QWE AI Academy That Morning You Opened Your Chart and the Price Had Vanished
I still remember the first time it happened. Stock closed at $48.20. Next morning the open printed $45.10. No candles in between. Just empty space. My stop was sitting at $47.50 like a useless decoration. That blank strip is a gap down, and once you see how the order book actually creates it, charts stop feeling magical.
Gap up: today’s open higher than yesterday’s close. Gap down: open lower than that close. The chart leaves a blank because the regular session never printed the prices in between – most of the move already happened after hours or in the opening auction. Investopedia’s gap definition frames the same discontinuity: open differs from prior close with little or no trade printed between.
How the Jump Actually Forms (Order Book Reality)
Markets close. News drops – earnings, Fed speaker, geopolitical headline, overseas futures. After-hours and pre-market traders (or algorithms) reprice. When the regular open auction runs, the matching engine finds the new clearing price where supply meets demand. If that price sits far from the prior close, you get the visual gap.
Trading-forum explanations nail the mechanics: if the highest resting buy is $100 while the last print was $105, a market sell can open the next session at $100. The exchange doesn’t invent the jump; traders do by shifting bids and offers overnight while the continuous book is thin or closed.
Pro tip: Always check the pre-market high/low and volume before the bell. That range often becomes the first support/resistance of the regular session and tells you whether the gap has already been “traded through” by institutions.
Partial gaps stay inside yesterday’s high-low range. Full gaps break completely outside it. That single cut changes how far price often travels.
| Type | Definition | Typical implication |
|---|---|---|
| Full gap up | Open > prior high | Stronger overnight demand shift |
| Partial gap up | Prior close < open ≤ prior high | Weaker; more likely to slip back into range |
| Full gap down | Open < prior low | Stronger supply shift |
| Partial gap down | Prior low ≤ open < prior close | Often shallower move |
Full gaps generally need bigger order flow to clear, so they tend to trend farther once the open settles – the same full-vs-partial split laid out in StockCharts ChartSchool.
Spotting Them Fast for Data Work
Flag rule is one line of logic. On daily OHLCV, gap up if open > prior close; gap down if open < prior close. Full gap: open > prior high or open < prior low. Add a percent floor (2% or 4%) and a volume floor so penny noise drops out.
import pandas as pd
# assume df has columns: date, open, high, low, close, volume
# sorted ascending by date
df['prior_close'] = df['close'].shift(1)
df['prior_high'] = df['high'].shift(1)
df['prior_low'] = df['low'].shift(1)
df['gap_pct'] = (df['open'] - df['prior_close']) / df['prior_close'] * 100
df['full_gap_up'] = df['open'] > df['prior_high']
df['full_gap_down'] = df['open'] < df['prior_low']
gappers = df[(df['gap_pct'].abs() > 2) & (df['volume'] > 500000)]Then dump the ticker list plus overnight headlines into an LLM: “Summarize the catalyst and whether volume supports continuation.” Scanner first, summary second. You get a shortlist worth reading – not a magic autotrader. Opening-range and volume-profile labels sit nearby once these events are tagged.
What the Numbers Say About Filling
Traders love “gaps get filled.” Size and regime disagree.
Same-day SPY fills in recent multi-month windows (as of mid-2025 through 2026 updates from TradeThatSwing / Edgeful-style SPY gap stats) land roughly 53-69% overall and slide with trend. Tiny gaps 0-0.19% filled about 79-92% of the time; 0.2-0.39% still decent; larger than ~0.4% often sat under 50%. Plenty of the fills that did print showed up by noon. Weekday skew shows up too – Friday gap-downs looked weaker in some of those samples.
A near-century look at US indexes (Plastun et al., 2020, North American Journal of Economics and Finance) finds real post-gap anomalies – and also shows the tidy “they always fill” story breaks once costs and regimes enter the model. Fill odds are conditional data. Not destiny.
Where Beginners Get Burned
Overnight gap risk is the quiet account dent for swing holds. Your stop only works after the open is liquid. A 5% hole through it is just a worse fill. Cut size before binary events, or flatten. Thin names can gap on almost nothing, then whip once the crowd chases – the exhaustion pattern after a multi-day run.
Common / breakaway / runaway / exhaustion labels mostly resolve in hindsight. Volume on the gap day is the fast filter you actually have at 9:35: heavy volume leans continuation; light volume leans fill attempts. Mis-tag a breakaway as exhaustion and you fade the start of the real move.
Honest limit: no free public table locks fill rates for every ticker and gap size forever. Stats drift when volatility regimes change. Backtest your own universe.
FAQ
Do most gaps fill the same day?
No. On SPY, only the smallest buckets show high same-day fill rates (often 80%+). Larger gaps frequently skip a same-session fill.
What’s the difference between a full gap and a partial gap in practice?
Yesterday’s range $50-$52, close $51.40. Open $52.30 = full gap up (outside the range). Open $51.80 = partial. The full print usually means a larger overnight imbalance and historically travels farther before mean-reverting; the partial often gets absorbed back into the prior day’s range within hours. Pre-market volume is the quick conviction check either way.
Can I use AI tools to trade gaps automatically?
Scanners yes. Full autopilot – usually no, not without hard risk rules. Gap risk, open slippage, and fill stats that drift by regime keep most retail edges semi-discretionary. Build features (gap size, volume ratio, pre-market range, sector RS), score context with a model if you want, then paper the signals hard. The edge lives in the filter, not in the word “gap.”
Open a daily chart of any liquid name that moved overnight this week. Mark full or partial, note volume, write one sentence on the catalyst. Five names. You’ll feel the pattern faster than any glossary entry.