Actionable insights — Portfolio Update: DGX + Mid-Caps

1. Buy the retest of a breakout that coincides with the 200-day — and be willing to miss the breakout itself

1. On the long chart (five years is the working span here), mark the prior overhead resistance — the ceiling the stock repeatedly failed at before it finally cleared.
2. Accept that you will often miss the range expansion itself. Chasing the initial breakout is a different, worse trade; the tactic here is explicitly a second-chance one.
3. Wait for the retracement back toward that former ceiling. A level that capped the stock for years becomes the natural floor once it is above it, because the marginal seller who was trapped there is now gone.
4. Require a second, independent support at the same price before committing — the 200-day moving average is the one used here. One support is a level; two coinciding supports are a setup, because they bring different buyers (technical traders and trend followers) to the same price.
5. Buy "relatively close to" the confluence rather than exactly at it. The rule is stated as a tendency — "typically a profitable tactic" — not a precision entry, which is what makes it usable without perfect fills.
6. Pair the technical entry with a fundamental thesis you would hold anyway. The chart chooses the when; it does not choose the what.

• A decisive close back below the old resistance level — that is the falsification, and it is also the natural stop, because the whole premise was that the level had flipped from ceiling to floor. Watch too for the 200-day drifting away from the breakout price over time: the confluence that made the setup only exists for a window, and once the two levels separate you are back to a single, weaker support.

2. Separate the view on the business from the view on the position size — and give each its own rule

1. When a winner gets extended, resist framing the decision as "do I still like it?" That question conflates two independent judgements and usually resolves into an unnecessary sale.
2. Ask the business question first, and answer it against the original thesis: is it still executing as the write-up described? If yes, the thesis is not the reason to act.
3. Ask the portfolio question separately: has the position's weight grown into something you would not choose to hold today from scratch? Is your overall market exposure where you want it?
4. If the answer to the business question is yes and the portfolio question is the problem, the correct action is a partial trim, not an exit. State a specific fraction — one-third — so it is executable rather than aspirational.
5. Anchor the hold to the next scheduled catalyst rather than to a price level. That converts "hold" from an open-ended posture into a decision with a date on it, and it is testable.
6. Publish both instructions at once. Different readers have different weights and different cash needs; a single instruction would be wrong for most of them.

• The catalyst date itself (October 27) — a hold anchored to an event has to be re-decided when the event passes, not silently rolled forward. Watch also for the failure mode this rule is designed to prevent: trimming on price strength, then trimming again on the next strength, until the winner is gone and the thesis was right the whole time.

3. Before abandoning a lagging call, decompose the gap — and check whether the causes favour you going forward

1. Mark the position honestly and in full: price return, total return with dividends, and the annualised rate, against the benchmark over the same window. No selective start dates.
2. State the shortfall in the benchmark's favour without softening it. A call that has lagged has lagged; the credibility of everything that follows depends on saying so.
3. Now decompose the benchmark's win into its sources: how much came from multiple expansion (investors paying more per dollar of earnings) versus earnings growth, and how much came from construction (weighting, concentration) rather than from breadth.
4. For each source, ask the forward question: is it repeatable, and is it exhaustible? Multiple expansion from an already-high starting point is exhaustible. Earnings growth of non-operating origin is exhaustible. Concentration cuts both ways.
5. Only then decide. Keep the position if the causes of the lag are the kind that reverse; abandon it if the causes are durable advantages you simply do not have.
6. Re-underwrite the starting valuation of your own holding too, honestly, since a lagging asset is not automatically a cheap one.

• A relative-performance turn that starts with the identified cause reversing rather than before it — mid-caps outperforming while mega-cap earnings still accelerate would mean your decomposition was wrong even if the trade works. The falsifier: the S&P's earnings continuing to compound from operations after the non-operating gains are stripped out.

4. Reprice an index on normalized profit margins before comparing its multiple to anything

1. Never compare two P/E multiples until you have asked whether the two E's are of the same quality. A headline multiple computed on peak-margin earnings understates the true price being paid.
2. Look for the tell that margins are doing the work: a period where the index's price rose sharply while its forward P/E fell. That combination proves the return came entirely from earnings, which relocates the entire risk into the earnings line.
3. Interrogate the composition of those earnings. Separate operating profit from non-operating items — mark-ups on stakes in private companies, investment gains, one-time revaluations — which are real in the accounting sense and absent from cash flow.
4. Quantify the dependence where you can. A single company contributing a majority of its own profit from revaluation gains, inside an index where the top ten are ~38% of the weight, is a concentrated exposure to one accounting convention.
5. Ask whether the comparison asset had the same booster. If it did not, its duller reported earnings were cleaner, and the multiple gap between the two is narrower than it looks — possibly inverted.
6. Conclude in relative terms rather than as a market call: mean reversion in margins is a statement about which asset wins, and its timing is unknowable.

• Private-AI valuation marks flattening or reversing — that is where the flattered earnings stop, and it shows up in reported profit a quarter later. Watch too for the honest counter: if operating growth alone keeps the index's earnings compounding after the mark-ups fade, the normalized-margin case weakens rather than merely being early. And note the standing weakness of this method — it identifies an overvaluation, never a date.

5. Treat index inclusion as a mechanical transfer — and prefer the index that is doing the selling

1. Recognise that index membership is decided by market capitalisation, i.e. by price. A company is promoted after its shares have already exploded, not before.
2. Follow the flow: the receiving index's funds are forced buyers at the peak; the donating index's funds are forced sellers of the same shares at the same moment.
3. Ask which side of that transfer you want to be on structurally. The mid-cap index systematically sells its parabolic winners into the promotion and keeps the steadier compounders; the large-cap index systematically buys them at their most expensive.
4. Size the effect honestly per event — a new entrant is a small weight, so any single case is negligible — then aggregate: "multiplied by dozens, if not hundreds," it becomes a measurable long-run drag.
5. Use it as a structural preference, not a trade. It explains a decade-scale return differential; it will not tell you anything about the next quarter.
6. Apply the same mechanic in reverse when hunting individual names: an index removal is forced selling into weakness and can be an opportunity, the mirror image of the graduation problem.

• Announced S&P 500 additions of recently parabolic names — each one is a live instance, and the post-inclusion drawdown is the measurable evidence. The counter-case to respect: a promoted name that keeps compounding makes the drag argument look like cherry-picking, so the claim is about the average of many events, not about any one.

6. Put the conviction on the horizon where the argument actually operates — and say what you are cautious about in between

1. Identify the time scale of your evidence. Structural arguments — margin normalisation, index-mechanics drag, starting valuations — operate over years, not quarters. Saying so is not hedging; it is matching the claim to the data.
2. Give the near term its own, separate assessment, and make it about risk conditions rather than about the thesis: leverage, breadth, positioning, the count of live threats.
3. Name the specific near-term risk instead of gesturing at "uncertainty." Margin debt going vertical is a checkable series, and it is the channel by which a valuation problem becomes a forced-selling problem.
4. Translate the near-term caution into a sizing instruction, and the long-term conviction into a holding instruction. Those are compatible; a single blended instruction would be neither.
5. Address the audience segment explicitly — "for risk wary types" — because the correct action genuinely differs by risk tolerance, and pretending otherwise produces advice that is wrong for most readers.
6. State the long-horizon claim with a number attached (a decade), so it can be judged later rather than quietly abandoned.

• Margin debt rolling over — historically the leverage unwind, not the valuation, is what sets the timing of a drawdown. Watch also for the long-horizon claim's own falsifier: mid-caps' 15-year lag is admitted in the same post ("despite that it has badly lagged over the last 15 years"), so a further multi-year stretch of underperformance is inside the range of what has already happened and would not, by itself, prove the structural case wrong — which is exactly why the position is sized cautiously rather than pressed.