Actionable insights — Software vs. Semis: the next battle in AI

01:22 1. Check what's under the index before trusting its level

1. Pair the index's distance from its high with the share of members above their 50-day moving average.
2. Compare trend signals across caps (S&P vs Russell 2000) and across credit (high yield) for divergence.
3. Run a quick bulk trend check over a sector's names to see how many are still in bullish trend.

• Breadth recovering (more names back above the 50-day) as confirmation, or the index signal flipping to join the Russell.

02:32 2. Read credit stress at the bottom tier, and map it to who borrows there

1. Track BB and CCC spreads separately; the lowest tier moves first and most.
2. Convert the CCC spread into an all-in borrowing cost (spread + Treasury yield).
3. Ask which borrowers live at that tier; here, private-equity-owned companies.

• CCC widening further while BB holds: stress concentrating in levered, PE-owned credit.

08:25 3. Use physical-vs-futures convergence to judge where commodity prices lean

1. Compare physical/prompt prices with the front futures contract as expiry approaches.
2. A large physical premium must close: either physical falls or futures grind up, usually partly both, which cushions futures.
3. Set that against where your trend signal would break; relief dips above it stay trend-bullish.

• Physical premiums weakening — the tell that the tightness is easing and the one-factor market can breathe.

17:22 4. Compare short- and long-window momentum, then overlay positioning

1. Rank names on 3-month momentum and on 12-minus-1-month momentum (skip the latest month so a fresh pullback isn't penalized).
2. Names leading the short window but not yet the long one are migrating into momentum; long-window leaders sliding in the short window are migrating out.
3. Overlay positioning: migrants coming from very light positioning have the most room to run as they graduate into longer windows.

• Software names holding strength into the 4–6 month window; a broad breakout with both groups rising together.

25:04 5. Inside a rotating theme, split haves from have-nots on supply and competitive position

1. For commodity-like products (memory), watch global capacity additions even from suppliers you don't buy from; the global price still moves.
2. For differentiated businesses, ask whether AI hurts, doesn't hurt, or strengthens the moat, and whether the valuation already assumes it.
3. Expect the momentum basket to keep the haves and shed the have-nots.

• Memory price momentum rolling over; 2027 comps against this year's memory revisions.

21:31 6. Avoid peak-on-peak setups before the second derivative turns

1. Flag groups trading at peak multiples on peak growth against an "uncompable comp."
2. Expect multiple compression and slowing revisions together once the rate of change turns, even if growth stays positive.
3. Apply it to capex-driven cycles: a pause in spending built at peak input prices is itself the second-derivative hit.

• Any hyperscaler "slow down a little" signal; power and permitting limits capping build growth.