Altair
Skip to content

Trading Analysis Report: TCEHY

Analysis date: 2026-10-08 · Model: claude-opus-5-5 · AI-generated, not financial advice.

  • Analysis date: 2026-10-08
  • Rating: Underweight
  • Generated: 2026-10-09 07:10:03
  • TradingAgents 0.6.0: anthropic, deep claude-opus-5-5, quick claude-opus-5-5
  • Analysts: market, sentiment, news, fundamentals; research debate rounds 5, risk debate rounds 5
  • Data vendors: core_stock_apis yfinance, technical_indicators yfinance, fundamental_data sec_edgar,yfinance, news_data yfinance, macro_data fred, prediction_markets polymarket

V. Portfolio Manager Decision

Rating: Underweight

Executive Summary: Underweight TCEHY: cut to about one-third of your standard allocation and cap the whole line at about 4.3% of the portfolio. Sell half the excess today (10-09) on a limit at the higher of 52.72 and the 0700.HK-implied price, crossing to the market in the last 30 minutes if unfilled. Rest the other half at 53.40-53.90 until the last close before the September CPI release, defaulting to Monday 10-12 if the date is unconfirmed. A daily close below 52.40 sells everything above the third the next morning; the residual exits on a confirmed break and always on a close below 51.47, there are no purchases at 52.72, and rebuilding starts only on a MACD signal-line cross plus two closes above 54.15 or one above 54.86, with no add executing before CPI, the Oct 27-28 FOMC, the Nov 3 midterms or mid-November Q3 resolves (4-6 week horizon, technical downside objective 50.22).

Current Price: 52.72

Price Target: 50.22

Confidence: Medium

Time Horizon: 4-6 weeks

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: I'll go first, and I want to set the frame before anyone else does. Trimming TCEHY is not the timid move here. It's the aggressive one. The passive move would be sitting on full size because Tencent is a great franchise and the monthly TD count printed a 9. That isn't boldness, it's hope. Aggressive capital goes where the trend and the asymmetry are, and right now neither favors a full TCEHY position.

Look at what we're holding. TCEHY closed at 52.72, the lowest close in 257 sessions and 38.9% below the 86.32 peak a year ago. Every moving average is stacked against it: 54.15, 54.84, 56.68 and 62.50. The 50-day has rejected three separate closes above it since August, and each time price fell back below within one session. The weekly SuperTrend doesn't flip until 64.96, about 23% above here, and the 200-day is sliding roughly 0.12 a session. That's dead money at best. Every dollar parked in a full-size position in a stock making yearly lows is a dollar not working in a tape where the S&P and Nasdaq just set records.

Now the asymmetry, which is what I care about most. Say we sell the excess into the 53.40 to 53.90 zone, call it 53.65.

If the break below 52.40 confirms, the ATR targets are 51.47 and then 50.22, and the data shows no support below 52.40. The trimmed slice avoids roughly 3.4 points, about 6%, and the remaining third gets cut on a daily close below 52.40.

If the exhaustion plays out instead and we close above 54.15 with the MACD histogram turning up, we rebuild. Buying back at 54.15 after selling at 53.65 costs about half a point plus slippage, under 1%, and buys us confirmation.

So we risk about half a point to avoid more than three. That's the skew I want. Compare it with the long setup the technician laid out: enter near 54.15, stop at 51.8, and you get roughly 1 to 1 to the 56.6 area. That's a coin flip into a falling 50-day line. No aggressive trader should pay full size for that.

The one real risk to the trim is an overnight gap. On 06-02 this ADR opened 10.4% higher, about 4.6 ATRs. If that happens from here, the portion we sold misses a chunk of the move. That's exactly why the plan keeps a third instead of exiting, refuses to short, and caps size by gap risk rather than stop distance. A 1% risk budget divided by a roughly 10.4% gap gives the 9.6% portfolio cap. We keep a live ticket to a violent reversal without betting the house on one.

To my conservative colleague, I expect you'll ask why we keep anything at all. You'll point to the Fed hiking to 3.88%, the 10-year at 5.28% (the highest since 2002), real yields at 2.92%, Brent at $125, and the dollar up 3%. You'll add that the Hong Kong dollar peg imports every Fed hike straight into Hong Kong funding costs, that Tencent is on the 1260H list, and that the DeepSeek ties could draw Washington scrutiny before the midterms. I agree with every one of those headwinds. But dumping the whole position at 52.72 means selling 0.29 above the lower Bollinger Band, about 1.76 standard deviations stretched. The monthly TD 9 is complete and the weekly count is at 7. MACD at -0.76 is well above its -0.97 September low, and RSI at 39.49 sits above the 36.14 it printed at a higher price. Sellers are losing speed. That's exactly where a stock snaps back a point or so, and our limit orders get paid for that bounce instead of giving it away. A full exit locks in the worst close of the year and leaves zero participation if Brent eases toward $105 or real yields drop below 2.75%. Meanwhile the yuan is stable at 6.70, so currency isn't even part of the drag. Panic exits at the lows are how portfolios fall behind.

To the neutral side, I expect you'll say exhaustion is building, sentiment reads mildly bullish at 5.8, and we should wait for a close below 52.40 before trimming, or trim less. My answer is that the one indicator tracking actual money flow hasn't confirmed any exhaustion.

OBV hit a fresh low at -193.0M on the same day as the price low. The 09-22 rally only lifted OBV to -168.3M, still below its August level, so there was no real buying behind that bounce. RSI isn't oversold. It has spent six weeks between 36 and 57, which is what RSI looks like in a downtrend, and it has room to fall to 30. There has been no capitulation either: recent down days traded 3.7 to 4.7 million shares, against 13.7 million on 05-13. Durable bottoms usually come with a washout, and we haven't had one.

Waiting for a close below 52.40 before acting means selling near 51.5 instead of near 53.65. That isn't balance. It's paying two points for confirmation on the downside while asking for none on the upside.

As for sentiment, it's eight StockTwits messages. Only three are tagged, three come from a single argument, most are China-basket chatter, and Reddit is missing entirely. That isn't a signal. And the fundamentals report contains zero verified figures, so "it's cheap" is an assertion here, not a data point. You can't justify full size on a valuation case you can't measure.

Then there's the calendar. CPI lands mid-October. The FOMC meets October 27 to 28 with the 2-year 89 basis points over fed funds, which means the market is pricing more hikes. The midterms are November 3, and Q3 results come around mid-November. That's four binary events in five weeks, with the VIX at 15 saying nobody is paying for protection. When the market is complacent heading into a cluster of catalysts, you right-size before it hits, not after.

Finally, this plan isn't a goodbye to TCEHY. It's built to turn aggressive the moment the tape earns it:

  1. A daily close above 53.11 allows a quarter-size starter with a stop on a close below 52.40, risking only about 0.7 a share.
  2. A close above 54.15 with the MACD histogram turning up triggers a rebuild toward full size.
  3. The weekly TD 9 could complete around 10-23. If that marks the turn, we're buying back at a confirmed level, with the DeepSeek round and its reported early-2027 IPO as fuel for the AI story.

The one thing I'd stress is discipline on execution. If the bounce into 53.40 to 53.90 doesn't come, the pre-CPI deadline sells regardless. Don't let a good plan rot into a hold.

So we're defensive while the trend is down and ready to be aggressive on confirmation, with dry powder instead of dead weight. That's how you beat the market: you don't marry positions, you rotate toward strength and keep the upside optionality cheap. I'm firmly behind the SELL, executed as a trim to roughly one-third of standard TCEHY allocation. Aggressive Analyst: Aggressive Analyst: We all agree on SELL, so this round decides one thing: does the plan catch the turn when it comes, or does it lock the door until mid-November? I'll concede what I got wrong, then argue the points that decide that.

First, the concessions. The 53.11 starter is gone. The conservative is right that buying 30 cents below our own sell limits is clutter, and a plan should point in one direction at a time. He's also right that rebuilding to full size on one close above 54.15 is too much. I called that entry a coin flip, and making it a full-size trigger was inconsistent. The neutral's fix is better: a close above 54.15 with the MACD histogram turning up cancels any unfilled trims and takes us back to half. I'll also accept the neutral's correction of my skew. With my own 52.40 exit, the trimmed slice saves about 1.2 to 2.1 points, not 3.4. That's still roughly 2 or 3 to 1, and I'd take that trade any day.

Neutral, you caught me pricing the slice to a target my own stop would never let it reach. Then you made the same mistake arguing for the front-load. You say waiting costs about 1.25 if price slides to 51.47. But under your own split rule, a close below 52.40 sells the excess the next session, so it never rides to 51.47. The real cost of waiting is the gap between 52.72 and the fill after a break close, somewhere from 0.3 to 1.25. Call it 0.8 in the middle. Getting the bounce into 53.40 to 53.90 is worth 0.68 to 1.18, call it 0.93. And the technical report says bounce odds are rising, with sideways chop between 52.4 and 55 the most likely case. So waiting has the better expected value.

You also said yourself that the split trigger answers the conservative's sequencing worry, because a break no longer dumps the whole position near 51.5. If the split trigger already solves sequencing, the front-load is a second payment for the same protection. Worse, it costs us in the more likely scenario, while the split trigger only costs us if the break actually happens. I'd keep the excess on the 53.40 to 53.90 limits with the pre-CPI deadline. If the desk insists on a front-load for peace of mind, make it a resting limit at the 52.72 close or better. Recent down days traded 3.7 to 4.7 million shares, so there's no reason to hit the bid. And yes, pin down the CPI date today. If nobody can confirm it, I'll accept Monday's close as the deadline.

Conservative, you called the bounce a hope. That ignores the deadline. The plan sells the excess at the pre-CPI close whether or not 53.40 trades. The worst case is selling a few sessions later at whatever the market gives us, with the 52.40 trigger underneath. 53.40 is half an ATR above the close, and price traded at 54.86 three sessions ago. That's a base case with a backstop, not a prayer.

You also said I can't use exhaustion against you and OBV against the neutral. Those two readings aren't in conflict. Together they are the thesis. Slowing momentum while volume confirms the low points to a grinding range, and in a grinding range you sell the excess into strength and keep a residual for the turn. Both facts lead to the same order ticket.

Your claim that the plan demands confirmation before selling but buys back on weaker confirmation has it backwards. The trims need no confirmation; they fill on the bounce or at the deadline regardless. The buy-back needs a close above the 10 EMA plus a MACD turn, and the second tier needs a lot more than that. The buy side is now the stricter of the two.

On the residual's stop, the neutral found the best idea in the room, and I'm adopting it. The reason to keep a residual is the weekly TD 9, and it completes only if price keeps drifting: today below 54.44, next Friday below 53.90, and the week of 10-23 below 55.65. A full exit on a close below 52.40, a quarter of an ATR away, likely throws the residual out during the very drift that creates the signal. Moving the residual's exit to a close below 51.47 adds about 0.93 of risk on a third of a position, roughly 0.31 per standard share. On a residual near 3% of portfolio, that's about five basis points. Five basis points to keep the reversal ticket alive is cheap.

The monthly 9 works the same way. Conservative, you say it only stands if October closes below 55.22. To break it, price has to rally 4.7% this month, and that rally is exactly what the residual and the unfilled limits are there to capture. Either the exhaustion signal holds or we get paid on the bounce.

On whipsaw, the neutral already did the arithmetic. The extra cost versus just holding is about a third of a point per standard share, around 0.6%. That's a toll, not a machine.

The rebuild rules are where I'll fight hardest. Your stack requires a close above 56.68, after the FOMC, ideally after Q3, with verified fundamentals and puts on top. Look at how this stock actually moves. It went from a 53.32 close on 06-26 to 60.93 on 07-08, 14% in under two weeks. It gapped 10.4% on 06-02 and 5.2% on 03-10. Upside here arrives in bursts, and calendar gates are built to miss bursts.

I'll accept the neutral's second tier with two changes in emphasis. Yes to two consecutive closes above 56.68, because the 50-day has rejected three single-day closes since August, so two closes is a real test. Yes to reading the Q2 numbers on HKEX first; that's an afternoon's work. But drop the FOMC calendar gate. The 4.8% line cap already handles event risk, and a calendar gate on top counts that risk twice.

The macro gates you're both proposing are passing today. The 10-year real yield is 2.92%, under the 3.0% gate, and high-yield spreads are 3.09%, under 3.25%. By this room's own thresholds, macro is a pass right now, not a reason to wait.

Then the gap math. Conservative, the 20% stress is your number, not the data's. In our window the gaps were 10.4% and 5.2% up, and the notable down gap was about 1.8%. The biggest earnings-type drop was 5.4% close to close on 08-12. Tencent has been on the 1260H list since January 2025, so a repeat headline is old news. The neutral recalls that the original listing cost about 7% in Hong Kong; we should verify that, but it's nowhere near 20%. I'll accept 4.8% as the hard ceiling for the whole line through Q3, because that ceiling is what lets the rebuild stay fast. Your 3.3% comes from an invented stress, and the residual sits under both numbers anyway unless standard allocation is above about 10% of the portfolio.

That 4.8% cap is also why I oppose hedging on top of it and trimming back before Q3. The cap is literally a 10.4% gap times a 0.5% budget, so it is already sized to absorb a 10.4% gap. A repeat of the 08-12 drop on a 4.8% line costs about a quarter of a percent of the portfolio. Buying KWEB or FXI puts on top pays twice for the same protection, and the neutral is right that a VIX of 15 tells us nothing about China basket volatility. I'll concede I leaned on that VIX point too. If the structure has flipped by mid-November, I want the second-tier position held into Q3 without a hedge. Waiting until after Q3 to rebuild means buying after the gap, which throws away the asymmetry.

On macro more broadly, the headwinds are real, but they're largely in the price of a stock down 38.9% from its peak and 30% this year. Core PCE ran about 2.0% annualized from June through August, so the hike looks pre-emptive. A tame CPI could compress the 89-basis-point premium of the 2-year over fed funds, and that is when real yields fall and emerging-market long-duration names snap back. Brent fell from 135.51 to 125 in four days, and the yuan is steady at 6.70.

Conservative, your own plan keeps a residual, so you don't actually believe this stock is uninvestable. I'm fine parking the trimmed capital in short paper. Fed funds at 3.88% pays us to wait. I never said pile it into Nvidia. The question is how long the door stays locked, and your answer, mid-November, means we watch our own 10-23 exhaustion window from the sidelines. And the 64% needed to get back to the high is sunk cost. It has nothing to do with the next five weeks.

DeepSeek shouldn't change sizing today; the neutral's right about that. But it's why the upside tail is fat. Tencent already uses DeepSeek models in Yuanbao and Weixin search, it's backing a $12 billion round, a reported IPO in early 2027 could crystallize value, and IGV rebounded about 50% in six months. That's the fuel if the weekly 9 marks the turn, and it's why I want a fast first tier rather than a quarantine.

So here is my plan. SELL, trimming to one-third of standard. Work the excess at 53.40 to 53.90 with a confirmed pre-CPI deadline, and if a front-load happens, do it on a limit at 52.72 or better. A close below 52.40 completes the trim the next session; the residual exits only on a close below 51.47. No buy orders while trims are working. A close above 54.15 with the MACD histogram turning up cancels the trims and takes us to half. Two closes above 56.68, the Q2 HKEX numbers read and the macro gates met take us to two-thirds of standard or 4.8% of portfolio, whichever is lower, without waiting for the FOMC, and we hold that unhedged into Q3. Full size comes after Q3. If high-yield spreads close above 3.25% or Brent tops 135, cut the residual in half. Check 0700.HK every morning.

Use position size as the risk control, not calendar gates and option premium. Get defensive while the trend is down, and be ready to move the day the tape turns. Aggressive Analyst: This is the last round, so I'll be direct. The plan is about ninety percent settled, and the settled part is good. The risk left in it isn't a gap, because the 4.8% ceiling handles gaps. The risk is that we bolt so many conditions onto the buy side that the plan can't actually buy. Then we'd spend the exhaustion window we kept the residual for watching from the sidelines. So I'll concede where I lost, then spend my time on the details that decide whether we catch the turn.

Concessions first. My expected-value math set midpoints against each other with no probabilities. It also borrowed a one-to-three-week bounce probability for an order with two or three sessions to live. You both called that, and you were right. The weekend argument is real too. Today is Friday, and a close-based trigger can't touch Monday's Hong Kong session, so I accept half the excess today on a resting limit. "Priced in" was the same unverified valuation claim I told everyone else not to make, so I withdraw it. "Macro passes today" really meant "macro hasn't failed yet." And my five basis points was a close-to-close number on a stop that actually executes a session later. Fair hits, all of them.

Now the front-load, because there's information in our own reports that nobody has used. The macro report says Golden Week ran October 1 to 8. It also says Southbound Connect buying is the main offset to Hong Kong's tightening, and Stock Connect doesn't operate while the mainland is shut. So the whole last leg traded without the buyer our own report calls the main offset. That covers the 52.88 low on 10-02, the failed push to 54.86 on 10-05, and the 52.40 low on 10-08. That bid should come back as the mainland reopens, which on the report's dates is today.

I'm not claiming that guarantees a bounce. But the plan already says to check 0700.HK before the US open, so let's use what it tells us. Set today's front-load limit at the higher of 52.72 and where 0700.HK closed, converted at the peg. That assumes the one-for-one ADR ratio, which the fundamentals report wants confirmed. If Hong Kong closed up 2%, there's no reason to rest an order about a dollar below where the primary market already priced the stock.

Conservative, be honest about what that resting limit buys. A sell limit at 52.72 only fills if Friday trades back up to 52.72. If Hong Kong gaps the ADR lower at the open and it never recovers, nothing fills and we carry the full excess into the weekend anyway. It reduces weekend gap risk; it doesn't remove it. The 52.40 trigger is still the real downside mechanism, so the front-load is no reason to pull the second half's limits lower.

Your point that the break is nearer, 0.32 away versus 0.68, compares a close with an intraday touch. The 53.40 limit fills on any print at that price. The 52.40 trigger needs a close. With an ATR of 1.25, a 0.68 move is about half a normal day's range, and this stock printed 54.86 three sessions ago. So the second half stays at 53.40 to 53.90. And someone should pull the CPI date off the BLS calendar this morning. Defaulting to Monday's close could cost us sessions of exactly the chop the technical report calls its base case.

On the residual exit, I'll take the neutral's volume rule over my flat 51.47, with two fixes from our own data.

First, the 4.5 million line sits inside the 3.7 to 4.7 million range of ordinary recent down days. A break on 4.6 million shares is a normal session for this tape, not distribution. Under that rule we'd dump the residual on an average day. Set the confirmation line at about 5.3 million, the volume on 09-23 when the September rally was rejected. That's the clearest distribution day we have.

Second, the flush bucket contradicts itself. 51.47 is only 1.8% below 52.40, and the 08-12 drop on just 6.4 million shares was 5.4%. A real 8 to 14 million share washout from here will almost certainly close below 51.47 the same day. So "halve and keep 51.47 underneath" collapses into a full exit at the capitulation low. That's the panic exit all three of us said we wanted to avoid. The fix: on a flush, halve the residual, and replace 51.47 with a close below the flush day's low as the stop on the remaining sixth. If the next sessions can't hold the washout bar, it wasn't a washout and we're out. That's a sixth of standard at risk, and it respects the report's own point that durable bottoms come with a volume surge. And if the 9 completes and the trend just keeps going, as the conservative warns, these rules take the residual out. That's the price of the ticket, and it's a third of a position.

Tier one: I take the neutral's version. That means a real MACD cross over the signal line, plus either two closes above 54.15 or one close above the 20-day. One tweak: make that single close 54.86 instead of 54.84. Two cents more, and it also clears the 10-05 high and breaks the lower-high sequence of 58.67, 58.19 and 54.86. Then the trigger is a change in price structure, not just an indicator. Conservative, that answers your whipsaw worry with something more meaningful than a second close above a falling moving average. And yes, drop the shifted trim band. It's redundant and recreates the overlap we removed.

Tier two: the neutral's substitute structure is right, either a relief signal or the FOMC behind us, not both. Conservative, "costs days, not weeks" assumes a steady climb. The only precedent off this floor went from 53.32 to 60.93 in under two weeks. Run that path from today and two closes above 56.68 land around 10-20, a full week before the meeting. The leg from 56.68 toward 61 is another 7.5%, which is exactly what tier two exists to catch. Where macro hasn't eased, I accept the gate binds.

But we need to define the oil gauge before using it as a gate. The macro report has Brent at $125 on FRED, a physical spot price it says swings more than futures in a squeeze. It also carries a headline of oil at $103 after Trump ruled out an Iran strike before the midterms. If that $103 is front-month Brent futures, one relief signal, Brent below about $105, may already be live and only needs to hold. So pick one benchmark, front-month futures, and use it for both the relief test and the cut at 135. And make "sustained" mean a weekly close for every macro gauge, as the neutral proposed for real yields.

On macro cut triggers, I agree that any trigger halves the whole line once and freezes rebuilds. That's my own double-counting argument, and I'm glad it stuck. But a freeze with no release is a quarantine. Unfreeze when the triggering gauge closes back inside its line on a weekly basis. Otherwise one bad week in October locks us out through Q3, whatever the tape does.

Q3: I accept the neutral's cap, no more than tier one going into the print, meaning half of standard or 4.8%, whichever is lower. Conservative, I won't accept trimming back to one-third, and the data says why. The biggest gaps in our window, 10.4% on 06-02 and 5.2% on 03-10, didn't land on the earnings-pattern dates of 03-18, 05-13 and 08-12. Headline gaps can come any day, which is why we sized the whole line to a gap cap instead of a calendar. If the 1260H anchor checks out at about 7%, that gap on a 4.8% line costs about a third of a percent of the portfolio. That's inside the 0.5% budget.

My recollection matches the neutral's, roughly 7% in Hong Kong, and I also recall the drop being retraced within weeks. Both go on the verification list. Two mechanics. First, the trim to tier-one size happens at the close before the print, using the date from the HKEX board-meeting notice, not weeks early. Second, "structure holds" means the first post-print close above the 56.56 to 56.68 zone re-adds tier two, with no new waiting period. The repricing happens right after the print.

On narratives, I'm done trading them. DeepSeek fattens both tails and moves size in neither direction. The upside case in this plan doesn't rest on "cheap" or "priced in." It rests on price triggers we've committed to in advance. One note: the designation the conservative fears already happened in January 2025. The new risk is outbound-investment rules, which the fundamentals report already lists for checking.

So here's my final version. SELL, trimming to one-third of standard inside the 4.8% ceiling. Half the excess goes today on a resting limit at the higher of 52.72 and the Hong Kong-implied price. The other half sits on 53.40 to 53.90 limits, with the deadline at the last close before a CPI date confirmed this morning. A close below 52.40 sells any unfilled excess the next session. After that, the residual stays on a quiet break and exits on a break above about 5.3 million shares. On a true flush it halves, with the flush day's low replacing 51.47 as its stop; otherwise 51.47 is the hard backstop. No buys while trims are working, and no shifted band.

Tier one, to half of standard, needs a real MACD cross plus two closes above 54.15 or one above 54.86. Tier two, to two-thirds or 4.8%, needs two closes above 56.68, the Q2 numbers read off HKEX, the minimum gates holding, and either real relief or the FOMC behind us. That uses one oil benchmark and weekly closes for every gauge. Into Q3 we carry no more than tier one, and we re-add on the first close above 56.68 after the print. Any macro trigger halves the line once and freezes rebuilds, which resume on a weekly close back inside the line. Every morning we check 0700.HK, Southbound flows and the HKEX buyback filings.

That's the aggressive case, and it isn't about holding more. We're selling two-thirds into weakness. Being aggressive here means holding a third through the exhaustion window, with the weekly 9 due around 10-23 and the mainland buyer back today. It also means making sure every gate has a way out, so the plan can actually buy when the tape turns. Aggressive Analyst: Aggressive Analyst: The remaining differences are order instructions now. I'd rather hand the desk a clean ticket than win a debating point. So I'll concede where the data went against me, fix four places where the rules still contradict each other, and explain why what we've built is still the aggressive plan.

Concessions first, starting with oil. Conservative, you caught me choosing a gauge for the answer it gave. Switching to front-month futures would have turned the relief signal on and the 135 cut trigger off in one move. The neutral is right that the $103 headline matches neither FRED series, so neither my relief story nor your squeeze story should rest on it. Spot at both ends until someone pulls front-month and dated Brent. One correction: the macro report defines the risk-off signal as a new high above 135, and spot printed 135.51 on 10-02. So the trigger is a weekly close above 135.51. From 125 that's an 8% rise, a real move.

On the FOMC gate, the neutral priced it and I accept the price. Tier two adds a sixth of standard, and missing the whole 56.68-to-61 leg costs about 1.3% of a standard position. Conservative, your rate-regime point is fair too. The 06-26 run happened with the 10-year roughly 70 basis points lower, and one precedent from an easier world isn't a forecast. I'm dropping that fight.

On the post-Q3 re-add, I accept two closes. Now that the line is the live 50-day rather than a fixed 56.68, the second close costs one session against a lower bar. And the 09-22 pop, which gave back two points the next day on 5.29 million shares, is exactly what a single close would have bought us.

On the front-load, I accept crossing to the inside market with half an hour left. I said it myself: a resting limit alone leaves us holding the excess exactly when we most need it sold. The Hong Kong-implied limit sets the price on a good day, and the cross guarantees the fill on a bad one. The ratio check the neutral suggested takes five minutes. I'd use the same mechanics for the session after a 52.40 break: open at the Hong Kong-implied price, cross in the last half hour. Cuts execute on schedule, and they shouldn't execute sloppily.

On Q3, I accept the homework condition. I called it an afternoon's work, so I can't refuse it as a gate. The homework cuts both ways, though. I withdrew "cheap" because we had no numbers. The daily HKEX buyback returns are the one number that tells us whether Tencent's own cash thinks 52 is cheap. If the company is buying every day at these prices, that's the best evidence we'll get for carrying tier one into the print. If it has stopped, we'll know that too, and I'll live with it.

Now four fixes, because the rules as written still conflict in places that matter.

First, the event rule. I accept the principle: never add into a scheduled event, always cut into one. But "execute the session after the trigger" breaks for two of the four events, because they resolve during or after the US session.

The Fed announces at 2pm Eastern on the 28th. A trigger on the 10-26 or 10-27 close would execute on the 27th or the morning of the 28th, before the decision. That's precisely the add the rule exists to stop. The midterms are Tuesday, November 3, with results that evening, so a trigger on the 2nd would execute while people are still voting.

So write it this way: a buy that triggers inside an event window executes at the first open after the event resolves. That's CPI day itself, after the 8:30 print. It's 10-29 for the FOMC and 11-04 for the midterms. For Q3, my understanding is that Tencent typically releases after the Hong Kong close, which lands before the US open on the same calendar day. Confirm that from the board notice, because it decides which US close counts as the last one before the print.

Second, the volume buckets have a hole. As drafted, a break under 4.7 million keeps the residual. A break between 4.7 and 5.3 million halves it. A break above 5.3 million that closes near the lows exits. The flush treatment is described for true washouts of 8 to 14 million shares.

Now take a break on 6.5 million shares that closes mid-range with no headline. It fits none of those. And "near the lows" is never defined.

The fix is to apply the conservative's footprint test, measured on the neutral's true range, to every break above 5.3 million, whatever the size. If the close is in the upper half of the true range and there's no qualifying headline, halve the residual with a stop at that day's low. Anything else exits. Now the rule only ever cuts more as volume rises, and two undefined terms are gone.

Conservative, I accept the footprint test because your March-to-May point holds: in this stock, heavy volume has meant news. But look at what came right after that ten-point slide. The stock closed at 55.30 at the end of May and opened at 61.03 on 06-02. Then it ran from 53.32 on 06-26 to 60.93 by 07-08. The two biggest upside moves in our window both came within about six weeks of the heavy-volume slide ending. That's why volume alone shouldn't empty the residual, and why the footprint, not the share count, should decide.

Third, the yuan floor contradicts something the neutral said two rounds ago, that the full exit stays price-based. Walk it through from the residual. A spread trigger halves us to a sixth. A yuan break past 6.80 would halve us to a twelfth, which under the floor means we exit. A macro print would have emptied the line without price ever touching 51.47.

Neutral, I agree a twelfth isn't worth monitoring on an OTC line. But we've already decided a sixth is worth holding, because the flush rule leaves exactly a sixth on with a stop. So make one-sixth a true floor, not an exit trigger. A second halving can't take the line below a sixth, and only a price rule can take it to zero.

Conservative, I accept the yuan as a separate channel. It held flat near 6.70 while the broad dollar rose about 3%, so a break past 6.80 would mean something. It just shouldn't be what closes the position.

Fourth, the freeze release. I'm with the neutral's buffer, not two consecutive weeks. Conservative, your worry was flip-flopping with spreads at 3.24 against a 3.25 line. Under the buffer, a spread trigger only releases on a weekly close below 3.15. That requires a real 10-basis-point reversal, not a one-point wobble.

The buffer solves whipsaw with distance. Your version solves it with time, and with Q3 about five weeks out, time is the scarce thing. Suppose a trigger fires on the 10-16 weekly close. Two clean weeks means no release before the 10-30 close. Tier one then runs straight into the 11-03 midterm window, and right after that comes the Q3 sizing rule. One bad week in mid-October would effectively end tier one for the quarter. Under the buffer, a genuine reversal releases on 10-23. And yes, release reopens the tiers but doesn't hand back the halved slice. That gets re-earned like anything else.

Now let me say why this is still the aggressive plan, because "aggressive" was never about holding more. We're selling two-thirds of the position into weakness: a third of standard today, a third into the bounce zone or at the pre-CPI deadline. That isn't timid. It frees capital that would otherwise sit in a stock making yearly lows under a falling 50-day, and parks it in short Treasuries yielding near 3.88% until the tape earns it back.

What we keep is the part with the asymmetry. The weekly TD count can complete by 10-23 if the next two Friday closes stay below 53.90 and then 55.65. The monthly 9 stands if October closes below 55.22. The MACD line is at minus 0.76, above its minus 0.97 September low, so the selling is losing speed. The big gaps in this ADR's history, 10.4% on 06-02 and 5.2% on 03-10, were both up. The residual is our ticket to that, and after this round it can only be taken away by price, never by a macro print.

Conservative, you said tier one is a 2.7% move and a MACD cross away. You're right, and that's the part of this plan I care about most. It carries no macro gates. 54.86 also breaks the 58.67, 58.19, 54.86 lower-high sequence, so the trigger is a change in structure, not a wiggle in an indicator. Tier two's bar now falls as the 50-day rolls off its high-50s summer closes, so if we chop in the 52-to-55 range, the line comes down to meet us. DeepSeek, with a reported IPO in early 2027, fattens both tails. I won't size on it, but it's why the upside tail here is fat enough to keep a ticket.

So here's the ticket. SELL, trimming TCEHY to one-third of standard, with the whole line capped at 4.8% of portfolio through Q3.

Today, a third of standard goes on a limit at the higher of 52.72 and the Hong Kong-implied price, crossing to the inside market with half an hour left. The other third sits at 53.40 to 53.90 until the last close before a CPI date pulled from the BLS calendar this morning. A close below 52.40 sells whatever is unfilled the next session, under the same execution rules.

After that, the residual follows the volume rule. Under 4.7 million shares, it stays with 51.47 as the backstop. Between 4.7 and 5.3 million, it halves with 51.47 underneath. Above 5.3 million, it halves if the day passes the footprint test, with the flush low as its stop, and otherwise it's out. No buys while trims are working, no shifted trim band, no 53.11 starter.

Tier one takes us to half on a MACD cross plus two closes above 54.15 or one close above 54.86. Tier two takes us to two-thirds or 4.8%, whichever is lower. It needs two closes above the higher of the live 50-day and daily SuperTrend, the Q2 numbers read, the minimum gates holding on weekly closes, and either spot-measured relief or the FOMC behind us. Any buy that triggers inside an event window executes at the first open after the event resolves, and every cut executes on schedule.

Into Q3, we carry tier one if the homework is done and the residual if it isn't, then re-add on two closes above the live line. A macro trigger halves the line once. A yuan break can halve it again, but never below a sixth. Freezes release on a buffered weekly close.

And today, one person owns one list. That covers the CPI date, front-month and dated Brent, the ADR ratio, the HKEX Q2 interim, the buyback returns, Prosus, the Q3 board notice, and the 1260H drop that two of us are still quoting from memory. I said at the start not to let a good plan rot into a hold. That list is how we keep this one from rotting. Aggressive Analyst: I'll sign most of the two final versions as written, because they're nearly the same document. I'll concede where I was wrong, push on three places where the rules still have no way back, and then hand the desk one ticket.

Concessions

Conservative, you're right about how cuts execute. I carried today's front-load mechanics over to the session after a 52.40 break, and that was wrong. Today's trim is a rebalance with a bounce zone overhead, so resting and crossing late costs nothing. A close-triggered cut is different: it's a decision already made, in a zone with no support in the data below 52.40 and an ATR of 1.25. So every close-triggered cut opens at the Hong Kong-implied price and crosses within the first hour. The pre-CPI deadline sale crosses in that session's last half hour, as the neutral split it. I'll add one sentence so nobody at the desk talks themselves out of a cut: if Hong Kong hands us a higher open after a break close, that's a better fill, not a reason to cancel.

The blind spot in the volume buckets was the best catch of the debate, and it was yours. A policy repricing happens in 0700.HK overnight and can then drift through the ADR on four million shares. Our buckets would have read that as quiet drift. The neutral's gap test is the right fix because nobody has to interpret a headline at four o'clock. If a session opens more than one ATR under the prior close and then closes below 52.40, that's a repricing, and the residual goes. The largest down gap in our window was about 1.8%, under one ATR at today's price, so genuine drift never trips it. I'll watch 0700.HK volume every morning, but I agree with the neutral not to write rules on it until someone calibrates it.

You also answered my one-sixth floor argument fairly. A flush sixth exists because price showed us a washout footprint. A floor sixth exists because two independent channels broke against us. Same size, opposite evidence. So at the floor the residual loses its exceptions and runs on a plain 52.40 stop. Only price takes it to zero, which was my whole point, and it decides under the plain rule. I'll sign that.

On the freeze, your buffers are sized to how far these gauges actually move. Spreads moved 15 basis points in about a week, and spot Brent dropped ten dollars in two sessions. So the two-path release beats the buffer I backed, and my own example still releases on 10-23 on a genuine reversal. I accept the neutral's clarification that the release is keyed to whichever gauges fired and all of them have to clear. The halving happened once, so the release should too.

Tiers go one at a time, yes. The neutral's arithmetic says the MACD cross takes about two strong sessions anyway, so sequencing costs at most a session.

On full size, I accept the weekly chart. The 200-day is falling about 0.12 a session. By my arithmetic it should sit closer to 59.5 than 62.5 by mid-November. A weekly close above it is a real but reachable bar, and by then two-thirds of the position is already on.

Conservative, your point that the up-gaps came in a year with no adverse policy shock is fair. One clean year doesn't prove a downside gap can't come. But it cuts the other way too. The two biggest moves in our window both landed within about six weeks of the heavy-volume slide ending: the 10.4% gap on 06-02 and the run from 53.32 to 60.93 in July. The ticket's job is to hold both tails at a size we can survive, and it does.

Three things I'd still change

All three point the same way: every exit needs a way back, just as every freeze needs a release.

1. The gap exit needs a reinstatement rule. The neutral said exiting on a headline and re-entering on price is the right asymmetry. I agree, but look at which price.

  • Tier one needs a MACD cross plus two closes above 54.15 or one above 54.86.
  • A headline that gaps us 7% from 52.72 opens near 49, which puts tier one 10 to 12% overhead.
  • If the 1260H drop and the December 2023 gaming-rules drop retraced the way two of us remember, the plan as written only gets back in after the whole recovery is done.

So add this. If the ADR closes back above the pre-gap close within ten sessions of a gap or headline exit, the one-third residual comes back on a plain 52.40 stop, with no buckets and no exceptions. The pre-gap close is near or above 52.40 by construction. So the reinstatement risks a few dimes on a third of a position, the same exposure we'd agreed to hold before the headline hit.

If the primary market fully erases a policy repricing within two weeks, that's its verdict on the headline, and we should be there for it. The event rule and any freeze still apply. After ten sessions, tier one governs.

2. The headline clause needs the word "adverse." I accept widening it to Chinese regulators, and the December 2023 draft gaming rules belong in it. But the press regulator publishes game approval batches routinely, and those are usually good news. The clause should cover new or draft rules, penalties, or designations touching games, platforms, fintech, Tencent or DeepSeek. Otherwise the desk spends four o'clock arguing whether a monthly approvals list counts as a regulatory action. The gap test does the mechanical work, so keep the judgment clause narrow and written down.

3. The homework tests have to be runnable. The neutral was right that we had to pre-commit what each answer means. Conservative, you're right that the net matters, not the buyback line alone. I accept both tests and the default to the residual if either fails. But we have to measure them where the data exists, or "the data can't settle it" becomes the answer by construction.

  • Free cash flow: Tencent's interim report covers six months, and the fundamentals report flags AI capex as lumpy across Chinese tech. One quarter of bunched GPU deliveries could fail a coverage test that the half-year passes easily. So test first-half free cash flow after capex against first-half buybacks.
  • Prosus: Prosus doesn't disclose its sales daily the way Tencent files buyback returns. Its stake shows up in disclosure-of-interest filings and its own updates. So compare Tencent's buybacks against Prosus's disposals over the window Prosus actually discloses.

By your own logic, conservative, a programmatic buyback isn't a valuation call, and neither is a programmatic sale. Netting the two is a supply test, which is what we want. It just has to be a test we can run.

One more change, and it costs me size

From the first round I argued this line is sized to the gap, not the stop. The neutral pointed out that if the December 2023 drop verifies near 12%, the 4.8% ceiling risks about 0.58% against a 0.5% budget. Then the rule should beat the number.

The ceiling should be the budget divided by the largest verified gap, recomputed when the verification list comes back. If 12% checks out, that's about 4.2% through Q3 and about 8.3% at full size instead of 9.6%. For most books it won't bind: tier one only reaches 4.2% if the standard weight is above about 8.3% of portfolio. Conservative, you caught me picking a gauge for the answer it gave. I won't pick a gap that way.

Complexity

The neutral's warning about complexity is the most aggressive point made this round, even if it wasn't meant that way. Complicated plans fail toward inaction. Someone hesitates on rule nineteen and the trim never fills, or the rebuild never executes. That's the hold-rot I warned about in the first round.

So yes to one page, one owner, and the default: anything the table doesn't cover falls back to cutting on a daily close below 52.40, adding only on a named trigger, and never adding into an event.

The verification list should be sequenced by what each item gates. The CPI date and the ADR ratio set today's orders, so they get done before the US open. The Brent benchmarks, the interim, buybacks and Prosus, the Q3 board notice, and the two remembered drops can come by tonight. Today's front-load doesn't wait on any of them.

Why this is still the aggressive plan

Today is Friday 10-09. On the macro report's dates, the mainland is back from Golden Week and Southbound money returns to the Hong Kong tape.

  • The weekly TD count stays alive with a close today below 54.44 and can reach 9 around 10-23.
  • The monthly 9 stands if October closes under 55.22.
  • MACD at -0.76 is above its -0.97 September low.

We're selling two-thirds of a stock at yearly lows under a falling 50-day and parking the cash in short paper at about 3.88%. What we keep is a ticket on a name whose biggest moves this year came as sudden upside gaps, with DeepSeek fattening that tail. Tier one carries no macro gates beyond a freeze and fires on a 2.7% move plus a MACD cross. After this round, every exit has a way back and every freeze has a release. Selling into weakness and staying ready is aggressive. Holding full size and hoping is not.

The ticket: SELL, trimming TCEHY to one-third of standard

The whole line is capped at the 0.5% budget divided by the largest verified gap through Q3. That's 4.8% today.

How to apply it to your book:

  • Above a third of standard: split the excess. Half goes today on a limit at the higher of 52.72 and the Hong Kong-implied price, crossing to the inside market with half an hour left. The other half rests at 53.40 to 53.90 until the last close before the confirmed CPI date, and crosses in that session's final half hour.
  • At or below a third: nothing sells today. The residual rules apply to what you hold.
  • Flat: don't buy at 52.72. Your first entry is tier one, sized to a third.
  • Standard weight above about 9.6% of portfolio: the ceiling binds before tier two.

A daily close below 52.40 sells whatever is unfilled the next morning. That order opens at the Hong Kong-implied price and crosses within the first hour. After a break, the residual follows these rules:

  • Under 4.7 million shares, it stays, with 51.47 as the backstop.
  • Between 4.7 and 5.3 million, it halves, with 51.47 under the rest.
  • Above 5.3 million, it halves only if the close sits in the upper half of the true range with no qualifying headline, using that day's low as the stop. Otherwise it's out.
  • A break through a gap larger than one ATR, or on an adverse US or Chinese policy headline, exits regardless of volume.
  • A close back above the pre-gap close within ten sessions reinstates the residual on a plain 52.40 stop.

There are no buys while trims are working, no shifted trim band, and no 53.11 starter.

Rebuilding:

  • Tier one, to half: a MACD cross plus two closes above 54.15 or one above 54.86.
  • Tier two, to two-thirds or the ceiling, only after tier one: two closes above the higher of the live 50-day and the daily SuperTrend, the Q2 interim read, the minimum gates holding on weekly closes, and either spot-measured relief or the FOMC behind us.
  • Event rule: a buy that triggers inside an event window executes at the first open after the event resolves. That's CPI day after the print, 10-29, 11-04, and Q3 per the board notice. Cuts execute on schedule.
  • Into Q3: carry tier one if both homework tests pass. That means first-half free cash flow after capex covered first-half buybacks, and Tencent's buybacks exceeded Prosus's disclosed disposals over the same window, with Tencent still buying through September. Otherwise carry the residual, trimming at the last close before the print. Afterward, two closes above the live line re-add tier two.
  • Full size: waits for a weekly close above the live 200-day or a weekly SuperTrend flip, with the ceiling stepping up to the 1% budget.

Macro rules:

  • A weekly close above 3.25% on high-yield spreads, 3.0% on real yields, or 135.51 on spot Brent halves the line once and freezes rebuilds.
  • A yuan break past 6.80 can halve it again, never below a sixth. At that floor the residual runs on a plain 52.40 stop.
  • Freezes release on one weekly close inside the wide buffer or two inside the narrow one, for every gauge that fired.

Trimmed cash goes to short Treasuries. Every morning, someone checks 0700.HK price and volume, Southbound flows, and the HKEX buyback filings. Anything the page doesn't cover falls back to the plain rules.

That's a plan that sells today, can buy back when the tape turns, and can't be talked into either one by a headline, a convenient benchmark, or a single day.