Does SPY fill the gap?

Partially, essentially always. Fully, rarely.

Across 167 sessions where SPY opened between 0.3% and 0.7% above the prior close, every one traded below its own opening price at some point — 167 of 167, average maximum dip 0.55% below the open. A complete fill, meaning a close back below the prior day's close, happened on 18.6% of them (31 of 167).

Between those two poles sits the number that describes the cohort best: 68% of these gap days (114 of 167) gave nearly all of the gap back intraday, trading down to +0.2% or less above the prior close at some point in the session.

SPY gap up odds: what a +0.3% to +0.7% gap did next

Rates are measured against the prior day's close unless the row says otherwise.

| Outcome | Rate | Count | | --- | --- | --- | | Touches +1.0% intraday | 32.9% | 55/167 | | Touches +1.5% intraday | 10.8% | 18/167 | | Closes ≥ +1.0% | 18.6% | 31/167 | | Closes ≥ +1.5% | 6.0% | 10/167 | | Closes below the prior close (full fade) | 18.6% | 31/167 | | Closes below its own open | 44% | 74/167 | | Trades below its own open at some point | 100% | 167/167 |

The averages fill in the shape. The cohort's average close was +0.48% above the prior close. The average move from the open to the close was +0.02%. The gap carried the day; what followed the bell averaged out to nothing.

How the gap cohort compared with an average session

Measured against all 888 sessions in the sample, the gap cohort was better than average — modestly.

| Metric | Gap +0.3% to +0.7% (n=167) | All sessions (n=888) | | --- | --- | --- | | Touches +1.5% intraday | 10.8% | 7.5% | | Closes ≥ +1.0% | 18.6% | 11.5% | | Closes ≥ +1.5% | 6.0% | 4.4% |

For downside reference, the all-session baseline touched −1.5% intraday on 9.3% of days and closed at or below −1% on 10.4%.

What happened after the gap faded

The 114 sessions that surrendered the gap are the more informative half of the cohort, because the fade usually did not end the day.

| Path after the gap | n | Outcome | Average close | | --- | --- | --- | --- | | Faded to ≤ +0.2% | 114 | 72% closed green again | +0.26% | | Continued to ≤ −0.2% | 56 | 54% finished red | −0.17% |

Recovery was common; a large recovery was not. Of the 114 faded sessions, roughly one in three reached +0.5%, and the average close sat at +0.26% — a green day, but a thin one. When the fade pushed through to −0.2% or worse, the distribution flipped: those 56 sessions averaged −0.17% at the close, and a majority finished red.

What time does SPY bottom intraday?

Early, more often than not. This section rests on 38 regular-hours minute-bar sessions from 2026-06-10 to 2026-08-04 — a short, recent window and the smallest sample here. It describes two summer months, not all market regimes.

| When the session low printed | Share of sessions | | --- | --- | | 09:30–10:00 ET | 37% (average −0.31%) | | Through 11:15 ET (cumulative) | 58% | | After 14:00 ET | 29% |

The first move below the open followed the same clock. A crossing of 0.3% below the open occurred at all on 66% of sessions (25 of 38): 24% had crossed by 10:00, 32% by 10:30, and 53% by 11:15 — meaning about 80% of the sessions that crossed at all had already crossed by 11:15.

Deeper crossings ran later. A move 0.55% below the open happened on 50% of sessions (19 of 38), and 10 of those 19 first crossed after 11:15. The shallow dip was a morning event. The deep one was not reliably anything.

The 11:15 checkpoint

Where a session stood at 11:15 said less about the rest of the day than the timing data might suggest.

| State at 11:15 ET | n | Closed above its 11:15 level | 11:15 → close average | | --- | --- | --- | --- | | Above the open | 20 | 45% | −0.03% | | ≥ +0.3% above the open | 9 | 33% | −0.12% | | Below the open | 18 | — | 0.00% |

The strong subset is nine sessions. That is too few to carry a claim, and we are not making one from it. Across all three groups, the average rest-of-session move rounded to roughly flat.

SPY consecutive up days statistics

Big up days clustered, but weakly, and the clusters were short.

| Prior run | n | Next day ≥ +1% | Next day closed up | Next-day average | | --- | --- | --- | --- | --- | | One day ≥ +1% | 101 | 10% | 60% | +0.13% | | Two consecutive days ≥ +1% | 9 | ~22% | 67% | +0.50% |

The two-day row is nine observations. It points the same direction as the one-day row and it is not evidence on its own.

Streak lengths across the 888 sessions: 84 runs of a single ≥+1% day, 6 runs of two, 2 runs of three, and never four or more. The pattern rules behind our own long signals sit on this cohort — how our SPY daytrade signals work sets out which rules fire and what they measured.

The day after two big days

A narrower cohort: a +0.3% to +0.7% gap opening the session after a day that closed ≥+1.5%. Nine sessions. Small enough that the counts, not the percentages, are the honest presentation.

Touched +1.5%: 0 of 9
Closed ≥ +1.0%: 0 of 9
Closed red: 2 of 9
Closed below its own open: 4 of 9
Open→close average: −0.30%

Nine sessions produced zero upside extensions. That is a real absence in the record and a sample too thin to generalize from.

Where SPY's move showed up across the tape

Correlation and beta to SPY over the 250 sessions ending 2026-08-04.

| Symbol | Correlation | Beta | | --- | --- | --- | | QQQ | 0.93 | 1.41 | | IWM | 0.81 | 1.22 | | XLI | 0.72 | 0.95 | | NVDA | 0.66 | 1.87 | | TSLA | 0.62 | 2.24 | | AVGO | 0.58 | 2.16 | | AMD | 0.57 | 3.14 | | GOOG | 0.55 | 1.36 | | XLF | 0.55 | 0.62 | | AMZN | 0.54 | 1.46 | | MU | 0.54 | 3.38 |

Correlation and beta are separate readings, and the table's extremes make the point. QQQ tracked SPY's direction most closely in the set, at 0.93, and moved 1.41 times its magnitude. MU moved 3.38 times SPY's magnitude at a correlation of 0.54 — the loosest tracking in the table alongside AMZN, at more than double QQQ's amplification. A high beta describes size of move, not reliability of direction.

What the same setups looked like a month out

Over 21-session forward windows across the full sample (n=868), SPY closed ≥+3% 36% of the time, touched +3% at some point 52% of the time, and averaged +1.56%. Following two consecutive ≥+1% days (n=9), 4 of the 9 windows closed ≥+3%, averaging +2.75%, with the worst window at −3.0%. Nine windows, again — the count is the story, not the average.

How we measured this

Every figure above is derived from our own stored bars: SPY daily bars from 2023-01-18 through 2026-08-04 (889 sessions) and SPY one-minute bars from 2026-06-10 through 2026-08-04 (38 regular-hours sessions). Cohorts are defined by the open-versus-prior-close relationship and recomputed from the stored series, not hand-selected.

The results that argue against activity are in here on purpose. A 100% rate of trading below the open, an open-to-close average of +0.02%, a 44% rate of closing red from the open, and a nine-session cohort with zero upside extensions all belong in the same table as the 32.9% touch rate. Base rates are only useful when the unflattering ones are printed next to the rest.

Derived, delayed data. Impersonal research, not investment advice. We publish our own measured aggregates and cite public sources.