Wheel Research Log for Hermes Chat
> 用途：Hermes chat research memory / decision log
>
> 本文件係俾 Hermes Agent 記錄 Wheel 討論、候選股研究、已完成工作及限制；
> 唔係 IBKR Tools Development specification，亦唔係交易指令或自動落單設定。
Scope
呢個 chat 專門討論：
- Potential Wheel Candidates
- 宜家是否適合開 Wheel
- CSP premium、strike、breakeven、downside buffer
- Assignment 後 Covered Call 可行性
- US／HK candidates 比較
- Futu Cash Plus 抵押及資金配置限制
持倉 refresh、IBKR portfolio、現有倉位 P&L 留返另一個 chat。
User Rules and Preferences
- 候選以大型藍籌、熟悉嘅科技股、防守性消費股及金融股為主。
- Assignment 後要願意持有數年；唔係單純追求最高 premium。
- 可考慮：MSFT、GOOG、AAPL、TSM、0700、9988、KO、SBUX、HSBC 等。
- 一般油股／油價 beta 唔作優先候選；只有發電、受 Data Center 電力需求支持嘅能源標的先另行考慮。
- Wheel 可使用嘅資金上限，係 Futu Cash Plus 入面實際有幾多錢，唔係預先設定嘅固定金額。
- Cash Plus 唔可以直接當成即時可提取現金；要分開顯示 cash、Cash Plus、power 及 margin。
- 每次 Research 先 refresh Futu account，再用最新 Cash Plus amount 作 candidate collateral 上限。
- Futu API／CLI 只作 read-only research；唔會用 Futu 落單。
- 任何 candidate 必須分開驗證 Put liquidity 及 assignment 後 Covered Call liquidity。
- HKEX option contract size 必須逐隻核實，唔可以假設 100 股。
- 任何 Wheel 分析都要檢查：現價、52-week high/low、P/E、IV、delta、bid/ask、volume、OI、expiry、collateral、breakeven 及 downside buffer。
- 如果標的接近 52-week high、估值過貴或 premium 太薄，應列為 WAIT_PULLBACK，而唔係硬開。
- 所有資料要標示 source 及 timestamp；即時 quotes 會過期。
Candidate Status Labels
ACTIVE— 值得進一步查 option chainWAIT_PULLBACK— 公司質素可以，但現價／估值唔適合WATCH— 有研究價值，但未足夠資料REJECTED— 質素、波動、估值或 assignment risk 不符合要求DATA_UNAVAILABLE— 缺少可靠 Put／Call 或 contract data
Futu Account Snapshot
2026-08-27 07:06 UTC — source: ibkr futu account
Futu status READY / read-only
Account environment REAL
HKD cash 0
Cash Plus HKD 284,716.92
Futu power HKD 540,965.05
Initial margin HKD 14,236.13
Risk status LEVEL3 / SAFE
Research limit Current Futu Cash Plus amount (not fixed)Interpretation: 當時 Cash Plus 約 HK$284.7k；Wheel candidate collateral 上限應跟隨每次最新 Futu Cash Plus snapshot，唔係固定數字。唔會將 HK$540.9k power 當成可全數部署資金。
Research Log — 2026-08-27
First candidate universe
US：GOOG、MSFT、AAPL、TSM、NVDA、AMZN、META、KO、SBUX、QQQ、QQQM
HK：0700 Tencent、9988 Alibaba、1810 Xiaomi、3690 Meituan、9618 JD、0005 HSBC
Futu quote results
Symbol Price PE TTM 52W high Initial status
GOOG $339.10 17.0 $404.23 ACTIVE
MSFT $496.37 27.7 $549.20 ACTIVE / price elevated
KO $90.08 27.1 $92.49 WAIT_PULLBACK
SBUX $108.49 62.7 $109.88 WAIT_PULLBACK / valuation high
0700 HK$448.0 15.0 HK$677.70 WATCH
0005 HK$162.1 14.7 HK$168.92 WATCH / defensiveAdditional quote observations:
- 0700 Tencent: lot size 100 shares; strong stock liquidity; PE TTM about 15.
- HSBC 0005: lot size 400 shares; one board-lot not the same as option contract size; PE TTM about 14.7.
- KO: current price close to 52-week high, so do not chase CSP.
- SBUX: current price close to 52-week high and PE TTM about 62.7; not defensive enough at this valuation.
- Futu US quote was captured outside normal US session; where
lastand bid/ask differ, use live bid/ask and timestamp for option decisions.
Futu option-chain results
US chains available:
- GOOG — 2026-09-18 chain returned
- MSFT — 2026-09-18 chain returned
- KO — 2026-09-18 chain returned
- SBUX — 2026-09-18 chain returned
HK chains tested (corrected):
The earlier empty result was caused by querying invalid/non-listed HK option expiry dates (2026-09-24). It was not an API or Futu permission failure. The Futu expiration API returned valid dates including 2026-08-28, 2026-09-04, 2026-09-29, 2026-10-29, 2026-11-27, 2026-12-30, 2027-03-30, and 2027-06-29.
Using the valid 2026-09-29 expiry, the chain returned:
0700 Tencent 174 rows
0005 HSBC 140 rows
9988 Alibaba 172 rows
1810 Xiaomi 138 rows
9618 JD 138 rowsConclusion: Futu HK option-chain API and the CLI routing are working. Always query the underlying's expiration list first, then pass an exact returned strike_time to ibkr futu chain. Do not infer HK expiry from US third-Friday conventions or manually invent a date.
Current Research Ranking
1. GOOG — best first deep-dive: large-cap, familiar business, PE TTM about 17, option chain available.
2. MSFT — highest quality and defensive characteristics, but current price is relatively high; use lower CSP strikes rather than chase.
3. 0005 HSBC — strongest defensive/value profile among the tested HK stocks, but Futu option chain unavailable and HK contract details require verification.
4. 0700 Tencent — good quality/value and familiar, but Futu option chain unavailable in the tested query.
5. KO — high-quality defensive company, but currently too close to 52-week high.
6. SBUX — familiar and optionable, but current valuation and price position are not defensive enough.
Next Research Procedure
When continuing this research:
1. Use /root/project/kit-ibkr-v2/.venv/bin/ibkr futu commands only for Futu market/account data.
2. For the top US candidates, quote actual Put contracts around 20–45 DTE and delta approximately 0.15–0.20.
3. Calculate mid premium from bid/ask, not last price.
4. Check option IV, delta, volume, OI and bid/ask size.
5. Check Covered Call liquidity at likely assignment-stage strikes separately.
6. Compare collateral and remaining buffer against the latest Futu Cash Plus amount.
7. Record every new snapshot below with UTC timestamp, source and conclusion.
8. Keep this file focused on Hermes chat research; do not turn it into CLI implementation documentation.
Futu options market-open limitation and IBKR fallback
Futu OpenD may return a valid option chain while individual option quotes are still stale, zero-volume, unusually wide, or not fully active during the first minutes of the US session. Therefore, a first-open wide spread or zero volume is not automatically a final liquidity verdict.
Futu also rate-limits option-chain requests (observed maximum: 10 requests per 30 seconds). Do not retry in a tight loop; wait for the limit window and retry once in a bounded manner.
Required procedure:
1. Record underlying and option quote timestamps separately.
2. Check bid/ask, bid size, ask size, volume and last-trade timestamp.
3. Recheck after the opening market settles, preferably 10–15 minutes later.
4. If Futu chain/quotes are empty, stale, rate-limited or not active, use the IBKR option chain/quote path as a read-only fallback when Gateway and data entitlement are available.
5. Label source and timestamp clearly; align expiry, strike and option right before comparing Futu with IBKR.
6. Do not classify a candidate as AVOID based solely on the first open Futu snapshot.
The IBKR fallback is for research only and never implies order submission.
Publishing for Other Agents
呢份 Markdown 可以 publish 成公開 HTML，方便其他 agent 讀取。Project 係獨立嘅
Cloudflare Pages project：
kit-stock-research
https://kit-stock-research.pages.devPublish command
由 project root 執行：
.venv/bin/python scripts/publish_stock_research.py指定另一份 Markdown 或 project：
.venv/bin/python scripts/publish_stock_research.py \
stock-research/wheels.md --project kit-stock-researchScript 會：
1. 讀取 Markdown；
2. 轉成 self-contained HTML；
3. 呼叫 kit_ibkr.report.cloudflare_deploy.deploy_html；
4. 用 Wrangler deploy 到 Cloudflare Pages；
5. 輸出 production URL。
首次建立新 project（只需做一次）：
wrangler pages project create PROJECT_NAME --production-branch mainVerify published page
Browser harness 未必有 Chrome；可使用已安裝嘅 Lightpanda 或 Obscura。Obscura
request 必須指定 finite timeout，避免 hanging：
/opt/lightpanda/lightpanda fetch \
'https://kit-stock-research.pages.dev/' --dump markdown
timeout 30 /opt/obscura/obscura fetch --timeout 30 \
--dump original 'https://kit-stock-research.pages.dev/'如果 read-back 收到 403，唔好當成內容驗證成功；改用 Wrangler deployment list
確認 deployment 狀態：
wrangler pages deployment list --project-name kit-stock-research將 production URL、deployment URL、status 同任何 403／verification limitation
記返落本文件嘅新 snapshot。Publish 只係公開 research document，唔包含 credentials，
亦唔會觸發任何 IBKR／Futu order。
Portfolio concentration preference
Wheel 唔要求一定 hold 多隻，亦唔要求刻意分散。Kit prefers fewer operations and may concentrate on one underlying he genuinely wants to own after assignment. Single-stock concentration must be reported as a risk factor, but it is not an automatic rejection or hard requirement. Candidate ranking should prioritise assignment quality, option liquidity, acceptable return and minimal-management suitability.
Reddit: CSP LEAPS discussion
Link: https://www.reddit.com/r/thetagang/comments/1vzhb4t/csp_leaps/
Read via: Obscura rendered public Reddit page
Reviewed: 2026-08-27
Summary
The discussion compared long-dated CSPs (LEAPS) with weekly/monthly or 30–45 DTE CSPs. The original poster considered selling a MU September 2027 $800 put for approximately $15,000 premium, presented as roughly 19% annualised return.
Arguments supporting LEAPS CSPs:
- Large upfront premium with a lower strike than a shorter-dated trade.
- Potentially less frequent management and fewer roll decisions.
- Can suit an investor who genuinely wants to own the underlying for the long term.
- Long-dated options may have less short-term mark-to-market noise than very short-dated options.
Arguments against LEAPS CSPs:
- Buying power is locked for a long period and cannot be pivoted easily.
- The underlying's valuation, business thesis and market regime can change materially before expiry.
- Shorter DTE trades may generate more cumulative premium, but require more active management and can produce more assignment/roll decisions.
- A large upfront premium and annualised percentage can hide substantial tail risk.
- The discussion specifically warned against selling a long-dated CSP on a cyclical stock near a high; a high premium is not proof of a safe entry.
Hermes conclusion for Kit's Wheel framework
- LEAPS CSP is not automatically lower risk. Longer DTE changes the timing and mark-to-market profile, but assignment downside remains strike × contract size minus premium.
- The simple annualised formula (
premium / collateral × 365 / DTE) is a static yield estimate, not a guaranteed or risk-adjusted return. It must be adjusted mentally for locked capital, opportunity cost, gap risk and funding cost. - With Cash Plus as the hard collateral cap, one US GOOG CSP could consume most of the available pool. That concentration is materially different from several smaller HK positions such as 0700, 9988 or 1810.
- Default DTE: 30–45 days. 2–4 months can be researched selectively. 12–24 month CSPs are exceptional cases only, requiring unusually strong valuation, long-term conviction and sufficient unused collateral.
- For every future US candidate, record
DTE efficiency,collateral concentration,funding drag,gap risk, and whether assignment is genuinely acceptable.
This external discussion is qualitative context only. Its examples and opinions are not treated as live market data or as a trading recommendation.
Operating Feedback — 2026-08-27
Kit confirmed the following framework decisions after reviewing the HK snapshot:
Four key decisions
1. Cash Plus is the hard collateral cap. Futu Cash Plus (about HKD 284.7k at this snapshot, roughly USD 36.5k) is the real collateral reference. Futu purchasing power (about HKD 540k) must not be treated as deployable cash, because doing so could turn a market drawdown into forced deleveraging.
2. Bid/ask spread is a hard liquidity filter. HSBC 0005 145P (12.2% spread, volume 41) and JD 9618 100P (10.2% spread, volume 15) demonstrate that defensive stock quality does not compensate for poor option execution.
3. Tencent versus Alibaba is a useful two-track comparison. In this sample, 0700 400P/490C represented the more balanced quality/liquidity track, while 9988 105P represented the higher-yield/higher-risk track. Do not generalise either as universally safest or the only liquid stock without a broader scan.
4. Valuation and 52-week position filters are active. KO near its 52-week high (PE about 27.1) and SBUX near its 52-week high (PE about 62.7) stay WAIT_PULLBACK rather than being chased for premium.
Pre-entry risks to record
- Single-stock concentration: one US GOOG CSP around the USD 30k–32k strike collateral range could consume approximately 85%–90% of the current HKD Cash Plus pool after FX conversion. The return must be judged against concentration, not premium alone.
- Cross-currency funding drag: before any US CSP, verify in Futu order preview whether HKD Cash Plus supports the USD option collateral directly or creates USD borrowing/margin interest. A quoted option yield is not a net yield until this is known.
- Alibaba gap risk: 9988 105P had 9.2% buffer, delta -0.193 and IV 41.9%. A 5%–8% overnight gap is plausible for a volatile China technology name; the candidate requires genuine willingness to take assignment, not just premium chasing.
Current actionable tiers (research only)
Tier Candidate Why Main risk
1 0700 400P HKD40k collateral; 10.8% buffer; liquid Lower ~5.5% annualised return
2 9988 105P HKD52.5k collateral; ~17.4% annualised Gap risk; IV 41.9%; delta -0.193
US GOOG 300-315P Quality/PE, option liquidity Single-stock concentration; FX interestThese are research tiers, not orders or recommendations to enter now. Any next-stage comparison must show net return after funding costs, remaining Cash Plus buffer, and assignment scenario.
HK Open Research Snapshot — 2026-08-27 07:27 UTC
Source: ibkr futu quote, ibkr futu metrics, ibkr futu account
Session: Hong Kong market open / intraday
Cash Plus available for research: HKD 284,714.37
Futu power: HKD 540,957.31
Important: Cash Plus is the research collateral ceiling; power is not treated as the deployable ceiling.
KPI lines
Code Price PE TTM 52W range MA20 MA50 RSI14 RV20D Status
0700 HK$448.4 15.1 HK$411.8-672.2 460.91 453.62 46.49 30.37% WATCH
9988 HK$115.7 27.2 HK$89.5-184.1 122.16 112.35 46.02 53.11% WATCH / volatile
1810 HK$27.44 18.6 HK$21.4-59.5 27.31 26.19 50.71 47.94% WATCH / volatile
9618 HK$111.9 17.7 HK$93.0-138.9 120.05 114.80 40.09 46.25% WATCH / pullback
0005 HK$161.7 14.6 HK$93.3-167.4 162.34 156.93 53.48 19.82% BEST defensive
3690 HK$77.4 negative HK$64.3-122.3 87.98 82.41 31.72 33.11% OVERSOLD / cautionWheel return and liquidity
- 0700 Tencent: stock quality and valuation are reasonable; price remains below MA20/MA50. The corrected Futu HK chain test returned 174 contracts for 2026-09-29; option-level quotes and liquidity still need to be measured.
- 9988 Alibaba: valuation is moderate but 20-day realised volatility is high at 53.11%; not a defensive first choice until Put buffer and premium justify the volatility. Corrected chain test returned 172 contracts for 2026-09-29.
- 1810 Xiaomi: price is close to MA20 and realised volatility is 47.94%; familiar candidate, but assignment exposure is more volatile. Corrected chain test returned 138 contracts for 2026-09-29; contract size and liquidity still need checking.
- 9618 JD: below MA20/MA50 with RSI 40.09 and PE TTM about 17.7; potentially better entry zone. Corrected chain test returned 138 contracts for 2026-09-29; option liquidity and post-assignment Call still need measuring.
- 0005 HSBC: lowest realised volatility in this group (19.82%), PE TTM about 14.6, and price only moderately below 52W high. Strongest defensive/value candidate. Corrected chain test returned 140 contracts for 2026-09-29; verify contract size and liquidity before calculating returns.
- 3690 Meituan: RSI 31.72 and below MA20/MA50 may indicate oversold conditions, but PE is negative and the business is not a defensive assignment candidate. Keep out of the first Wheel group.
Return-rate status: HK option chains are now confirmed available. CSP annualised-return KPIs are still pending because the current snapshot only collected chain membership, not each contract's bid/ask, IV, delta, volume and OI. Do not substitute stock dividend yield for option premium return.
HK Option-Level Research — 2026-08-27 07:31 UTC
Source: ibkr futu chain + ibkr futu quote, expiry 2026-09-29 (33 DTE). Mid = (bid + ask) / 2. Annualised premium return = mid × contract size / collateral × 365 / DTE. These are indicative research figures, not execution quotes.
Under Leg Strike Bid-Ask Mid Delta IV OI Vol Size Collat Buffer AnnRet Spread
0700 P 400 1.96-2.02 1.99 -0.097 31.1% 30,086 519 100 40,000 10.8% 5.5% 3.0%
0700 P 380 0.82-0.90 0.86 -0.044 33.9% 8,102 46 100 38,000 15.3% 2.5% 9.3%
0700 C 490 4.30-4.41 4.355 0.194 31.3% 20,705 2,191 100 - 9.3% - 2.5%
0005 P 145 0.46-0.52 0.49 -0.078 27.4% 1,486 41 400 58,000 10.3% 3.7% 12.2%
0005 P 137.5 0.22-0.31 0.265 -0.040 32.5% 1,955 0 400 55,000 15.0% 2.1% 34.0%
0005 C 177.5 0.30-0.34 0.32 0.075 20.0% 685 1 400 - 9.8% - 12.5%
9988 P 105 1.63-1.68 1.655 -0.193 41.9% 9,621 573 500 52,500 9.2% 17.4% 3.0%
9988 P 97.5 0.61-0.65 0.63 -0.084 44.5% 12,271 1,983 500 48,750 15.7% 7.1% 6.3%
9988 C 127.5 2.24-2.28 2.26 0.261 43.8% 5,147 459 500 - 10.2% - 1.8%
1810 P 25 0.44-0.46 0.45 -0.210 43.1% 17,511 321 1000 25,000 8.9% 19.7% 6.7%
1810 P 23 0.14-0.17 0.155 -0.085 45.8% 6,872 19 1000 23,000 16.2% 7.5% 19.4%
1810 C 30 0.61-0.63 0.62 0.286 45.1% 13,329 1,267 1000 - 9.3% - 3.2%
9618 P 100 0.56-0.62 0.59 -0.109 32.2% 8,442 15 500 50,000 10.6% 6.5% 10.2%
9618 P 95 0.26-0.30 0.28 -0.053 35.4% 2,658 1 500 47,500 15.1% 3.3% 14.3%
9618 C 122.5 0.89-0.95 0.92 0.176 30.0% 1,019 24 500 - 9.5% - 6.5%Interpretation
- 0700 400P: good liquidity and 10.8% buffer, but annualised premium only about 5.5%. The 490C has active volume/OI and a relatively tight spread; Tencent is the cleanest quality/liquidity candidate in this sample.
- 0005 145P: defensive and low IV, but premium return only about 3.7%; Put/Call spreads and volume are weak. Good quality, not attractive income at this expiry.
- 9988 105P: about 17.4% annualised premium with 9.2% buffer, 573 volume and 9,621 OI; strongest income/liquidity combination, but 41.9% IV and delta -0.193 mean higher assignment volatility.
- 1810 25P: about 19.7% annualised premium and 321 volume, but delta -0.210, 43.1% IV and only 8.9% buffer. Attractive premium is compensation for meaningful Xiaomi volatility; not the defensive first choice.
- 9618 100P: 10.6% buffer and 6.5% annualised return, but only 15 volume and 10.2% spread. Liquidity is not strong enough for priority ranking.
Preliminary HK ranking
Income + liquidity 9988 105P (higher risk) > 0700 400P
Defensive quality 0700 > 0005
Higher-volatility play 1810 25P (not defensive)
Avoid for now 9618 (thin volume/wide spread)Covered Call quotes shown above are only first samples. A complete Wheel decision still needs assignment-stage Call selection at the actual assigned cost basis, dividend/exercise terms, and Futu order-preview collateral confirmation.
Three-Way Shortlist — GOOG vs 0700 vs 9988
Based on Kit's preference for Google, willingness to accept concentration, and desire to minimise the number of Wheel operations, retain these three candidates as the active shortlist:
Candidate Role Strength Main trade-off
GOOG US quality Wheel Best long-term conviction US$30.5k collateral; ~4.7% gross
0700 HK balanced Wheel Best liquidity/capital balance ~5.5% gross; lower income
9988 HK income Wheel Highest quoted income China gap risk; wide quote riskDecision framing:
- GOOG: promote to the primary US candidate. A single-stock concentration warning is required, but concentration is not an automatic rejection because Kit does not require diversification and prefers fewer operations. Confirm IBKR live bid/ask, Futu HKD-to-USD collateral treatment, USD funding interest and earnings timing before entry.
- 0700 Tencent: retain as the most balanced HK candidate. The 400P uses approximately HKD 40,000 collateral, has roughly 10.7% buffer and deep tested OI/volume. Its principal weakness is modest annualised income.
- 9988 Alibaba: retain as the higher-income HK candidate, not as the defensive candidate. The 105P has approximately 9.1% buffer, delta around -0.17 and higher IV. Quote freshness and executable spread must be reconfirmed: a wide bid/ask can make the earlier 17.4% annualised mid-price estimate unreliable.
Priority by long-term conviction: GOOG > 0700 > 9988
Priority by capital efficiency: 0700 > 9988 > GOOG
Priority by minimal operations: GOOG > 0700 > 9988This shortlist is for research comparison only; no order is implied.
After approximately one hour of the US session, Futu underlying quotes were more stable, but the option-level research was cross-checked against IBKR because Futu option quotes can refresh asynchronously during the opening window. Futu chain membership was available; IBKR was used as the final live bid/ask reference where available.
Cash Plus cap HKD 284,721.98 / about USD 36,322
Futu power HKD 540,971.78 (not deployable cash)
Symbol Price PE TTM 52W status Research read
GOOG 336.49 16.9 -16.8% from high Best quality; capital concentration
MSFT 504.97 28.1 -8.1% from high Good company; collateral too large
KO 89.41 26.8 -3.3% from high Defensive; premium too thin
SBUX 108.07 62.5 -1.6% from high Overvalued; poor option liquidity
BA 210.08 75.6 -17.4% from high Turnaround; wide option spreads
QQQM 295.74 N/A -3.9% from high Diversified; low option returnIBKR 2026-09-18 option samples (22 DTE):
Under Put Bid-Ask Mid Buffer Collateral GrossAnn Read
GOOG 305P 0.82-0.91 0.865 9.4% USD30,500 4.7% quality, concentration
MSFT 440P 0.57-0.62 0.595 12.9% USD44,000 2.2% exceeds cap
KO 80P 0.04-0.06 0.050 10.5% USD8,000 1.0% too thin
SBUX 95P 0.11-0.18 0.145 12.1% USD9,500 2.5% thin / poor liquidity
BA 190P 0.49-0.71 0.600 9.6% USD19,000 5.2% spread too wide
QQQM 265P 0.35-0.55 0.450 10.4% USD26,500 2.8% quote quality mixedConclusion: there are good candidates, but no perfect US setup in this snapshot. GOOG is the strongest quality candidate but consumes about 84% of the Cash Plus cap. MSFT exceeds the cap. KO is defensive but pays too little. BA remains a speculative turnaround Wheel. This does not invalidate the HK candidates: 0700 remains the balanced/liquidity candidate and 9988 remains the higher-income/higher-risk candidate. Future reports must distinguish “no perfect setup” from “no good stock”.
Next comparison step: at US market open, repeat the same KPI format for GOOG, MSFT, KO, SBUX, and selected blue-chip candidates. For US options, record expiry, strike, bid, ask, mid, IV, delta, OI, volume, collateral, breakeven, downside buffer, annualised premium return, and Covered Call liquidity.