Actionable insights — When Things Diverge

1. Date every major index's all-time high and read the scatter, not the level

1. Pick a fixed set of indices that span the market's capitalisation and style range — Paulo's set is Nasdaq 100, Russell 2000, S&P 500, Dow Industrials. Use the same four every time so the readings are comparable.
2. For each, write down the date of its most recent all-time high — a calendar date, not a percentage-from-high. The date is the datum.
3. Measure the spread between the earliest and latest date. Tight clustering (days to a couple of weeks) means the whole market is still advancing as one thing. A spread measured in months means the market has been topping in pieces while the headline index was carried by whatever rotated in next.
4. Ask the question in his framing, which is deliberately not rhetorical: "How are these divergences a healthy development?" Make the bull case explain the scatter; if it cannot, the scatter is information.
5. Archive the reading. The check is worth far more on its second run than its first, because the comparison is against your own prior dated list rather than a memory.

• The scatter widening rather than resolving — each new index high that fails to be confirmed by the others within days adds to it. The bullish resolution is the opposite: the laggards printing fresh highs and the dates re-converging.
• Which index is earliest. A small-cap or Dow high leading and the Nasdaq lagging is a different market from the reverse; the ordering tells you which cohort is being abandoned.

2. Repeat the same date test one level down — on the leadership names

1. Define the leadership cohort explicitly — the names that made this bull market. His grouping: Mag7 / hyperscalers / chips.
2. Date each name's own peak, the same way, on one chart so the dispersion is visible at a glance rather than inferred from a table.
3. Apply the healthy-market standard: in a genuine advance, the leadership tops together, and late. Serial peaking spread over quarters or years means the index has been passed hand to hand rather than led.
4. Pay particular attention to dispersion inside a single sub-group. Names with the same narrative and the same customers should top on roughly the same week; when they do not, the narrative is no longer what is moving them.
5. Resist the temptation to convert a peak date into a stance on the company. It is a market-internals reading, not a fundamental one — he offers no view on any of the seven names.

• The oldest peak in the cohort getting older with no new high — a leader more than a year past its high is functionally no longer leading, whatever its index weight.
• A re-synchronisation: several cohort names printing new highs in the same window would invalidate the reading and is the cleanest bull signal from this test.

3. Do not accept "broadening out" from a single equal-weight chart

1. When you hear the breadth argument, identify the exact evidence being offered. Usually it is one chart — the S&P Equal Weight Index turning up — and nothing else.
2. Separate the two claims hiding in it: (a) participation is broadening, and (b) broadening participation is bullish. The first is a measurement; the second is an inference that needs its own support.
3. Test the measurement against the tests above. Genuine broadening should show up as re-converging index highs and leadership names making new highs together. If those two tests fail while the equal-weight chart rises, the "breadth" is rotation out of the leaders, not new demand.
4. Give the other side its due before dismissing it — the dispersion tape has genuinely been lucrative for fast active managers. The disagreement is about what it implies, not whether it exists.
5. Distrust the speed of adoption. "Too quickly embraced" is itself the tell: a bullish reading that the consensus reaches immediately and from one chart has not been stress-tested by anyone.

• Equal-weight rising because cap-weight is falling — check whether the ratio is improving on absolute gains in the average stock or on losses in the giants. Only the first is participation.
• Breadth measures that disagree with each other (advance-decline, new highs vs new lows, percent above the 200-day). Unanimity is meaningful; a single supportive series is not.

4. Size the concentration before you call a rotation harmless

1. Accept the premise if it is true — leadership has changed — then ask the arithmetic question the bullish version skips: how big is the thing being rotated out of, as a share of the index?
2. Compute the weight of the vulnerable cohort on more than one definition, because the tightest definition flatters the answer. His three cuts: AI-related (~half the S&P), Mag7 (>30%), chips (~20% at their highs earlier this year).
3. Compare that weight against the capacity of whatever is rotating in. A rotation is only benign when the destination is large enough to carry the index without the source; when the source is roughly half the index, it structurally cannot be.
4. Note the direction of travel of the weight, not just its level — a cohort's share peaking (chips "at their highs earlier this year") is itself part of the divergence picture.
5. State the conclusion as a constraint on the index, not as a call on the names: concentration is why a leadership change cannot be waved through as neutral for the S&P.

• The AI/Mag7 share falling without a drawdown in the index — that would be the orderly de-concentration the bulls are assuming, and it has not happened.
• New entrants inflating the concentration further (index-eligible mega-listings), which makes the same rotation more dangerous, not less.

5. Cross the equity read against credit — look for spread/curve reflexivity, not spread levels

1. Pick the credit cohort that overlaps your concentrated equity leadership. The overlap is what makes the two readings one trade rather than two.
2. Watch for a reclassification in how that cohort is described. His marker: firms "previously considered 'asset light'" — borrowers the market used to treat as capital-unintensive and is now repricing as heavy, debt-funded issuers.
3. Look for the reflexive loop rather than the spread level: issuance requires duration hedging in Treasuries → the curve moves → spreads press wider → the next deal costs more and needs more hedging. A loop is dangerous at any spread level; a wide spread that is not looping is just a price.
4. Overlay the supply calendar against the buyer's capacity: an "onslaught of future issuance" only matters if the marginal buyer has no dry powder. Check cash levels and equity commitment on the manager surveys.
5. Trigger the conclusion on the conjunction, not any single leg — his sentence is built as "now that [spreads looping with the curve] just as [managers fully committed, low cash]…"

• IG spreads widening on days the equity index is flat or up — the divergence that says credit is leading.
• Deals being pulled, upsized concessions, or new issues breaking issue price — the market-structure evidence he has tracked all year alongside this.
• Cash levels rebuilding on the manager surveys, which would defuse the "no one left to buy the supply" leg.

6. Convert a divergence conclusion into near-dated volatility, not a directional short

1. Notice what the checklist actually established: fragility and poor internals — not a date, a level, or a catalyst. That evidence supports a convexity position, not a directional one.
2. Ask whether the option market is charging for that fragility. Compare implied volatility against the state of the tape you just documented; when the internals are this incoherent and near-dated vol is still cheap, the mispricing is the trade.
3. Choose near-dated deliberately: it is the cheapest expression, it does not require being right about the timing over a long horizon, and it is what gets crushed most by a suppressed-volatility regime — so it is where the mispricing concentrates.
4. Buy the volatility in both asset classes when the same levered names sit behind each. "Equity (and credit)" is one position expressed twice, because the leadership and the IG borrowers are the same companies.
5. Do not attach a target or a short. The checklist's honest output is "the tone for Risk overall may be changing" — own the optionality on that, and let the tape decide.

• Near-dated implied vol staying pinned while the divergence readings deteriorate further — that widens the mispricing rather than closing it, and is the case for adding.
• The carry cost: near-dated vol decays fast, so the position needs either a re-check schedule or an explicit budget for how many expiries you are willing to pay for.
• The invalidation: index highs re-converging and leadership names making new highs together would remove the premise, cheap vol or not.