← Analysis page  ·  Jay Singh hub  ·  Research hub

Actionable insights — Weekly SSR: the "bad news is good news" regime read, the arb-spread decomposition, the sell-the-news test, the miscategorized-AI-beneficiary screen, the distressed-refi put-sell

The repeatable analysis behind the calls: not what he holds, but how he works — an ex-Goldman special-situations process written so it can be rerun on the next data-dependent Fed tape, the next too-wide merger-arb spread, the next blow-out that sells the news, the next SaaS name the market wrongly treats as an AI victim, the next distressed name that just fixed its balance sheet.
2026-JUL-05 · Weekly SSR research call (premium) · Jay Singh (Special Situations Report; ex-Goldman Sachs) · ▶ Transcript (PDF) · full analysis · report · deck · notes
How to read this page: each insight is a method — the trigger that put him onto an idea, the steps that turned it into a position, and the signal to watch when re-running it. The boxed line shows how it played out in this call. (Premium recording with no public video, so no per-step video deep-links — section timestamps live in the saved notes.)

1. Read the market's regime before you read the data — "bad news is good news"

The repeatable method
  1. Before reacting to any macro print, decide which regime you're in: is the market driven by growth (good data = up) or by the Fed's reaction function (bad data = up, because it caps hikes)? The regime flips, so name it explicitly each week.
  2. In a Fed-driven tape, treat a weak labor/inflation number as bullish for the relief rally — but verify the number is "bad for the right reasons." A jobless rate that falls can still be dovish if it fell because people left the workforce (participation down), not because hiring rose.
  3. Cross-check the market's own fast-moving tells (inflation-swap rates, OIS-implied hike count) to confirm the repricing is underway before positioning — and don't trust overnight futures ("a very fickle" tape) as the whole story.
Here: NFP 57k vs 113k, with participation 61.8→61.5 (750k left the labor force) and April/May revised −74k — "bad for the right reasons." Paired with a dovish Warsh Europe speech, inflation swaps collapsed to pre-war levels and OIS fell to ~1.5 hikes → a relief rally (Nasdaq futures +1.4%).
Watch for

2. Decompose a too-wide merger-arb spread into its discrete risk buckets — then check the downside floor

The repeatable method
  1. When an announced-deal spread is unusually wide, don't treat it as one number — enumerate the discrete reasons: (a) regulatory/approval risk, (b) jurisdiction/geopolitical risk, (c) the target's own financing/operational risk, (d) the commodity/price backdrop that could make the buyer re-strike.
  2. Weigh each bucket on its merits (e.g., will the acquirer's home regulator actually block a strategically-desirable deal?) and identify which one is really driving the discount (often a timeline extension, which arbs reflexively sell).
  3. Size the trade against the downside floor: if the stock has already round-tripped to its pre-announcement price, the deal-break downside is limited, so a wide spread on a well-funded cash buyer is asymmetric.
Here: AAUC — a ~34% spread on Zijin's C$44/US$32 all-cash bid vs a ~$23 stock. Buckets: outstanding Chinese MOFCOM/SAFE/NDRC approvals after a July 29 outside date; West-Africa jurisdiction risk (Mali junta); a Q1 $58M loss + Kurmuk capex; and a 20% gold drop → re-strike fear. But it's back to its pre-deal price, so "the risk-reward is quite decent."
Watch for

3. Separate the print from the position — the sell-the-news / positioning test

The repeatable method
  1. When a name posts a genuine blow-out and still sells off, don't conclude the fundamentals broke — ask whether over-extended retail/levered positioning (single-stock levered ETFs, options volume, a big pre-print run) set up a technical unwind regardless of results.
  2. Look for an independent confirming signal that demand is real: a competitor raising prices into the same "glut" fear is a hard tell that the shortage is intact, not cyclical.
  3. Anchor to the durable fundamentals (contracted volumes, capital return) and treat the positioning-driven dip as an entry, sizing for continued volatility.
Here: MU printed 85% GM, $18.3B FCF, 16 five-year SCAs ($22B deposits) and still faded to ~$992 on retail positioning — then SMCI-style panic aside, Samsung announced a 20% Q3 DRAM hike and UBS extended the shortage to 2028, confirming the pullback was "positioning, not fundamentals."
Watch for

4. Find the AI beneficiary the market has miscategorized as an AI victim

The repeatable method
  1. When a whole sector de-rates on an AI-threat narrative (here SaaS on "agents will replace software"), screen its members for the one actually using AI to gain share — with hard adoption metrics, not slideware.
  2. Demand evidence the flywheel is turning: usage stats (assistant weekly-active up multiples, apps built, % of workflows AI-generated), a new channel out-converting the old one, and an industry standard that rivals feel forced to join.
  3. Check that the compressed multiple can be grown-into: is growth durable (multi-quarter acceleration) and is the AI cost a margin drag today but a customer-acquisition investment tomorrow? Then set a target off the re-rate to a growth-appropriate multiple.
Here: SHOP — sold off with SaaS, but Sidekick weekly use +4x, ~half of Q1 flows AI-built, AI-search bringing 2x the new buyers of legacy search, and Amazon/Meta/Microsoft/Salesforce/Stripe joining its UCP council. ~35% GMV growth into a compressed multiple → ~$190 vs $119.50 (~60%).
Watch for

5. Split the index into theme vs non-theme to recover the true economic signal

The repeatable method
  1. When one theme dominates the index weight, stop reading the headline index as "the economy" — decompose it into the theme cohort and everything else.
  2. Compare the two lines over the same window: if the index is up but the non-theme majority is roughly flat, the "strong economy" read is an illusion created by a handful of names.
  3. Use that to calibrate breadth risk (how few names carry the tape) and to avoid mistaking index strength for a green light on cyclicals.
Here: 41 AI stocks = ~45% of the S&P; since Feb 28 the index is +7% but the 459 non-AI names only +1.5%. Just 10 names (32% of the index) drove 60% of the Q2 return, and 33% of stocks actually fell — "two asset classes: AI and non-AI."
Watch for

6. Trade the mechanical index event — Russell reconstitution and "style blurring"

The repeatable method
  1. Treat the semiannual FTSE Russell reconstitution as a forced-flow event: names migrating between growth/value or small/large trigger non-discretionary institutional buying and selling that front-runners chase.
  2. Watch for "style blurring" — when the formula pushes mega-caps into the Value index (Mag 7 now 17% of Russell 1000 Value), value-mandated managers are forced into tech, and the growth/value diversification you were relying on quietly disappears.
  3. Hunt for names dumped purely for index-mechanical reasons (a stock that shrank out of one index and became a rounding error in the next) as dislocations to buy, separate from fundamentals.
Here: the largest reconstitution on record — Amazon shifted almost fully to Value ("cheap" signal), MSFT/AAPL split 50/50, 43 AI names graduated out of the Russell 2000. Bloom Energy sold off simply because it shrank in the Russell 2000 and is small in the Russell 1000 — a mechanical, not fundamental, move.
Watch for

7. Use the respected 13F as confirmation of a name you already underwrote — not as the thesis

The repeatable method
  1. When a top-tier investor discloses a position in a name you've already researched, treat it as a confidence check on your own work, not a reason to buy blind — you keep the thesis, they add conviction.
  2. Discount for the disclosure lag: a 13F is weeks-to-months stale, so it confirms the setup was attractive at their cost, not necessarily today's price.
  3. Weight the source: a generational track record (a "GOAT") on a name matching your own SOTP is a stronger tell than a crowd of me-too filers.
Here: SEDruckenmiller added to the same Sea Limited that Singh laid out at ~$91 last week; "because he's the GOAT, this is good to mention," while noting the 13F is a couple months old.
Watch for

8. Shadow the specialist short — extract the screen, not the ticker

The repeatable method
  1. When a credible specialist discloses a short, resist copying the ticker — reverse-engineer the repeatable screen behind it so you can apply it yourself.
  2. Name the mispricing pattern: a stock up huge on flat fundamentals (price/fundamental divergence) is a valuation short; a lightly-regulated financial that quietly loaded up on an opaque, illiquid asset class is a hidden-leverage short.
  3. Prefer to "shadow" (a smaller, defined-risk version) rather than mirror the size, and pair with the macro catalyst that could trigger the re-rate.
Here: CATBurry's first-ever short after +86% in H1 on "sales that aren't growing that fast" (the valuation-short screen, alongside NVDA/TSLA/AMAT); and Altana shorting LNC/MET on unhedged private-credit exposure at under-regulated insurers (the hidden-leverage screen).
Watch for

9. Monitor private-credit redemption gates as a leading systemic-stress signal

The repeatable method
  1. Track the redemption-request % versus the structural quarterly cap across retail-facing private-credit vehicles (non-traded BDCs / interval funds). A request rate far above the 5% cap means investors want out and the fund is illiquid.
  2. Read repeated gating (multiple quarters at the cap, or several managers gating at once) as the leading edge of a broader liquidity squeeze, not an isolated event.
  3. Convert it to positioning: it corroborates shorts on the most private-credit-exposed managers and insurers, and flags contagion risk if a slowdown hits suppliers → credit → households.
Here: OWL — OCIC ($34B) took 18.8% redemption requests and Technology Income Corp a wild 38%, both hard-gated at 5%. "This private-credit issue is not over" — feeding the LNC/MET insurer-short thread and the BIS circular-financing warning.
Watch for

10. Buy the distressed name right after a solvency-fixing (dilutive) refi — via selling puts, not shares

The repeatable method
  1. When a shaky balance sheet completes a refinancing that removes near-term default risk but floods the market with new stock/converts, the equity overhang creates a washed-out bottom even though the company is now safer.
  2. Express the long through selling deep out-of-the-money puts into the elevated post-raise implied vol: you collect large premium and only take assignment (owning the stock cheaply) if it falls further.
  3. Frame the payoff as return-on-capital-at-risk and stagger strikes; keep it a defined trade tied to the stock reclaiming a technical level (e.g., the convert strike).
Here: SOC — Sable repaid its ~$956M Exxon loan via a $675M TLB + $345M convert + $115M equity (stock to ~$3.08), erasing default risk → +30% toward the ~$4 convert strike. Singh is long via sold 2/2.5-strike puts at high premium (~70-80% ROC), eyeing more if it clears $4.
Watch for

Methods distilled from the premium Special Situations Report weekly call (2026-07-05; transcript, report & deck PDFs in this folder; notes in transcript.md) for personal study. Not investment advice. © Special Situations Report for source material.