This week’s read: this was a steady week of selling rather than one bad session. Only 2 of our 59 names offer an entry that pays at least one to one against its own stop, so the bar for new money is higher than the index suggests.
The S&P 500 lost 2.7% across the five sessions and the Nasdaq-100 5.0%, with no single day accounting for much of it. 34 of the 59 are now classed Breakdown, up from 27. What has changed is the price of starting something new; positions you already hold are governed by their own stops below.
Every Sunday we sort those 59 stocks and ETFs into four trend regimes and publish three prices for each: where the name may be bought, where to take something off, and where we step aside. They are fixed before Monday opens, because a price set in advance beats a decision made while a position is already moving.
Our record first. Every entry we priced over 18 weeks, taken at its published level and held to its stop or to Friday, compounds to -11.0% on 512 fills after dealing costs, against +0.3% for the S&P. The five we feature each week compound to -22.0% on the same rule, a gap that is mostly the volatility of a five-name sample rather than worse picks, and the methodology section below has that test. Neither is a portfolio anyone ran and neither shows an edge. What we can defend is the process: levels fixed in advance, a stop on every one, and a record you can check. If a level looks wrong to you, say so in the comments.
Market Conditions
The broad tape softened this week. The S&P 500 (SPY) closed at 735.29, Repair and down 2.7% on the week, while the Nasdaq-100 (QQQ) at 656.10 is Breakdown and down 5.0% on the week. The damage was done before the Friday, which closed higher: SPY rose 0.31% and QQQ rose 1.40% in that session, with 27 of our 59 names lower on the day and a median move of +0.3%. Every level below is set from that close. Those regimes are the daily ones; the weekly frame has SPY down to Transitional and QQQ holding Transitional. That is where price sits against its weekly averages, not a prediction about what happens next.
Beneath the surface, regime breadth deteriorated: Breakdowns rose from 27 to 34, Repairs fell from 9 to 6, and Trend-Holds fell from 13 to 9. Regime downgrades outnumbered upgrades 16 to 4, a clear risk-off tilt. The selling concentrated in ASTS -21%, LRCX -16%, NBIS -16%, CRWV -15% on the downside, while BABA +12%, AMZN +8%, BIDU +7%, BRK.B +5% held up against the tape. Fresh damage showed up in ADBE, ASML, FTI and ITA, among others, dropping to Breakdown. Trend-Hold was reclaimed in AMZN and MSFT.
Read those counts as a statement about where our own rules will let us buy, not as market breadth. 59 names are not a representative sample, and they are not 59 independent bets: 11 semiconductor names, 4 defence, 4 metals, 2 index ETFs and 2 crypto vehicles move largely together. The practical consequence is narrow: of the names we cover, 2 publish an entry that pays more than the risk it carries, and the rest are levels to watch. The breakdown is in the Bottom Line.
What Changed From Last Week
Reclaimed Trend-Hold: AMZN, MSFT. These reclaimed while the benchmarks did not, so they are isolated pockets of strength rather than the leadership of an advancing tape, and the weekly chart still has to hold what the daily just took. We rank fresh reclaims behind names that already held the regime.
Lifted out of Breakdown into Transitional: BABA. An early bounce that still has to prove itself.
First repair off the lows: BIDU. A damaged name showing the first sign of basing.
Cooled out of Trend-Hold into Transitional: LMT, RBRK, UNH, XOM. Leaders that lost their trend and now need to re-prove it.
Lost the trend, into Repair: PANW, SOFI. Downgraded a full step, treat these as damaged until reclaimed.
Broke down: ADBE, ASML, FTI, ITA, NOC, PLTR, QQQ, RTX, SMSN.L, TSM. Fresh damage, the framework moves these to the avoid list. NOC and RTX went from Trend-Hold to Breakdown in a single week, which is a three-regime fall and the sharpest kind of break the framework registers.
How Our Trade Plans Work
Each name is classified into a regime (Trend-Hold, Transitional, Repair or Breakdown) from the 1D EMA stack. The five detailed setups print the 1W regime beside the 1D one; the condensed tables carry the daily classification only, so a name can be Trend-Hold in the table and weaker on the weekly chart. Every level traces back to a specific indicator. T1 is the first profit-take, T2 the structural objective, T3 the stretch target. Regime sets how much risk a name is allowed and when: Trend-Hold earns full risk but only on a pullback to support or a confirmed breakout; Transitional gets half or nothing; Repair gets a starter only after a reclaim; Breakdown gets none until a reclaim confirms. Those words are shares of a fixed risk budget and nothing more: if a full position risks half a percent of the account, half size risks a quarter of a percent and a starter less again. They are not statements of conviction, and a smaller size does not improve a trade, it only limits what a bad one costs.
Two stops, two jobs, and the tighter one governs. The trade stop is the tactical level where we step aside and let the position go, and it is the number the Stop column carries. The regime invalidation is the deeper level at which price sits under the whole moving-average stack and the trend structure is gone. Be clear that the second is not the first thing to change: the classification is the count of moving averages price closes above, so it steps down one notch at the nearest average below price, usually a few per cent away, and each detailed setup now prints that level too. Every stop here is a daily close condition rather than a resting order; what that costs you is in the sizing note below.
Which level wins when two of them fire. In order: the tactical stop closes the position, whatever else is true; the trailing level only ever tightens that exit and never loosens it; a regime step down closes nothing but caps the size, so an open position is held and not added to; and the invalidation sits below the tactical stop, reached only if that stop was passed through on a gap. Entries are read intraday and classification at the close, so a dip can fill in the morning of a day that ends in a weaker regime. The stop still governs; the new class applies to what you do next. For most setups the invalidation was the only number we showed, and across the 436 fills from those issues it sat a median of 10% away while the typical trade never travelled more than 3.4% against us: only one trade in six ever reached it. A stop that far out is a thesis level, not risk control, and printing it alone was our mistake.
The counts below are not interchangeable. Across 21 weeks we have published 1023 entries carrying a price: 617 triggers above the close, of which 305 were reached, 49%, and 406 dip bands at or below it, of which 264 filled, 65%. A band fills more often because price only has to come back to it. Together 569 of 1023 filled, 56%. Six further rows printed “on reclaim” with no level at all, an old fallback we have since removed; they could not fill and are excluded. The 436 figure beside the invalidation is narrower again: the fills from the fifteen issues that published an invalidation rather than a governing stop. None of these is a count of independent experiments. They are 59 overlapping names across consecutive weeks, so one bad week moves all of them, and the rules were being adjusted as the history accumulated.
How long a trade lasts, and when to take something. Every level here is a one-week level: take it during the plan week, and if neither the first target nor a stop has been reached, mark it at Friday close and re-read the structure next issue. T1 is touched at some point in the week on 53% of 569 filled setups, and what you do there matters more than where you entered. Held to Friday or the stop, 41% of setups finish green and the average returns -0.8%. Closing the entire position at T1 lifts the green share to 63% but cuts the average to -2.1%, because it sells every winner at the first target and keeps every loser. Half off at T1 with the rest left to run lands between the two: 42% green, -1.4% average. Pick on that basis rather than on instinct, and be clear what none of the three does: every one of those averages is negative, so T1 management changes the shape of the record, not its sign.
When a level is a trade and when it is only a level. Every entry here is measured from its own price to the stop that governs it, and on to the first target that entry can still reach. If that pays less than one to one we print the level and say it does not qualify, because below that we would rather hold the cash. The threshold is our choice rather than a law, and clearing it is not proof of a profitable trade: it is the floor, not the case. A failing entry is never rescued by a smaller position, because size changes what a bad trade costs and not whether it is bad.
What that floor costs, and what it is not shown to buy. This is a third population again, and a narrower one: the 413 fills that priced an entry, a governing stop and a first target above that entry, which is what it takes for a ratio to exist at all. Across them the test refuses 79%, four entries in five, and the way it does that is worth knowing. The ratio rises when the stop is near, and near stops are closed through more often: of the entries paying under one to one, the ones we refuse, 7% were closed through their stop, against 18% of those between one and two and 29% of the few paying better than two. Set against outcomes the test is a wash. The entries it kept returned -0.88% and those it dropped -0.47%, so the 0.85 points of difference run against the ones we kept, on a p-value of 0.47 across the 19 weeks that offered both. That is no measurable difference in either direction, and we will not have the history to settle it for years. We keep the floor because we would rather hold cash than take an entry whose arithmetic we cannot defend, not because we have shown that it pays.
How the record is calculated. Both figures in the opening take every entry at the level we published, fill a gap through it at the open, exit at the first daily close through the governing stop or else mark at Friday close, chain the weeks together and charge 10 basis points of dealing costs once per trade. Charging nothing, 20 or 40 basis points instead, the featured series reads -20.6%, -23.5% and -26.2%, so costs move it about 1.4 points per 10. Two limits are specific rather than general. 12 of the 90 featured setups, all of them between 13 July and 28 September, printed only the regime invalidation and no tactical level, so those were graded on the looser stop. And three weeks carry no issue at all; they are absent from the record rather than counted as flat. On the gap between the two: the featured five averaged -1.20% a week against -0.53% for the whole priced list, a paired difference of 0.67 points with a p-value of 0.49, and they trailed in 9 of 18 weeks. That is a coin flip, so the compounded gap is the cost of compounding five names a week instead of the whole list, not evidence about which names we chose.
A note on what our own record does and does not show. The one place we thought we had measurable value was the avoid tag, and it does not survive being measured as a policy you could run. Counting only the avoid names that never reclaimed, they fell 2.9%, which looks decisive until you notice that you can only identify that group after the week has finished. Counting every name we tagged avoid, the figure is -0.9% against -1.2% for the rest of the book. So the tag has neither helped nor hurt, and we would rather say that than quote the flattering version.
One name you cannot size in. XRPI clears too little daily volume for a retail order to fill without moving the price, so its classification is coverage rather than a tradeable signal.
Size by risk, not by dollars. Among the names this issue says you may take, stop distances run from 1.7% to 2.6%, so the same dollar amount in two names can mean risking 1.5 times as much on one as the other. Decide what a single losing trade may cost you, say half a percent of the account, and divide it by the distance from entry to stop. A 1.7% stop carries 1.5 times the position of a 2.6% stop for identical risk. The wide-stop names are not off limits; they are simply small.
Two limits on that formula, because on its own it will concentrate you. Risking half a percent of the account against a 1.7% stop puts about 29% of the account in that one position, which is more than any single idea in this letter deserves: cap any one name at a fixed share of the account as well, and treat two names in the same sector or two index ETFs as one position for that purpose. And every stop here is a daily CLOSE condition rather than a resting order, so the distance to it is not a maximum loss. A gap can close well through it, and an intraday break that recovers by the bell does not trigger it at all. Size for a worse outcome than the arithmetic shows.
This update ranks 59 names. You are not meant to trade all of them, or even most of them. Most weeks a handful sit at a level worth acting on and the rest are there so you can check our work.
The Week Ahead
These names in the universe report during the plan week:
Monday: ASTS
Tuesday: CRWV, IFX.DE
Thursday: AMAT, NBIS
One note on these dates. AMAT, ASTS, CRWV, IFX.DE, NBIS appear in only one of our two calendars, so those dates are unconfirmed. Check the company before you size into any of them.
Market-wide releases this week, from the BLS calendar: Tuesday 8:30 a.m. ET, Consumer Price Index (October) and Friday 8:30 a.m. ET, Producer Price Index (October). These move the whole book at once rather than one name, so a single level surviving them is not the test. The first lands Tuesday, before most of these levels have had a chance to fill, and the last on the Friday.
Earnings are binary, and the framework does not trade into them: if a name on this list reaches its entry before it reports, we do not take it. There is no reduced-size version of that, because a gap through the stop costs the same multiple of the risk whatever the size. Wait for the print, then re-read the structure. A level that was valid on Friday is not wrong after a gap, it is simply no longer the level.
Just beyond the plan week: BIDU on Monday November 16, NVDA on Tuesday November 17, PANW on Wednesday November 18, TGT on Wednesday November 18, BABA on Thursday November 19, WMT on Thursday November 19, INTU on Thursday November 19. Worth knowing now if you are sizing a position you intend to hold through it.
This Week’s Playbook
Same four regimes as above, in the order they earn risk. Regime decides how much a name is allowed; the level beside it decides when. Detailed setups for the top names follow, and the watchlist carries every level.
Within each group the names are ordered by how close the entry sits to Friday close, nearest first, which is what the Trigger odds column in the tables below reports. That shows where a decision is likely to be needed, not where the reward is. Of the 617 triggers we have published, the ones we would have called likely were reached 76 percent of the time, even odds 46 percent, and unlikely 19 percent, with roughly half never touched at all. Reaching a level is not the same as profiting from it: of the 152 likely ones that filled, the average came back about one percent negative, with about four times in ten closing green. Read the top of each list as the levels most likely to demand a decision, and size them off the stop underneath.
Where that leaves your exposure. The 59 names we cover split in one way that matters: 2 publish an entry that clears our one-to-one minimum (AAPL, BRK.B), and the remaining 57 do not, so they are reference levels this week and not purchases. Of those 2, 1 passes every other rule as well; BRK.B is held back as overbought, which is a wait rather than a buy. Both are dips you would buy into weakness. That is a separate question from whether a level gets reached, which the Trigger odds column answers, and from whether the level is worth watching at all, which is what the watchlist is for. So little new money goes to work until a level prints, and that is about fresh capital for new trades, not an instruction to sell what you already hold: positions you own are governed by their own stops and the trail lines in each setup. Cash is an allocation rather than the absence of one, and it is not free: it loses purchasing power while it waits, and whether it beats the names it is waiting on is exactly what none of us knows in advance. What it buys is the ability to take a level when one prints. The S&P regains its trend on a daily close above 762.65, +3.7% from Friday (its row in the table shows 754.17, the 1D Tenkan, which is the nearest level back rather than the full reclaim). The Nasdaq-100 regains its trend on a daily close above 713.71, +8.8% from Friday (its row in the table shows 701.19, the 1D EMA20, which is the nearest level back rather than the full reclaim). Those two levels are context rather than a gate. Every entry here is permitted by its own regime and its own level, so a qualifying setup is takeable with the benchmarks where they are; what the index regime changes is how much of the book should be working at once, not whether a single trade is allowed. What would make us less defensive: either benchmark reclaiming its trend, the Breakdown bucket shrinking two weeks running, or dip bands reappearing on names that clear one to one to their first target. What would make us more defensive: more names dropping into Breakdown, the weekly classifications following the daily ones lower, or stops hit on the setups below. Until something moves, do not manufacture a trade out of a name you like at a price the plan never gave you.
Trend-Hold, full risk, but only at a level. On a pullback to support: AAPL. On COST, AMZN, TGT, MSFT, WMT, LLY, DG the first target does not clear the distance to the stop, so those are a level to watch and not a position: see the payoff line below. BRK.B is also Trend-Hold and carries the overbought instruction below rather than a level here.
Transitional, half size or nothing. The trend is not confirmed, so a position here is a probe rather than a holding. On LMT, UNH, OSCR, LULU, BABA, SHOP, RBRK, INTU, XOM, NKE the first target does not clear the distance to the stop, so those are a level to watch and not a position: see the payoff line below.
Repair, nothing to take this week. A reclaim buys a trend that has not formed yet, which is only worth paying for when the first target is further away than the stop. On every Repair name this week it is nearer, so these are reference levels and not starters: BIDU, SOFI, NVO, AMD, PANW, SPY.
Breakdown, no exposure until a reclaim confirms. Not every dip is a setup, and cash is the position: GLD, NOC, ITA, AIXA.DE, PALL, RTX, ADBE, ETHA, and 26 more carrying the same instruction in the table below.
Overbought inside their trend: BRK.B. The trend is real but the entry is not. A Trend-Hold does not earn its full size at any price, so wait for the reset.
Where the payoff justifies the stop. A wide stop is not a bad setup, it is a small one, but some names are not paying enough to be worth much size. Both figures are measured from the published entry to the governing stop and to T1, the first target, which is the only one the condensed tables carry; T2 and T3 sit further out and are in the detailed setups.
- Tight and paying: AAPL (2.6% stop for 6.4% to T1)
- Do not take, the first target does not clear the risk: NKE (11% stop for 0.6% to T1), BABA (6.8% stop for 0.6% to T1), XOM (6.3% stop for 0.8% to T1), INTU (11% stop for 1.8% to T1)
Top Actionable Setups
One name clears every rule this week, so there is one detailed setup below rather than five. The screen takes Trend-Hold names only, then drops anything overbought above RSI 73, anything whose EMA stack is inverted, anything reporting earnings inside the plan week, anything without one average daily range between its entry and its stop, and anything whose entry pays less than one to one to the first target it can reach. Survivors rank by distance to the 20-day EMA, with ADX breaking ties and a name that only reclaimed its regime this week placed behind one that already held it, and a name too closely correlated with one already chosen swapped for the next candidate. “Actionable” means a level exists and the rules permit acting on it, not that the trade is shown to pay.
