NVDA: Sit-Out, Then Surge, Week of July 6
NVDA's week of July 6, 2026: green through the rout, red on the rally, then the tape's heaviest name by Friday. Five sessions, with every number receipted.
NVDA spent the week of July 6, 2026 out of phase with its own sector, then out in front of it. On Tuesday's chip rout it was the only green close among the five majors tracked below; on Thursday, when all four peers closed up, the only red one. Friday it put up its best session of the week and finished as the heaviest ticker on that day's tape. Net result across 5 sessions: +8.5%, a $211.1 high, $107.1B of regular-hours dollars, the 3rd-biggest ticker of the week. Every number is a stored query; expand any panel for the SQL.
The week on one row
One row carries the whole week: the prior Thursday's close (July 3 was the Independence Day closure), open, close, both extremes with timestamps, volume.
| prior_friday_close | week_open | week_close | week_change_pct | peak_close_date | peak_close | week_high | week_high_first_bar_et | week_low | week_low_bar_et | week_shares_m | week_dollar_bn | rth_dollar_bn | session_days_observed |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 194.51 | 194.42 | 210.96 | 8.5 | 2026-07-10 | 210.96 | 211.1 | 2026-07-10 16:01 | 190.6 | 2026-07-07 07:35 | 565.2 | 113.6 | 107.1 | 5 |
The exact SQL behind every number
WITH
(
SELECT (toString(argMax(et_date, c)), max(c), argMax(c, et_date))
FROM (
SELECT toDate(toTimeZone(window_start, 'America/New_York')) AS et_date,
argMax(toFloat64(close), window_start) AS c
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
GROUP BY et_date
)
) AS closes,
(
SELECT max(toFloat64(high)) FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA' AND window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
) AS hi,
(
SELECT min(toFloat64(low)) FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA' AND window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
) AS lo,
(
SELECT argMax(toFloat64(close), window_start) FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA' AND window_start >= toDateTime('2026-07-02 00:00:00') AND window_start < toDateTime('2026-07-03 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
) AS prior_week_close
SELECT
round(prior_week_close, 2) AS prior_friday_close,
round(toFloat64(argMinIf(open, window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199)), 2) AS week_open,
closes.3 AS week_close,
round((closes.3 / prior_week_close - 1) * 100, 1) AS week_change_pct,
closes.1 AS peak_close_date,
round(closes.2, 2) AS peak_close,
round(hi, 2) AS week_high,
formatDateTime(toTimeZone(minIf(window_start, toFloat64(high) >= hi - 0.011), 'America/New_York'), '%Y-%m-%d %H:%i') AS week_high_first_bar_et,
round(lo, 2) AS week_low,
formatDateTime(toTimeZone(minIf(window_start, toFloat64(low) <= lo + 0.011), 'America/New_York'), '%Y-%m-%d %H:%i') AS week_low_bar_et,
round(toFloat64(sum(volume)) / 1e6, 1) AS week_shares_m,
round(sum(toFloat64(close) * toFloat64(volume)) / 1e9, 1) AS week_dollar_bn,
round(sumIf(toFloat64(close) * toFloat64(volume), (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) / 1e9, 1) AS rth_dollar_bn,
uniqExact(toDate(toTimeZone(window_start, 'America/New_York'))) AS session_days_observed
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')From $194.51 to $210.96: +8.5%. The bookends tell the arc: the $190.6 low printed at 2026-07-07 07:35 ET, Tuesday's pre-market, hours before the rout opened, and the $211.1 high at 2026-07-10 16:01 ET, the bar carrying Friday's closing auction. Bottomed before its second session, topped on very nearly its final print; 565.2 million shares, $113.6B with extended hours.
Session by session
Five sessions, no two alike, which days did the work?
| et_date | close_usd | change_pct | shares_m | dollar_bn |
|---|---|---|---|---|
| 2026-07-06 | 195.62 | 0.6 | 84.3 | 16.54 |
| 2026-07-07 | 196.93 | 0.7 | 110.4 | 21.52 |
| 2026-07-08 | 204.14 | 3.7 | 126.8 | 25.43 |
| 2026-07-09 | 202.76 | -0.7 | 112.1 | 22.67 |
| 2026-07-10 | 210.96 | 4 | 131.6 | 27.41 |
The exact SQL behind every number
SELECT
et_date,
close_usd,
round(if(prev_close = 0, NULL, (close_usd / prev_close - 1) * 100), 1) AS change_pct,
shares_m,
dollar_bn
FROM (
SELECT et_date, close_usd, shares_m, dollar_bn,
lagInFrame(close_usd) OVER (ORDER BY et_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW) AS prev_close
FROM (
SELECT
toDate(toTimeZone(window_start, 'America/New_York')) AS et_date,
round(argMaxIf(toFloat64(close), window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199), 2) AS close_usd,
round(toFloat64(sum(volume)) / 1e6, 1) AS shares_m,
round(sum(toFloat64(close) * toFloat64(volume)) / 1e9, 2) AS dollar_bn
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-07-02 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
GROUP BY et_date
)
)
WHERE et_date >= toDate('2026-07-06')
ORDER BY et_dateThe arc: modest green Monday and Tuesday, Tuesday being the chip rout day, when closing up +0.7% at all made NVDA the complex's lone survivor (receipts next section). Wednesday +3.7%, the week's second-best close, while the broad tape steadied. Thursday -0.7%, the sit-out: red on the summer's wildest open while every peer closed green. Friday +4% to $210.96 on 27.41B of full-day dollars, the heaviest ticker on Friday's tape, receipted below. Dollar volume climbed every day but Thursday: 16.54B Monday to 27.41B Friday.
The chip complex, day by day
'Traded against its sector' is a claim a reader should get to check: NVDA beside AMD, MU, INTC, and TSM (the TSMC ADR), all five sessions.
| et_date | nvda_pct | amd_pct | mu_pct | intc_pct | tsm_pct |
|---|---|---|---|---|---|
| 2026-07-06 | 0.6 | 6.5 | 0.9 | 1.5 | 4 |
| 2026-07-07 | 0.7 | -6.4 | -4.6 | -9.6 | -4.2 |
| 2026-07-08 | 3.7 | 0.1 | 1.1 | -0.2 | 0.9 |
| 2026-07-09 | -0.7 | 5.7 | 4.3 | 2.1 | 0.1 |
| 2026-07-10 | 4 | 2.1 | -1.2 | -2.4 | -0.6 |
The exact SQL behind every number
SELECT
toString(d) AS et_date,
round(maxIf(pct, ticker = 'NVDA'), 1) AS nvda_pct,
round(maxIf(pct, ticker = 'AMD'), 1) AS amd_pct,
round(maxIf(pct, ticker = 'MU'), 1) AS mu_pct,
round(maxIf(pct, ticker = 'INTC'), 1) AS intc_pct,
round(maxIf(pct, ticker = 'TSM'), 1) AS tsm_pct
FROM (
SELECT ticker, d,
(c / lagInFrame(c) OVER (PARTITION BY ticker ORDER BY d ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW) - 1) * 100 AS pct
FROM (
SELECT ticker,
toDate(toTimeZone(window_start, 'America/New_York')) AS d,
argMaxIf(toFloat64(close), window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) AS c
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker IN ('NVDA', 'AMD', 'MU', 'INTC', 'TSM')
AND window_start >= toDateTime('2026-07-02 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
GROUP BY ticker, d
)
)
WHERE d >= toDate('2026-07-06')
GROUP BY d
ORDER BY dMonday the complex ran, AMD +6.5%, while NVDA crawled +0.6%. Tuesday's rout: AMD -6.4%, MU -4.6%, INTC -9.6%, TSM -4.2%, NVDA green at +0.7%, the basket's only positive close. Thursday inverted it: all four peers up, AMD ahead at +5.7%, NVDA alone red at -0.7%. Friday NVDA's +4% led while MU (-1.2%), INTC (-2.4%), and TSM (-0.6%) slipped red. Three of five sessions, NVDA's close opposed the sign of at least three of its four peers. MU's deep-dive covers the name that set the week's volume pace.
The 3rd-biggest ticker on the tape
Rank claims need a stated basis and a receipt, the leaderboard for context, then a one-row receipt against every other ticker.
| ticker | rth_dollar_bn |
|---|---|
| MU | 164 |
| SPY | 137.3 |
| NVDA | 107.1 |
| QQQ | 103.4 |
| SNDK | 88 |
| TSLA | 69.3 |
The exact SQL behind every number
SELECT ticker, round(sum(toFloat64(volume) * toFloat64(close)) / 1e9, 1) AS rth_dollar_bn
FROM global_markets.delayed_stocks_minute_aggs
WHERE window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
AND ticker NOT IN ('SPCX')
GROUP BY ticker
ORDER BY rth_dollar_bn DESC
LIMIT 6| nvda_rank | nvda_dollar_bn | single_stocks_above_nvda | pct_of_leader | friday_rank | friday_nvda_rth_bn | friday_margin_over_next_bn |
|---|---|---|---|---|---|---|
| 3 | 107.1 | 1 | 65.3 | 1 | 26.56 | 0.27 |
The exact SQL behind every number
WITH (
SELECT sum(toFloat64(volume) * toFloat64(close))
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
) AS nvda_d,
(
SELECT sum(toFloat64(volume) * toFloat64(close))
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-07-10 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
) AS nvda_friday,
(
SELECT (countIf(d > nvda_friday) + 1, max(d))
FROM (
SELECT ticker, sum(toFloat64(volume) * toFloat64(close)) AS d
FROM global_markets.delayed_stocks_minute_aggs
WHERE window_start >= toDateTime('2026-07-10 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
AND ticker NOT IN ('SPCX', 'NVDA')
GROUP BY ticker
)
) AS friday_field
SELECT
countIf(d > nvda_d AND ticker != 'NVDA') + 1 AS nvda_rank,
round(nvda_d / 1e9, 1) AS nvda_dollar_bn,
countIf(d > nvda_d AND ticker NOT IN ('SPY', 'QQQ', 'NVDA')) AS single_stocks_above_nvda,
round(100 * nvda_d / max(d), 1) AS pct_of_leader,
friday_field.1 AS friday_rank,
round(nvda_friday / 1e9, 2) AS friday_nvda_rth_bn,
round((nvda_friday - friday_field.2) / 1e9, 2) AS friday_margin_over_next_bn
FROM (
SELECT ticker, sum(toFloat64(volume) * toFloat64(close)) AS d
FROM global_markets.delayed_stocks_minute_aggs
WHERE window_start >= toDateTime('2026-07-06 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
AND ticker NOT IN ('SPCX')
GROUP BY ticker
)Only MU, the week's volume monster, and SPY printed more regular-hours dollars; 1 single stock ranked above NVDA all week, and NVDA ran 65.3% of the leader's total. The Friday columns pin the intro's other claim: on Friday alone, NVDA's $26.56B ranked #1 on the entire tape, $0.27B ahead of the next ticker, a whisker, but the top spot. Basis: regular-hours dollar volume, the July 6–10 window, one reused-symbol ticker excluded (see Data notes).
Where the week leaves NVDA
The natural next question is positional: how close is NVDA to its own highs, and what did the rally actually reclaim?
| friday_close | high_52w | high_52w_date | pct_below_52w_high | low_52w | low_52w_date | close_vs_52w_low_pct | ma_50d | close_vs_ma50_pct | ma_200d | close_vs_ma200_pct | trading_days_observed |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 210.96 | 236.54 | 2026-05-14 | 10.8 | 162.02 | 2025-07-14 | 30.2 | 209.16 | 0.9 | 191.75 | 10 | 251 |
The exact SQL behind every number
WITH daily AS (
SELECT
toDate(toTimeZone(window_start, 'America/New_York')) AS et_date,
argMaxIf(toFloat64(close), window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) AS c,
maxIf(toFloat64(high), (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) AS h,
minIf(toFloat64(low), (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) AS l
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2025-07-11 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
GROUP BY et_date
HAVING countIf((toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) > 0
)
SELECT
round(argMax(c, et_date), 2) AS friday_close,
round(max(h), 2) AS high_52w,
argMax(toString(et_date), (h, et_date)) AS high_52w_date,
round(100 * (1 - argMax(c, et_date) / max(h)), 1) AS pct_below_52w_high,
round(min(l), 2) AS low_52w,
argMax(toString(et_date), (-l, et_date)) AS low_52w_date,
round(100 * (argMax(c, et_date) / min(l) - 1), 1) AS close_vs_52w_low_pct,
round(arrayAvg(arrayMap(t -> t.2, arraySlice(arrayReverseSort(t -> t.1, groupArray((et_date, c))), 1, 50))), 2) AS ma_50d,
round(100 * (argMax(c, et_date) / arrayAvg(arrayMap(t -> t.2, arraySlice(arrayReverseSort(t -> t.1, groupArray((et_date, c))), 1, 50))) - 1), 1) AS close_vs_ma50_pct,
round(arrayAvg(arrayMap(t -> t.2, arraySlice(arrayReverseSort(t -> t.1, groupArray((et_date, c))), 1, 200))), 2) AS ma_200d,
round(100 * (argMax(c, et_date) / arrayAvg(arrayMap(t -> t.2, arraySlice(arrayReverseSort(t -> t.1, groupArray((et_date, c))), 1, 200))) - 1), 1) AS close_vs_ma200_pct,
count() AS trading_days_observed
FROM dailyFriday's $210.96 close sits 10.8% below the trailing-year high of $236.54 (set 2026-05-14) and 30.2% above the trailing-year low of $162.02, printed 2025-07-14, the far edge of the window. The averages carry the sharper reading: the week ended just +0.9% above the 50-day average of $209.16, while holding +10% above the 200-day at $191.75. A +8.5% week, in other words, carried NVDA back to roughly where its own two-month average already sat, with the May high still well overhead.
The week against its trailing half-year
Was this a big week by NVDA's own standards? Both axes, return and dollar volume, computed identically for every week of 2026.
| period_start | week_return_pct | week_rth_dollar_bn |
|---|---|---|
| 2026-01-05 | -3.1 | 113.6 |
| 2026-01-12 | 2.6 | 110.2 |
| 2026-01-19 | 3 | 92.4 |
| 2026-01-26 | 2.3 | 103.2 |
| 2026-02-02 | -1.8 | 127.5 |
| 2026-02-09 | -1.1 | 114.9 |
| 2026-02-16 | 4.5 | 86.4 |
| 2026-02-23 | -5.9 | 161.9 |
| 2026-03-02 | 2.8 | 119.8 |
| 2026-03-09 | 1.9 | 126.1 |
| 2026-03-16 | -5.4 | 143 |
| 2026-03-23 | -5.5 | 127.4 |
| 2026-03-30 | 5.1 | 100.9 |
| 2026-04-06 | 6.5 | 101 |
| 2026-04-13 | 8.4 | 127.5 |
| 2026-04-20 | 4.1 | 115.1 |
| 2026-04-27 | -5.4 | 144.6 |
| 2026-05-04 | 7.9 | 123.3 |
| 2026-05-11 | 5.3 | 153.2 |
| 2026-05-18 | -6.4 | 148.9 |
The exact SQL behind every number
SELECT
toString(wk) AS period_start,
round(ret, 1) AS week_return_pct,
round(dollar_bn, 1) AS week_rth_dollar_bn
FROM (
SELECT toStartOfWeek(toDate(toTimeZone(window_start, 'America/New_York')), 1) AS wk,
uniqExact(toDate(toTimeZone(window_start, 'America/New_York'))) AS sessions,
(argMaxIf(toFloat64(close), window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) / argMinIf(toFloat64(open), window_start, (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) - 1) * 100 AS ret,
sumIf(toFloat64(close) * toFloat64(volume), (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199) / 1e9 AS dollar_bn
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA'
AND window_start >= toDateTime('2026-01-05 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
GROUP BY wk
HAVING sessions >= 3
)
ORDER BY period_startMeasured open-to-close within each week, this week's +8.5% lands at the strong edge of NVDA's 2026 table, beside the 2026-04-13 week's +8.4%, and just two weeks after the 2026-06-22 week printed -9.3%. The volume axis is the quieter surprise: this week's 107.1B sits nearer the middle of the 2026 range than its top, the 2026-06-01 week moved 167.7B on a -4.9% move. A big price week on unexceptional volume. The baseline is prior tape, computed fresh; the June deep-dive holds the monthly version.
What the tape was made of
Weekly totals say how much traded; the tape's texture says how.
| prints_m | median_print_shares | odd_lot_pct_of_prints | nbbo_updates_m | clean_two_sided_pct |
|---|---|---|---|---|
| 12.17 | 3 | 84.4 | 9.86 | 98.99 |
The exact SQL behind every number
WITH
(
SELECT (round(count() / 1e6, 2),
round(100.0 * countIf(bid_price > 0 AND ask_price > 0 AND ask_price > bid_price) / count(), 2))
FROM global_markets.cache_stocks_quotes
WHERE ticker = 'NVDA'
AND sip_timestamp >= toDateTime64('2026-07-06 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)
) AS quote_census
SELECT
round(count() / 1e6, 2) AS prints_m,
quantileDeterministic(0.5)(toFloat64(size), toUInt64(abs(sequence_number))) AS median_print_shares,
round(100.0 * countIf(size < 100) / count(), 1) AS odd_lot_pct_of_prints,
quote_census.1 AS nbbo_updates_m,
quote_census.2 AS clean_two_sided_pct
FROM global_markets.stocks_trades
WHERE ticker = 'NVDA'
AND sip_timestamp >= toDateTime64('2026-07-06 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)12.17 million prints, a median print of just 3 shares, 84.4% of prints below the hundred-share round lot. Notice the ordering: the tape printed more often than the quote moved, 12.17 million trades against 9.86 million NBBO updates (98.99% clean two-sided). At that median size, essentially no order of size arrives at an exchange whole: parents are shredded into child slices by routers and algorithms. The modern retail-fragmented market in one row.
The spread stayed boring, on purpose
A rout, a bounce, a sit-out, a surge, every excuse for liquidity to flinch. Did it?
| session | med_spread_bps | quote_updates |
|---|---|---|
| 2026-07-06 | 1.02 | 1516856 |
| 2026-07-07 | 1.03 | 2411154 |
| 2026-07-08 | 1.52 | 1762259 |
| 2026-07-09 | 0.99 | 1798944 |
| 2026-07-10 | 0.96 | 2214184 |
The exact SQL behind every number
SELECT
session,
round(quantileDeterministicIf(0.5)((toFloat64(ask_price) - toFloat64(bid_price)) / ((toFloat64(ask_price) + toFloat64(bid_price)) / 2) * 10000, toUInt64(toUnixTimestamp64Micro(sip_timestamp)), bid_price > 0 AND ask_price >= bid_price), 2) AS med_spread_bps,
count() AS quote_updates
FROM global_markets.cache_stocks_quotes
WHERE ticker = 'NVDA'
AND sip_timestamp >= toDateTime64('2026-07-06 13:30:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)
AND (toHour(sip_timestamp) * 60 + toMinute(sip_timestamp)) BETWEEN 810 AND 1199
GROUP BY toDate(toTimeZone(sip_timestamp, 'America/New_York')) AS session
ORDER BY sessionNo. The regular-hours median spread never left the low single digits of basis points: 1.02 bps Monday, 0.96 bps Friday. The detail is worth a second look: the week's widest median was Wednesday's 1.52 bps, NVDA's big UP day, not the rout, while Tuesday's rout was quoted at 1.03 bps on the week's heaviest quote traffic, 2411154 updates. In a mega-cap, stress shows up as more quoting, not wider quoting (why the spread is the number to watch); the order-type economics never changed.
News flow: attention peaked before the move
If the week had a single narrative, the news feed should carry it, but not where you might expect.
| week_articles | publishers | peak_day_date | peak_day_articles | thursday_articles | peak_minus_thursday | top_publisher | top_publisher_share_pct | mu_co_articles | amd_co_articles | ai_titled_articles |
|---|---|---|---|---|---|---|---|---|---|---|
| 86 | 3 | 2026-07-06 | 25 | 20 | 5 | The Motley Fool | 77.9 | 24 | 17 | 29 |
The exact SQL behind every number
WITH
(
SELECT (toString(d), n)
FROM (
SELECT toDate(toTimeZone(published_utc, 'America/New_York')) AS d, count() AS n
FROM global_markets.stocks_news
WHERE has(tickers, 'NVDA')
AND published_utc >= toDateTime('2026-07-06 00:00:00')
AND published_utc < toDateTime('2026-07-11 04:00:00')
GROUP BY d ORDER BY n DESC, d ASC LIMIT 1
)
) AS peak_day,
(
SELECT (JSONExtractString(publisher, 'name') AS p, count() AS n)
FROM global_markets.stocks_news
WHERE has(tickers, 'NVDA')
AND published_utc >= toDateTime('2026-07-06 00:00:00')
AND published_utc < toDateTime('2026-07-11 04:00:00')
GROUP BY p ORDER BY n DESC, p ASC LIMIT 1
) AS top_pub
SELECT
count() AS week_articles,
uniqExact(JSONExtractString(publisher, 'name')) AS publishers,
peak_day.1 AS peak_day_date,
peak_day.2 AS peak_day_articles,
countIf(toDate(toTimeZone(published_utc, 'America/New_York')) = toDate('2026-07-09')) AS thursday_articles,
peak_day.2 - countIf(toDate(toTimeZone(published_utc, 'America/New_York')) = toDate('2026-07-09')) AS peak_minus_thursday,
top_pub.1 AS top_publisher,
round(100.0 * top_pub.2 / count(), 1) AS top_publisher_share_pct,
countIf(has(tickers, 'MU')) AS mu_co_articles,
countIf(has(tickers, 'AMD')) AS amd_co_articles,
countIf(positionCaseInsensitive(title, 'artificial intelligence') > 0 OR position(title, 'AI') > 0) AS ai_titled_articles
FROM global_markets.stocks_news
WHERE has(tickers, 'NVDA')
AND published_utc >= toDateTime('2026-07-06 00:00:00')
AND published_utc < toDateTime('2026-07-11 04:00:00')86 tagged articles from 3 publishers, 77.9% from a single outlet (The Motley Fool), one aggregated feed's tagging, weighted toward retail commentary, and read as exactly that. The peak was 2026-07-06, Monday, 25 articles, ahead of Thursday's 20 on the sit-out itself. The composition reads complex-wide rather than NVDA-specific: 29 headlines carried AI framing, 24 co-tagged MU, 17 co-tagged AMD. Nothing here isolates an NVDA-specific catalyst on either swing day, cause unknown from this data, and saying so beats inventing one.
Options: 17.28 million contracts, calls in charge
The options tape is where positioning shows its shape: compared to what, at which strikes, dated when?
| contracts_traded_m | prints_m | premium_notional_busd | week_put_call_ratio | trailing_4w_put_call_ratio | expiries_traded | busiest_contract | friday_close_over_busiest_strike | friday_expiry_share_pct | busiest_back_expiry_md | call_vol_strike_above_start_pct |
|---|---|---|---|---|---|---|---|---|---|---|
| 17.28 | 2.44 | 5.09 | 0.42 | 0.6 | 27 | $210 call, expiry 2026-07-10 | 0.96 | 34.9 | 07-17 | 96.6 |
The exact SQL behind every number
WITH
(
SELECT ticker
FROM global_markets.options_trades
WHERE startsWith(ticker, 'O:NVDA') AND length(ticker) = 21
AND sip_timestamp >= toDateTime64('2026-07-06 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)
GROUP BY ticker ORDER BY sum(size) DESC, ticker ASC LIMIT 1
) AS busiest_t,
(
SELECT round(toFloat64(sumIf(size, substring(ticker, 13, 1) = 'P')) / toFloat64(sumIf(size, substring(ticker, 13, 1) = 'C')), 2)
FROM global_markets.options_trades
WHERE startsWith(ticker, 'O:NVDA') AND length(ticker) = 21
AND sip_timestamp >= toDateTime64('2026-06-08 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-03 00:00:00', 9)
) AS trailing_pc,
(
SELECT argMax(toFloat64(close), window_start)
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA' AND window_start >= toDateTime('2026-07-10 00:00:00') AND window_start < toDateTime('2026-07-11 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
) AS friday_close,
(
SELECT argMax(toFloat64(close), window_start)
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'NVDA' AND window_start >= toDateTime('2026-07-02 00:00:00') AND window_start < toDateTime('2026-07-03 00:00:00')
AND (toHour(window_start) * 60 + toMinute(window_start)) BETWEEN 810 AND 1199
) AS start_close,
(
SELECT concat(substring(e, 3, 2), '-', substring(e, 5, 2))
FROM (
SELECT substring(ticker, 7, 6) AS e, sum(size) AS v
FROM global_markets.options_trades
WHERE startsWith(ticker, 'O:NVDA') AND length(ticker) = 21
AND substring(ticker, 7, 6) > '260710'
AND sip_timestamp >= toDateTime64('2026-07-06 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)
GROUP BY e ORDER BY v DESC, e ASC LIMIT 1
)
) AS back_expiry
SELECT
round(sum(size) / 1e6, 2) AS contracts_traded_m,
round(count() / 1e6, 2) AS prints_m,
round(sum(toFloat64(price) * size) * 100 / 1e9, 2) AS premium_notional_busd,
round(toFloat64(sumIf(size, substring(ticker, 13, 1) = 'P')) / toFloat64(sumIf(size, substring(ticker, 13, 1) = 'C')), 2) AS week_put_call_ratio,
trailing_pc AS trailing_4w_put_call_ratio,
uniqExact(substring(ticker, 7, 6)) AS expiries_traded,
concat('$', toString(round(toFloat64(toUInt32OrZero(substring(busiest_t, 14, 8))) / 1000, 2)),
if(substring(busiest_t, 13, 1) = 'P', ' put', ' call'),
', expiry 20', substring(busiest_t, 7, 2), '-', substring(busiest_t, 9, 2), '-', substring(busiest_t, 11, 2)) AS busiest_contract,
round(friday_close - toFloat64(toUInt32OrZero(substring(busiest_t, 14, 8))) / 1000, 2) AS friday_close_over_busiest_strike,
round(100.0 * sumIf(size, substring(ticker, 7, 6) = '260710') / sum(size), 1) AS friday_expiry_share_pct,
back_expiry AS busiest_back_expiry_md,
round(100.0 * sumIf(size, substring(ticker, 13, 1) = 'C' AND toFloat64(toUInt32OrZero(substring(ticker, 14, 8))) / 1000 > start_close) / sumIf(size, substring(ticker, 13, 1) = 'C'), 1) AS call_vol_strike_above_start_pct
FROM global_markets.options_trades
WHERE startsWith(ticker, 'O:NVDA') AND length(ticker) = 21
AND sip_timestamp >= toDateTime64('2026-07-06 00:00:00', 9) AND sip_timestamp < toDateTime64('2026-07-11 00:00:00', 9)17.28 million contracts and $5.09B of premium across 27 expiration dates, at 0.42 puts per call. Compared to what: NVDA's trailing four full weeks ran 0.6, the call tilt got MORE extreme during the rally week, not less. At which strikes: 96.6% of call volume traded above the week's $194.51 starting close, territory the stock had not yet reached when the week opened. Dated when: 34.9% of the volume sat in Friday's weekly expiry alone, and the busiest contract, the $210 call, expiry 2026-07-10, expired with Friday's close just $0.96 above its strike: within a dollar of the week's most-traded strike. Beyond Friday, the busiest dated expiry was 07-17, the July monthly, one week out (the expiration calendar; reading the ratio).
The shorts, with the week's asterisk
FINRA's daily files track short-marked sale volume, with one file-quality caveat this week, flagged rather than hidden.
| d | short_shares_m | nvda_rows_on_file |
|---|---|---|
| 2026-07-06 | 12.7 | 1 |
| 2026-07-07 | 22.1 | 1 |
| 2026-07-08 | 24.6 | 1 |
| 2026-07-09 | 19.5 | 1 |
| 2026-07-10 | 26.6 | 1 |
The exact SQL behind every number
SELECT toString(date) AS d,
round(sum(short_volume) / 1e6, 1) AS short_shares_m,
count() AS nvda_rows_on_file
FROM global_markets.stocks_short_volume
WHERE ticker = 'NVDA' AND date >= toDate('2026-07-06') AND date <= toDate('2026-07-10')
GROUP BY date
ORDER BY dateReported daily short volume ran 12.7M shares Monday and 26.6M Friday, roughly doubling as the week's dollar volume built. Less dramatic than it sounds: these files count short-marked SALES, mostly market-maker inventory management, short interest vs short volume separates the two. One asterisk: the July 7 file arrived truncated market-wide (receipted in the weekly recap), so treat that session's figures, NVDA's included, as provisional until refiled. The twice-monthly short interest print nearest this week settled June 30; days to cover puts a short position in exit-time terms.
FAQ
How did NVDA stock do the week of July 6, 2026?
NVDA rose 8.5%, from $194.51 to $210.96 on Friday July 10, with a $211.1 high and a $190.6 low. Friday was the strongest session at +4%.
Is NVDA at a record high after the week of July 6?
No. Friday's $210.96 close sat 10.8% below its trailing-year high of $236.54, set 2026-05-14.
Is NVDA above its key moving averages?
Yes: the July 10 close stood +0.9% above the 50-day average ($209.16) and +10% above the 200-day ($191.75).
What was NVDA's put-call ratio for the week, and is it unusual?
The week traded at 0.42 puts per call, more call-heavy than NVDA's trailing four-week 0.6, with 34.9% of contracts in the Friday weekly expiry.
Did NVDA move with the rest of the chip sector this week?
Mostly not. Against AMD, MU, INTC, and TSM, it was the only green close on Tuesday's rout (+0.7% against AMD's -6.4%) and the only red on Thursday's rebound (-0.7% against AMD's +5.7%).
Data notes
- Window and clock. Timestamps are stored UTC; regular hours are the 810–1199 ET-minute band (9:30 am–4:00 pm New York). The prior close is Thursday July 2 (July 3 was a market holiday).
- Leaderboard exclusion. The dollar-volume leaderboards exclude exactly one ticker by standing house rule: a symbol the vendor feed assigns to two different companies, the deep-dive on that listing carries the identity-verification receipts. Nothing else is excluded.
- Peer basket. A declared basket, AMD, MU, INTC, TSM (the TSMC ADR), of widely traded US-listed chip names, not a vendor sector classification.
- Moving averages. Simple averages of the last fifty and last two hundred regular-session closes through July 10; trailing-year extremes use regular-hours highs and lows.
- News feed. ONE aggregated feed's tagging, not global media. No licensed earnings-calendar dataset exists in the warehouse, so no forward earnings-date claims here.
- OCC parsing. Option tickers are parsed positionally (root O:NVDA; strike in the last eight digits). Positional and ordering claims are encoded as sanity bounds.
Methodology
- Source: consolidated tape,
delayed_stocks_minute_aggs,stocks_trades+cache_stocks_quotes,options_trades, FINRAstocks_short_volume, exchange-reportedstocks_short_interest. - Deterministic aggregates:
quantileDeterministicon stable keys; tuple tie-breaks on extremes. - Prior-period comparisons: computed live from the tape, never quoted from earlier posts; the weekly baseline carries a
period_startcolumn for drift coverage. - Warehouse as-of: July 12, 2026 (T+2 for the week's final session).
Cross-links: NVDA's June 2026 deep-dive, the five dailies via the week's market recap, DTE explained for the options vocabulary. Every panel's SQL runs as-is on the Strasmore terminal.