartExchange dark-pool/off-exchange levels can be used as secondary hidden-liquidity reference zones. They do not show buy/sell direction and should not be used for A-shares or Japanese stocks. A dark-pool level becomes technical support/resistance only after price confirms acceptance, rejection, repeated defense, or failed reclaim. Build the URL from the actual listing venue and ticker, e.g. `nyse-anet`, `nasdaq-nvda`, `nyse-spy`; SPY is a common ChartExchange `nyse-spy` case, and uncertain venues should be searched first.
- When the user's objective is to sell/trim near a target, analyze the pressure zone as an execution area, not as a prediction trophy. If price repeatedly spikes into resistance and is rejected, scaling into the zone can be more realistic than waiting for a perfect stand above the exact target.
- When the objective is trend holding, do not treat the first rejection at resistance as failure by itself. If the reclaimed lower level holds, pullback volume contracts, and momentum recovers, the setup can still transition from rejection to second attack.
- Always state separate plans for `intraday profit-taking` and `trend holding` when both are plausible. The same chart can justify trimming at resistance for one objective and holding a runner while a key support remains intact for another.
- If the stock is in a strong confirmed catalyst/theme and higher timeframe trend is intact, failed first attacks near resistance should be framed as `pressure-zone digestion` until the reclaimed support breaks. The invalidation level matters more than the emotional discomfort of the pullback.
- In a strong catalyst trend, a rejected target can become a liquidity test rather than a top if the next pullback holds above the reclaimed support, forms a higher low, and then reattacks with expanding price/volume. Update the active pressure level after that sequence; do not stay anchored to the first rejected target.
- When the trend is steep but orderly, overbought KDJ/RSI is a heat warning, not a standalone exit signal. Give more weight to higher lows, support conversion, and whether pullbacks are shallow versus whether oscillators are high.

### Intraday News And Expectation Gap

- For U.S. stocks with active headlines, especially large-cap names around political, regulatory, order, earnings, or deal news, compare `prior market expectation` versus `actual news`. Ask: what was expected, what actually landed, was it above expectation, merely in line/confirmation, or below expectation? A bullish headline can be bearish intraday if the market had priced a larger, faster, cleaner, or more certain outcome.
- Treat `good news but price sells off` as a technical signal, not a contradiction to ignore. If a stock gaps or spikes on positive news but cannot hold VWAP/short averages, loses the opening range, or makes lower highs on heavy volume, classify it as `sell-the-news / expectation gap` until price reclaims the broken pressure zone with volume.
- Apply this to numeric and qualitative news. Numeric gaps include order size, guide, margins, delivery pace, and deal value. Qualitative gaps include wording strength, timing, certainty, regulatory tone, management confidence, and whether the announcement solves the market's real concern. Separate `fundamental direction is positive` from `intraday tape is rejecting the news`.
- After a strong theme/catalyst gap or spike, watch for `upper shadow + lower successive highs + loss of event-day close/key round level`. That sequence is a warning that event premium is being distributed, even when the long-term story is intact. Do not let "the story is still good" soften a short-term technical invalidation.
- Treat key post-catalyst levels as a state machine. Above the event-day close/round-number support, the setup can be `digestion`; below it, and especially if price cannot reclaim it quickly, downgrade to `event-premium giveback` and shift the next support map lower. State the exact level that changes the read.
- Relative weakness matters more after good news. If the relevant sector/style is stable or only mildly weak but the stock loses key support on heavy turnover, attribute the move to stock-specific positioning or catalyst exhaustion before blaming the broad market.

## A-Share Cross/Trend Signals

Use golden/death crosses, divergence and moving-average alignment as auxiliary labels; require price/volume, support/resistance and market/sector confirmation. Do not publish screenshot-derived accuracy percentages without a reproducible sample, test window, entry/exit rule and survivorship treatment. A historical accuracy percentage is not a current trade probability.

In signal conflicts, prioritize price structure and volume: a golden cross below broken support is only a rebound candidate; a death cross that holds reclaimed support in a strong higher-timeframe trend can be digestion rather than failure.