Options / Tickers / NVDA
TICKER PAGE · NVDA
NVIDIA (NVDA)
Is protection on NVIDIA cheap or dear today, where has the money been placed, how did the past month play out, and how did the judgments we filed on NVIDIA settle? Updated after every close, with right and wrong judgments alike left on the record.
Data date 10.07 · close 238.94 (—) · −1.82% from the 52-week high
I
Where it stands today
One-year IV percentile0 / 100
Where one-month protection (30-day implied volatility) sits within its past-year range, 0 the cheapest and 100 the dearest. Protection on NVIDIA is now in the cheapest stretch of the past year.
30-day / one-year implied vol (%)30.21 / 37.88
The one-year sits 7.67 points above the 30-day, the usual shape: a longer policy covers more.
Expected move this week±1.7% ±4.11
Centred on the previous close of 239.24, option prices imply 235.13 to 243.35 at this week's expiry. A close outside it at expiry means the option sellers lose.
Put ÷ call (volume / open interest)0.66 / 1.06
Calls outtraded puts. It counts contracts, not direction: a put buyer may be insuring shares they own.
Levels from high to low, with distance from the close:
| Expected move, high | 243.35 | +1.85% | Centred on the previous close: the implied top at this week's expiry. |
| Call wall | 240 | +0.44% | The strike with the most call open interest: an upside target someone has already made a down payment on. |
| Close | 238.94 | ||
| Expected move, low | 235.13 | −1.60% | Centred on the previous close: the implied bottom at this week's expiry. |
| Put wall | 230 | −3.74% | The strike with the most put open interest: protection someone already paid for. |
| Flip level | 214.55 | −10.21% | Above it, dealer hedging damps moves; below it, hedging amplifies them. |
Data date 10.07. Updated once after each trading day's close; spot is the official close.
II
The last 30 trading days
Close, flip level, put wall and call wall, day by day: which side of the walls price is on, and whether the walls moved.
One-year and 30-day implied vol, day by day. The short line jumping while the long line holds means only the near-term event is being repriced; the long line rising too means the whole story is.
Shown: the last 30 trading days. Earlier records are in Chapter V (walls) and Chapter VII (implied vol) of the options page.
III
Today's big prints
No NVIDIA contract made today's leaderboard.
Data date only. The board for every name is in Chapter IV of the options page.
IV
Valuation and profitability
P/E (TTM)
29.2
Forward P/E
14.9
P/S
18.6
P/B
24.7
ROE
117.2%
Gross margin
74.7%
Net margin
63.7%
Dividend yield
0.43%
The latest P/E (2026.10.02) is 29.6, the 42nd percentile of 93 readings since 2000 (0 is the cheapest in that history, 100 the dearest; past readings are taken at quarter-ends). NVIDIA is compared only with its own past.
At the latest quarter-end: ROE 110.1% (2026.07.31), ROIC 94.6% (2026.07.31).
As of 2026.10.03, updated weekly. Full profit, return, valuation and dividend charts are in the Tech stock view.
V
Every case we filed on it
Each judgment is written down with its test before the fact, settled by that test, and left standing when wrong. These are all the cases that name NVIDIA, none removed. Open one for the original filing.
14cases since 07.22
13settled
1open
09.18After the second and third hikes, what gets repriced is the farthest cash flow in the expensive group , not the cheap group .Also: SNDK, MSTR, INTC, GOOGL, GSOpen · est. 11.30
Original filing and test, archived verbatim in Chinese
09.18 Filed (confidence: 中)
第二三次加息之后被重新定价的是贵的那组里现金流最远的，不是便宜的那组。贵组＝SNDK PB 16.67（一年 84% 时间便宜）/ MSTR PB 1.72（一年 89%）/ INTC PB 6.20（五年 99.8%）；便宜组＝NVDA PE 27.77（五年 99% 时间贵）/ GOOGL PE 17.51（五年 95%）/ GS PE 13.73（三年九成），均 09-18 收盘
Test: 10-28 之后一个月（到 11-30）贵组三只等权涨幅跑赢便宜组三只等权 ⇒ 收回；跑输或持平 ⇒ 成立。基准＝09-18 收盘
第二三次加息之后被重新定价的是贵的那组里现金流最远的，不是便宜的那组。贵组＝SNDK PB 16.67（一年 84% 时间便宜）/ MSTR PB 1.72（一年 89%）/ INTC PB 6.20（五年 99.8%）；便宜组＝NVDA PE 27.77（五年 99% 时间贵）/ GOOGL PE 17.51（五年 95%）/ GS PE 13.73（三年九成），均 09-18 收盘
Test: 10-28 之后一个月（到 11-30）贵组三只等权涨幅跑赢便宜组三只等权 ⇒ 收回；跑输或持平 ⇒ 成立。基准＝09-18 收盘
09.30Nvidia calls expiring Nov 20 were the mid-September position rolling up, not new bullish betsWithdrawn by its test
Settlement
Withdrawn by its test: it held only if, in October 1 morning settlement, the 230 and 275 calls each fell by more than 50000 contracts and the 245 and 295 calls each rose by more than 50000; any miss meant withdrawal. The 230 call went from 103083 to 61026 (down 42057) and the 275 call from 80576 to 33789 (down 46787), so neither old strike fell by 50000; the 245 call rose from 21327 to 92644 and the 295 call from 1774 to 71296. The positions did move up, but new positions exceeded closed ones by more than 50000, so this was not just a change of strikes
Withdrawn by its test: it held only if, in October 1 morning settlement, the 230 and 275 calls each fell by more than 50000 contracts and the 245 and 295 calls each rose by more than 50000; any miss meant withdrawal. The 230 call went from 103083 to 61026 (down 42057) and the 275 call from 80576 to 33789 (down 46787), so neither old strike fell by 50000; the 245 call rose from 21327 to 92644 and the 295 call from 1774 to 71296. The positions did move up, but new positions exceeded closed ones by more than 50000, so this was not just a change of strikes
Original filing and test, archived verbatim in Chinese
09.30 Filed (confidence: Leaning)
英伟达 11-20 到期看涨是九月中那批仓位往上滚，不是新增押涨：09-30 成交 230C 71918（持仓 103083）、275C 70590（持仓 80576），同日 245C 77447（持仓 21327）、295C 70108（持仓 1774，39.5 倍），四档张数几乎一一对应；230C/275C 即 J-0914-03 那批（09-14 成交 74676 / 71820）
Test: 10-01 晨券商结算持仓：230C 与 275C 各减少 ≥50000 张，且 245C 与 295C 各增加 ≥50000 张 ⇒ 成立；任一条不满足 ⇒ 收回。持仓不可回溯，最晚取数日 10-01
英伟达 11-20 到期看涨是九月中那批仓位往上滚，不是新增押涨：09-30 成交 230C 71918（持仓 103083）、275C 70590（持仓 80576），同日 245C 77447（持仓 21327）、295C 70108（持仓 1774，39.5 倍），四档张数几乎一一对应；230C/275C 即 J-0914-03 那批（09-14 成交 74676 / 71820）
Test: 10-01 晨券商结算持仓：230C 与 275C 各减少 ≥50000 张，且 245C 与 295C 各增加 ≥50000 张 ⇒ 成立；任一条不满足 ⇒ 收回。持仓不可回溯，最晚取数日 10-01
09.30 Corrected
补写跟踪合约：立案行（a2273e1）备注漏写跟踪合约，判断与判据不变
Result: 不变
补写跟踪合约：立案行（a2273e1）备注漏写跟踪合约，判断与判据不变
Result: 不变
10.01 Settled
收回，按判据收回：判据是 10-01 早上的结算持仓里，230C 与 275C 各减 5 万张以上、且 245C 与 295C 各增 5 万张以上才算成立，任一条不满足就收回。实际 230C 103083→61026（−42057）、275C 80576→33789（−46787），旧两档都没减到 5 万；245C 21327→92644（+71317）、295C 1774→71296（+69522）。方向确实是往上挪，但新开的比撤掉的多 5 万多张，这批仓位不只是挪了位置
Result: 收回
收回，按判据收回：判据是 10-01 早上的结算持仓里，230C 与 275C 各减 5 万张以上、且 245C 与 295C 各增 5 万张以上才算成立，任一条不满足就收回。实际 230C 103083→61026（−42057）、275C 80576→33789（−46787），旧两档都没减到 5 万；245C 21327→92644（+71317）、295C 1774→71296（+69522）。方向确实是往上挪，但新开的比撤掉的多 5 万多张，这批仓位不只是挪了位置
Result: 收回
09.24A new long-dated bet: the Nvidia Jan 2028 460 call traded 20388 contracts against 6463 open interest, a bet the stock doubles in just over two years, …confirmed
Settlement
confirmed
confirmed
Original filing and test, archived verbatim in Chinese
09.24 Filed (confidence: Leaning)
英伟达 2028-01-21 460C 一天成交 20388 张对原持仓 6463，是新开的长注：押两年多翻一倍，下在英伟达隐含波动率一年里最便宜的时候（52 周分位约 1%，低于 30 日实际波动）
Test: 09-25 晨结算 OI 较 6463 增加不到当日成交的一半（<10194）⇒ 当天进出，收回
英伟达 2028-01-21 460C 一天成交 20388 张对原持仓 6463，是新开的长注：押两年多翻一倍，下在英伟达隐含波动率一年里最便宜的时候（52 周分位约 1%，低于 30 日实际波动）
Test: 09-25 晨结算 OI 较 6463 增加不到当日成交的一半（<10194）⇒ 当天进出，收回
09.25 Settled
成立结清：09-25 晨读 09-24 结算，英伟达 2028-01-21 460C 持仓 6463 → 25300，ΔOI 18837 ÷ 当日成交 20388＝92%，远过收回线 10194。两年期、行权价两倍多的长注过夜几乎全部留下，是新开仓不是当天进出
Result: 成立
成立结清：09-25 晨读 09-24 结算，英伟达 2028-01-21 460C 持仓 6463 → 25300，ΔOI 18837 ÷ 当日成交 20388＝92%，远过收回线 10194。两年期、行权价两倍多的长注过夜几乎全部留下，是新开仓不是当天进出
Result: 成立
09.15Nvidia's roll moved the bet from year-end to earnings weekConfirmed
Settlement
Confirmed: the November 230 calls lost only 1570 contracts and NVDA closed at 228.87.
Confirmed: the November 230 calls lost only 1570 contracts and NVDA closed at 228.87.
Original filing and test, archived verbatim in Chinese
09.15 Filed (confidence: Leaning)
英伟达换仓＝把赌注从年底提前到财报周，不是新钱：12.18 245C/300C 减 5 万张，11.20 230C/275C 增 13.7 万张；230C 09-15 报 7.56 对 09-14 均价 7.13
Test: 11.20 前 NVDA 收 <200 且 230C OI 09-22 前掉一半 ⇒ 作废
英伟达换仓＝把赌注从年底提前到财报周，不是新钱：12.18 245C/300C 减 5 万张，11.20 230C/275C 增 13.7 万张；230C 09-15 报 7.56 对 09-14 均价 7.13
Test: 11.20 前 NVDA 收 <200 且 230C OI 09-22 前掉一半 ⇒ 作废
09.22 Settled
Result: 成立
Result: 成立
09.14The letter changed the overnight price and did not change the session's positionsAlso: SMH, MU, SNDK, AMD, INTC, QCOM, AVGO, ARMholds
Settlement
holds
holds
Original filing and test, archived verbatim in Chinese
09.14 Filed (confidence: Leaning)
AI 放缓那封信改变的是夜里的价，没有改变白天的仓位：SMH 跳空 −5.02 白天 +0.28（2.01x）；MU −7.10/+1.99、SNDK −6.85/+2.00、AMD −5.78/+1.47、INTC −6.92/+1.43、QCOM −6.15/+5.49 白天全被买回；NVDA −3.23/−0.13、AVGO −3.41/−1.40、ARM −8.46/−1.39 没人接。白天被买回的名单与 FT 09-11 报道 Situational Awareness 用深度价外看涨重建仓位的名单（AMD/INTC/SKHY/SNDK/CRWV）重合
Test: 09.15 SMH 开盘跳空超过 −2% 且白天不再买回 ⇒ 今日的买回是空头回补，收回。只引 每日跳空.csv EOD 17:15
AI 放缓那封信改变的是夜里的价，没有改变白天的仓位：SMH 跳空 −5.02 白天 +0.28（2.01x）；MU −7.10/+1.99、SNDK −6.85/+2.00、AMD −5.78/+1.47、INTC −6.92/+1.43、QCOM −6.15/+5.49 白天全被买回；NVDA −3.23/−0.13、AVGO −3.41/−1.40、ARM −8.46/−1.39 没人接。白天被买回的名单与 FT 09-11 报道 Situational Awareness 用深度价外看涨重建仓位的名单（AMD/INTC/SKHY/SNDK/CRWV）重合
Test: 09.15 SMH 开盘跳空超过 −2% 且白天不再买回 ⇒ 今日的买回是空头回补，收回。只引 每日跳空.csv EOD 17:15
09.15 Settled
保留：SMH 09-15 跳空 +0.94 白天 −0.82 收 +0.11，成交量 0.57x，缩量回落＝无人问津不是交货；MU +1.46/−1.06、SNDK +1.14/−2.47、INTC +1.69/−1.71、NVDA +1.14/−0.56、AVGO +0.38/−1.96；QCOM +0.21/+4.02 例外
Test: 收回条件（gap ≤−2% 且白天不买回）未触发
Result: 保留
保留：SMH 09-15 跳空 +0.94 白天 −0.82 收 +0.11，成交量 0.57x，缩量回落＝无人问津不是交货；MU +1.46/−1.06、SNDK +1.14/−2.47、INTC +1.69/−1.71、NVDA +1.14/−0.56、AVGO +0.38/−1.96；QCOM +0.21/+4.02 例外
Test: 收回条件（gap ≤−2% 且白天不买回）未触发
Result: 保留
09.14Protection sellers treated the bank guidance as new information and did not treat the letter as new informationAlso: GS, WFC, JPM, BAC, C, MU, SNDK, INTC, AMD, AVGO, TSMconfirmed
Settlement
confirmed
confirmed
Original filing and test, archived verbatim in Chinese
09.14 Filed (confidence: Leaning)
卖保护的人把银行的指引当成了新信息，没把那封信当成新信息：cm30 单日变化 GS +3.51 / WFC +2.99 / JPM +2.74 / BAC +2.39 / C +2.19 对 MU −0.04 / AMD +0.58 / AVGO +1.02 / NVDA +1.87 / TSM +2.01；cm365 MU 61.81→59.71（−2.10）、SNDK −1.61、INTC −1.01。起因＝美国银行 CEO 巴克莱会议：三季度交易收入持平、投行费用降一成
Test: 09.16 决议后 MU cm30 抬过 62 而股价不反弹 ⇒ 保护市场补定价了那封信，收回。只引 每日恒定期限IV.csv 15:45 主枪
卖保护的人把银行的指引当成了新信息，没把那封信当成新信息：cm30 单日变化 GS +3.51 / WFC +2.99 / JPM +2.74 / BAC +2.39 / C +2.19 对 MU −0.04 / AMD +0.58 / AVGO +1.02 / NVDA +1.87 / TSM +2.01；cm365 MU 61.81→59.71（−2.10）、SNDK −1.61、INTC −1.01。起因＝美国银行 CEO 巴克莱会议：三季度交易收入持平、投行费用降一成
Test: 09.16 决议后 MU cm30 抬过 62 而股价不反弹 ⇒ 保护市场补定价了那封信，收回。只引 每日恒定期限IV.csv 15:45 主枪
09.17 Settled
成立结清：MU cm30 09-17 57.86 <62（09-16 58.43）；银行三天涨的保费 09-17 一天退光，改价方向对了、持久性没有
Result: 成立
成立结清：MU cm30 09-17 57.86 <62（09-16 58.43）；银行三天涨的保费 09-17 一天退光，改价方向对了、持久性没有
Result: 成立
09.14The options market priced the letter as an entry point, not a turning pointhalf right
Settlement
half right
half right
Original filing and test, archived verbatim in Chinese
09.14 Filed (confidence: Leaning)
期权市场把那封信定价成买点，没定价成拐点：NVDA 11.20 230C 成交 74676 张（约 $53.24M，OI 32209）/ 275C 71820 张（$7.47M，OI 10918，205 笔）；12.18 245C 40023 张（$22.97M，OI 44402＝09.09 那批）/ 300C 41778 张（$3.54M，OI 66648）；四档合计约 $87.22M 全是看涨。两对同型：付低档卖高档，收益封顶在 275/300；另一读法为高档单独买的深度价外看涨
Test: 09.15 结算后 230C OI 增加不到 20000 张 ⇒ 当天来回的仓位维护，收回；245C OI 掉到 20000 以下 ⇒ 09.09 那批换仓，关于新钱的半句收回。最晚取数日 09.15（OI 不可回溯），过期后只能用当前 OI 作替代读数并标明
期权市场把那封信定价成买点，没定价成拐点：NVDA 11.20 230C 成交 74676 张（约 $53.24M，OI 32209）/ 275C 71820 张（$7.47M，OI 10918，205 笔）；12.18 245C 40023 张（$22.97M，OI 44402＝09.09 那批）/ 300C 41778 张（$3.54M，OI 66648）；四档合计约 $87.22M 全是看涨。两对同型：付低档卖高档，收益封顶在 275/300；另一读法为高档单独买的深度价外看涨
Test: 09.15 结算后 230C OI 增加不到 20000 张 ⇒ 当天来回的仓位维护，收回；245C OI 掉到 20000 以下 ⇒ 09.09 那批换仓，关于新钱的半句收回。最晚取数日 09.15（OI 不可回溯），过期后只能用当前 OI 作替代读数并标明
09.15 Settled
一半成立一半收回：11.20 230C OI 32209→100157（+67948，>20000 ⇒ 买点句成立）/ 275C 10918→80031；12.18 245C 44402→16361（<20000 ⇒ 09.09 那批换仓）/ 300C 66648→44531。到期日提前一个月、行权价压低 15、张数翻倍；净新钱约 4 万张≈$28M
Result: 一半成立
一半成立一半收回：11.20 230C OI 32209→100157（+67948，>20000 ⇒ 买点句成立）/ 275C 10918→80031；12.18 245C 44402→16361（<20000 ⇒ 09.09 那批换仓）/ 300C 66648→44531。到期日提前一个月、行权价压低 15、张数翻倍；净新钱约 4 万张≈$28M
Result: 一半成立
09.14Changed: somebody is betting on it directly, and the direction is up.confirmed
Settlement
confirmed
confirmed
Original filing and test, archived verbatim in Chinese
09.14 Settled
改判：有人对它本身下注，方向向上。NVDA −3.36% / 1.48x，cm30 +1.87，远端约 $87M 看涨（11.20 230C/275C、12.18 245C/300C）
Result: 改判
改判：有人对它本身下注，方向向上。NVDA −3.36% / 1.48x，cm30 +1.87，远端约 $87M 看涨（11.20 230C/275C、12.18 245C/300C）
Result: 改判
09.18 Settled
成立结清：「分位 >30 且续跌」未触发（09-18 iv_pctl 盘前 0.4，cm30 30.20，收 222.27 一周 +1.82%）；09-14 改判「有人押涨」并存，钱在 11.20 230C（OI 100784）
Result: 成立
成立结清：「分位 >30 且续跌」未触发（09-18 iv_pctl 盘前 0.4，cm30 30.20，收 222.27 一周 +1.82%）；09-14 改判「有人押涨」并存，钱在 11.20 230C（OI 100784）
Result: 成立
09.10Nobody is paying up at the front end of NvidiaRight
Settlement
Right
Right
Original filing and test, archived verbatim in Chinese
09.10 Filed (confidence: More likely)
英伟达近端没人加价，钱在远端：三十天 IV 三年分位 9.6 为全组最低、cm30 退到 33.10，而 12-18 到期 245C 成交 40,620 张、为持仓量四倍
Test: 09.11 盘前 期权链尺 EOD 轮结算 OI：若 245C 增量 < 成交量三成，则为当日来回，「钱去了远处」收回。⚠️ 取数窗口＝09-11 盘前，过期不可回溯
英伟达近端没人加价，钱在远端：三十天 IV 三年分位 9.6 为全组最低、cm30 退到 33.10，而 12-18 到期 245C 成交 40,620 张、为持仓量四倍
Test: 09.11 盘前 期权链尺 EOD 轮结算 OI：若 245C 增量 < 成交量三成，则为当日来回，「钱去了远处」收回。⚠️ 取数窗口＝09-11 盘前，过期不可回溯
09.11 Settled (confidence: More likely)
英伟达近端没人加价，钱在远端——终审
Test: 09.11 盘前 期权链尺 EOD 轮结算 OI：若 245C 增量 < 成交量三成，则为当日来回，收回
Result: 对。12-18 245C 结算 OI 9,924（09-10 EOD16:26）→ 44,402（09-11 EOD16:20），+34,478 张＝09-10 成交 40,620 的 85%，远超三成
英伟达近端没人加价，钱在远端——终审
Test: 09.11 盘前 期权链尺 EOD 轮结算 OI：若 245C 增量 < 成交量三成，则为当日来回，收回
Result: 对。12-18 245C 结算 OI 9,924（09-10 EOD16:26）→ 44,402（09-11 EOD16:20），+34,478 张＝09-10 成交 40,620 的 85%，远超三成
08.1908-19 was not a risk-off day: the four hardest-hit names all traded lighter than usual , so the decline came from absent buyers rather than selling pr…Also: SPY, QQQ, MSTR, COIN, AVGO, INTC, AMDRight — neither falsifying condition triggered
Settlement
Right — neither falsifying condition triggered. First: "if 08-20 brings a comparable decline on 1.5x or more volume, the no-sellers half fails." On 08-20 the four hardest-hit names printed AVGO +0.43%, RDDT -0.92%, INTC -0.72% and AMD +0.65% — none came close to the 3.71% to 4.61% declines of 08-19, so the premise never appeared and that half stands. Second: "if the high-beta group gives it all back, the long-end-and-weak-dollar half fails." MSTR rose 7.81%, COIN 7.58%, CVNA 1.53% and MARA 15.54%, while HOOD slipped 0.70% — four of five kept rising, which is not a full give-back, so that half stands as well.
Right — neither falsifying condition triggered. First: "if 08-20 brings a comparable decline on 1.5x or more volume, the no-sellers half fails." On 08-20 the four hardest-hit names printed AVGO +0.43%, RDDT -0.92%, INTC -0.72% and AMD +0.65% — none came close to the 3.71% to 4.61% declines of 08-19, so the premise never appeared and that half stands. Second: "if the high-beta group gives it all back, the long-end-and-weak-dollar half fails." MSTR rose 7.81%, COIN 7.58%, CVNA 1.53% and MARA 15.54%, while HOOD slipped 0.70% — four of five kept rising, which is not a full give-back, so that half stands as well.
Original filing and test, archived verbatim in Chinese
08.19 Filed (confidence: Leaning)
08-19 不是避险日：跌得最狠的四只全部缩量（0.68~0.94倍），下跌来自买盘缺席而非卖压；当天真正的定价事件是财政部扩大长债回购，不是 FOMC 纪要
Test: 次日两条各自可推翻：①若 08-20 出现同等跌幅配 1.5 倍以上量能，则「没人卖」这半不成立；②若高贝塔（MSTR/COIN/CVNA/MARA/HOOD）次日全数回吐，则「长端下行与美元走弱驱动」这半不成立。只引 每日跳空.csv 的 chg_pct 与本线量能倍数（当日量÷前20日均量）
08-19 不是避险日：跌得最狠的四只全部缩量（0.68~0.94倍），下跌来自买盘缺席而非卖压；当天真正的定价事件是财政部扩大长债回购，不是 FOMC 纪要
Test: 次日两条各自可推翻：①若 08-20 出现同等跌幅配 1.5 倍以上量能，则「没人卖」这半不成立；②若高贝塔（MSTR/COIN/CVNA/MARA/HOOD）次日全数回吐，则「长端下行与美元走弱驱动」这半不成立。只引 每日跳空.csv 的 chg_pct 与本线量能倍数（当日量÷前20日均量）
08.26 Settled (confidence: Leaning)
08-19 不是避险日：跌得最狠的四只全部缩量（0.68~0.94倍），下跌来自买盘缺席而非卖压；当天真正的定价事件是财政部扩大长债回购，不是 FOMC 纪要——终审
Test: 次日两条各自可推翻：①若 08-20 出现同等跌幅配 1.5 倍以上量能，则「没人卖」这半不成立；②若高贝塔（MSTR/COIN/CVNA/MARA/HOOD）次日全数回吐，则「长端下行与美元走弱驱动」这半不成立。只引 每日跳空.csv 的 chg_pct 与本线量能倍数（当日量÷前20日均量）
Result: 对(两条推翻条件均未触发)。①「若 08-20 出现同等跌幅配 1.5 倍以上量能则『没人卖』半不成立」——08-19 跌最狠四只在 08-20 的表现为 AVGO +0.43%、RDDT -0.92%、INTC -0.72%、AMD +0.65%，无一只出现同等跌幅(08-19 为 -3.71~-4.61%)，前提未出现 ⇒ 该半未被推翻。②「若高贝塔全数回吐则『长端下行与美元走弱驱动』半不成立」——MSTR +7.81%、COIN +7.58%、CVNA +1.53%、MARA +15.54%、HOOD -0.70%，五只中四只继续上涨、仅 HOOD 微跌 ⇒ 非「全数回吐」，该半亦未被推翻。
08-19 不是避险日：跌得最狠的四只全部缩量（0.68~0.94倍），下跌来自买盘缺席而非卖压；当天真正的定价事件是财政部扩大长债回购，不是 FOMC 纪要——终审
Test: 次日两条各自可推翻：①若 08-20 出现同等跌幅配 1.5 倍以上量能，则「没人卖」这半不成立；②若高贝塔（MSTR/COIN/CVNA/MARA/HOOD）次日全数回吐，则「长端下行与美元走弱驱动」这半不成立。只引 每日跳空.csv 的 chg_pct 与本线量能倍数（当日量÷前20日均量）
Result: 对(两条推翻条件均未触发)。①「若 08-20 出现同等跌幅配 1.5 倍以上量能则『没人卖』半不成立」——08-19 跌最狠四只在 08-20 的表现为 AVGO +0.43%、RDDT -0.92%、INTC -0.72%、AMD +0.65%，无一只出现同等跌幅(08-19 为 -3.71~-4.61%)，前提未出现 ⇒ 该半未被推翻。②「若高贝塔全数回吐则『长端下行与美元走弱驱动』半不成立」——MSTR +7.81%、COIN +7.58%、CVNA +1.53%、MARA +15.54%、HOOD -0.70%，五只中四只继续上涨、仅 HOOD 微跌 ⇒ 非「全数回吐」，该半亦未被推翻。
08.19The market is buying one date — 08-26 — rather than a stretch of volatilityPartly
Settlement
Partly.
Partly.
Original filing and test, archived verbatim in Chinese
08.19 Filed (confidence: More likely)
市场买的是 08-26 这一个日期而不是一段波动：定价跳跃整个压在跨过财报那一档上，没有随期限铺开
Test: 08-26 财报后看两件事：①08-28 那一档的实际波动是否落在冻结带 207.82–231.66 内；②若 08-31 与 09-18 的 IV 在财报前显著追上 08-28（差距收窄到 0.03 以内），则「集中在单日」这个读法被削弱。只引 09:09 盘前冻结那组读数，不与盘中盘后混比
市场买的是 08-26 这一个日期而不是一段波动：定价跳跃整个压在跨过财报那一档上，没有随期限铺开
Test: 08-26 财报后看两件事：①08-28 那一档的实际波动是否落在冻结带 207.82–231.66 内；②若 08-31 与 09-18 的 IV 在财报前显著追上 08-28（差距收窄到 0.03 以内），则「集中在单日」这个读法被削弱。只引 09:09 盘前冻结那组读数，不与盘中盘后混比
08.28 Settled (confidence: More likely)
市场买的是 08-26 这一个日期而不是一段波动：定价跳跃整个压在跨过财报那一档上，没有随期限铺开——终审
Test: ①08-28 那一档实际波动 vs 冻结带 207.82–231.66；②08-31 与 09-18 的 IV 在财报前是否追上 08-28（差距收窄到 0.03 以内则读法被削弱）。只引 09:09 盘前冻结那组读数
Result: 部分
市场买的是 08-26 这一个日期而不是一段波动：定价跳跃整个压在跨过财报那一档上，没有随期限铺开——终审
Test: ①08-28 那一档实际波动 vs 冻结带 207.82–231.66；②08-31 与 09-18 的 IV 在财报前是否追上 08-28（差距收窄到 0.03 以内则读法被削弱）。只引 09:09 盘前冻结那组读数
Result: 部分
08.17The gamma flip has stopped between 200 and 210 for three readings running, which says the market's center of gravity for pricing NVDA has not movedThe test itself was unfit, so no verdict is given and the ca…
Settlement
The test itself was unfit, so no verdict is given and the case is refiled. Read literally it settled on 08-24 (flip at 199.09, leaving the band on the downside), but across four days it never left by more than $1.23: -0.91 on 08-24, -1.23 on 08-25, -0.20 on 08-26 and -0.05 on 08-27, and by the 08-28 pre-open it was back at 205.96, up 5.96. Over the same stretch the stock ran from 208.48 to 227.98 while the flip barely moved. For a reading that oscillates about a dollar around the boundary, "leaves the band" is not a decidable test.
The test itself was unfit, so no verdict is given and the case is refiled. Read literally it settled on 08-24 (flip at 199.09, leaving the band on the downside), but across four days it never left by more than $1.23: -0.91 on 08-24, -1.23 on 08-25, -0.20 on 08-26 and -0.05 on 08-27, and by the 08-28 pre-open it was back at 205.96, up 5.96. Over the same stretch the stock ran from 208.48 to 227.98 while the flip barely moved. For a reading that oscillates about a dollar around the boundary, "leaves the band" is not a decidable test.
Original filing and test, archived verbatim in Chinese
08.17 Filed (confidence: More likely)
翻转位连续三期停在 200 到 210，说明市场对英伟达的定价重心未移动；值得等的是它离开区间那一天
Test: gex_flip 离开 [200,210] 的首日即开奖：向上离开＝重心上移，向下离开＝重心下移；仍在区间内则继续挂
翻转位连续三期停在 200 到 210，说明市场对英伟达的定价重心未移动；值得等的是它离开区间那一天
Test: gex_flip 离开 [200,210] 的首日即开奖：向上离开＝重心上移，向下离开＝重心下移；仍在区间内则继续挂
08.28 Settled (confidence: More likely)
翻转位连续三期停在 200 到 210，说明市场对英伟达的定价重心未移动；值得等的是它离开区间那一天——终审
Test: gex_flip 离开 [200,210] 的首日即开奖：向上离开＝重心上移，向下离开＝重心下移；仍在区间内则继续挂
Result: 判据不合格，不给结论，重立。按字面 08-24 即已开奖（gex_flip 199.09，向下离开）,但四天里最远只离开 1.23 美元：08-24 为 -0.91、08-25 为 -1.23、08-26 为 -0.20、08-27 为 -0.05，而 08-28 盘前又回到 205.96（+5.96）。同期正股从 208.48 涨到 227.98，翻转位纹丝不动。对一个每天在边界上下摆动一美元的读数，「离开区间」这四个字不构成可判定的判据。
翻转位连续三期停在 200 到 210，说明市场对英伟达的定价重心未移动；值得等的是它离开区间那一天——终审
Test: gex_flip 离开 [200,210] 的首日即开奖：向上离开＝重心上移，向下离开＝重心下移；仍在区间内则继续挂
Result: 判据不合格，不给结论，重立。按字面 08-24 即已开奖（gex_flip 199.09，向下离开）,但四天里最远只离开 1.23 美元：08-24 为 -0.91、08-25 为 -1.23、08-26 为 -0.20、08-27 为 -0.05，而 08-28 盘前又回到 205.96（+5.96）。同期正股从 208.48 涨到 227.98，翻转位纹丝不动。对一个每天在边界上下摆动一美元的读数，「离开区间」这四个字不构成可判定的判据。
08.17In the same-day-expiry layer, the overwhelming majority of volume has nothing to do with directionAlso: SPY, QQQ, MU, META (13 names in all)Not determinable
Settlement
Not determinable. The filing hard-coded its test as "only rows in the daily contract observation log whose source is the quote-screen OCR feed," and that source recorded nothing from 2026-08-18 through the settlement date — a direct check of 08-18 to 08-26 returns zero rows. The input the test required was never produced, so this case physically cannot be settled.
Not determinable. The filing hard-coded its test as "only rows in the daily contract observation log whose source is the quote-screen OCR feed," and that source recorded nothing from 2026-08-18 through the settlement date — a direct check of 08-18 to 08-26 returns zero rows. The input the test required was never produced, so this case physically cannot be settled.
Original filing and test, archived verbatim in Chinese
08.17 Filed (confidence: Leaning)
当日到期这一层，绝大部分成交量与方向无关：13 只里 11 只最热档是看跌，12 只行权价贴在收盘价 ±1% 内
Test: 后续三个到期日统计：最热档若系统性偏离现价（|行权价−收盘|/收盘 > 1% 的票占多数），则本判断不成立，08-17 记为单日形状；仍贴现价则加强。只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行
当日到期这一层，绝大部分成交量与方向无关：13 只里 11 只最热档是看跌，12 只行权价贴在收盘价 ±1% 内
Test: 后续三个到期日统计：最热档若系统性偏离现价（|行权价−收盘|/收盘 > 1% 的票占多数），则本判断不成立，08-17 记为单日形状；仍贴现价则加强。只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行
08.26 Settled (confidence: Leaning)
当日到期这一层，绝大部分成交量与方向无关：13 只里 11 只最热档是看跌，12 只行权价贴在收盘价 ±1% 内——终审
Test: 后续三个到期日统计：最热档若系统性偏离现价（|行权价−收盘|/收盘 > 1% 的票占多数），则本判断不成立，08-17 记为单日形状；仍贴现价则加强。只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行
Result: 无法判定。立案时把判据写死为「只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行」，而该来源自 2026-08-18 起至开奖日一行未入库(实查 08-18~08-26 共 0 行)⇒ 判据所需的输入从未产生，本判断在物理上无法执行开奖。
当日到期这一层，绝大部分成交量与方向无关：13 只里 11 只最热档是看跌，12 只行权价贴在收盘价 ±1% 内——终审
Test: 后续三个到期日统计：最热档若系统性偏离现价（|行权价−收盘|/收盘 > 1% 的票占多数），则本判断不成立，08-17 记为单日形状；仍贴现价则加强。只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行
Result: 无法判定。立案时把判据写死为「只引 每日合约观察日志.csv 里 source=行情尺截图OCR 的行」，而该来源自 2026-08-18 起至开奖日一行未入库(实查 08-18~08-26 共 0 行)⇒ 判据所需的输入从未产生，本判断在物理上无法执行开奖。
07.22The 09-18 190C and the 10-16 180C are the same position rolled forward, not two independent new positionsRight
Settlement
Right.
Right.
Original filing and test, archived verbatim in Chinese
07.22 Filed (confidence: More likely)
0918 190C⇄1016 180C 为同一持仓移仓(非独立新仓)
Test: 次晨双腿 OI:旧腿应减~80k且新腿应增~88k
0918 190C⇄1016 180C 为同一持仓移仓(非独立新仓)
Test: 次晨双腿 OI:旧腿应减~80k且新腿应增~88k
07.23 Settled
移仓假说
Result: 对
移仓假说
Result: 对
Cases for every name are in Chapter III of the options page. The ledger is append-only; a revision is a new row.
VI
Other names
Decades of NVIDIA price history and drawdowns are in The Archive · Tech.
SPY
QQQ
IWM
TLT
SMH
XLF
XLK
XLV
AAPL
AMZN
GOOG
GOOGL
META
MSFT
TSLA
SPCX
AVGO
TSM
AMD
INTC
MU
SNDK
MRVL
QCOM
SKHY
SMCI
DELL
ORCL
PLTR
CRWV
NBIS
COIN
HOOD
MSTR
SOFI
JPM
BAC
C
WFC
GS
KRE
V
MA
AXP
MS
SCHW
BLK
BRK.B
CRCL
WMT
COST
HD
TJX
MCD
KO
RACE
MC.PA
RMS.PA
How to read this page
· The one-year percentile and the 30-day and one-year implied vols come from two different sources; each is compared only with its own past, and the two are never spliced together.
· The 30-day and one-year vols are constant-maturity: interpolated between the two nearest expiries to exactly 30 and 365 days, read once before each close.
· A big-print row is one contract's full-day total, not a single trade. Without per-trade timestamps or aggressor side we state premium traded only.
· Every reading is cross-checked against two independent sources and updated once after the close. Raw data is not provided on this site.
· Valuation ratios, margins and returns are computed from the company's filed quarterly and annual reports, on the same basis as the sector stock view, and updated weekly.
· The P/E percentile compares the company only with its own quarter-end record, never with other companies, and is not spliced with the current readings above.
· Data and information only; not investment advice; no buy or sell recommendations.
· The one-year percentile and the 30-day and one-year implied vols come from two different sources; each is compared only with its own past, and the two are never spliced together.
· The 30-day and one-year vols are constant-maturity: interpolated between the two nearest expiries to exactly 30 and 365 days, read once before each close.
· A big-print row is one contract's full-day total, not a single trade. Without per-trade timestamps or aggressor side we state premium traded only.
· Every reading is cross-checked against two independent sources and updated once after the close. Raw data is not provided on this site.
· Valuation ratios, margins and returns are computed from the company's filed quarterly and annual reports, on the same basis as the sector stock view, and updated weekly.
· The P/E percentile compares the company only with its own quarter-end record, never with other companies, and is not spliced with the current readings above.
· Data and information only; not investment advice; no buy or sell recommendations.
Today's read on this name is in the subscriber email; past issues go public the next day.