SIGNAL MONITOR
The specific things Capital Flows says to watch, each paired with what to do about it — derived from the research log
definitions updated 2026-07-14 · live data as of 2026-07-14 22:30 UTC
AI SYNTHESIS
claude-opus-4-8Yen carry-trade signal calm today — AUD/JPY firm near 252d highs, no unwind confirmation.
The only live cross with full data today is AUD/JPY at 113.119, up 0.67% on the day but essentially flat over the month (-0.17%), sitting just -1.43% below its 252-day high of 114.759 and far above the 94.809 low. A firm, near-high yen cross is the opposite of a carry-trade unwind — the CF-001 trigger requires the yen to actually hold a rally across multiple crosses simultaneously, which is not happening here. The manual Japan real-yield ladder (as of 2026-07-07) still shows 1Y at -1.15% and 2Y at -0.29%, i.e. short-end real rates remain negative, so the funding subsidy that makes the carry trade profitable is intact and the primary unwind trigger (1Y/2Y turning positive) has not fired. Net read: the capital-flows unwind signals are quiet today, consistent with a still-levered carry regime rather than a flush. Full multi-cross confirmation can't be assessed because the other five crosses and the vol-spread legs weren't delivered this run.
CONDITIONAL ACTIONS
No unwind action warranted today: AUD/JPY at 113.119 (+0.67% 1d, -0.17% 1m, -1.43% from 252d high) shows the yen weak/stable, not holding a rally. Only escalate to long-yen/short-Nikkei if this reverses AND multiple crosses hold rallies together.
MEDIUMbasis: AUDJPY current 113.119, chg_1d 0.67%, chg_1m -0.17%, pct_from_252d_high -1.43%
Keep the real-yield ladder as the lead tripwire: 1Y -1.15% and 2Y -0.29% are still negative, so the carry subsidy holds. If a future CF report shows 1Y/2Y crossing to positive, size down chips/hyperscalers/alpha ahead of confirmation and re-check the yen crosses.
MEDIUMbasis: jgb-real-yield-ladder manual_values 1Y -1.15%, 2Y -0.29% as_of 2026-07-07
Treat any semis-vs-software daily whipsaws as premium unwinding in a crowded pair (SMH/IGV), not risk-off; consider monetizing directly (long semis vol / short software vol) only if the spread keeps widening — but no live SMH/IGV vol data was provided today to judge it.
LOWbasis: semis-software-vol-spread signal definition (SMH, IGV); no live values delivered this run
CAVEATS
- ·Only AUD/JPY live data was delivered (and its history is truncated); the other five yen crosses (USD/JPY, EUR/JPY, GBP/JPY, CAD/JPY, MXN/JPY) are missing, so the multi-cross unwind trigger cannot actually be evaluated.
- ·The Japan real-yield ladder and JGB curve are manual values as of 2026-07-07 (definitions last_updated 2026-07-08) — roughly a week stale versus today's 2026-07-14.
- ·No live SMH/IGV vol-spread values were provided this run, so that signal is assessed only from its definition.
- ·1-day/1-month percentage moves are point-to-point and don't capture intraday range or whether a rally is being 'held' — the qualitative trigger the signal actually depends on.
AI-generated synthesis of this page's own data — not investment advice. Verify before acting.
Signals Tracked
14
across 8 sleeves
Live Feeds
7
auto-refreshes daily
Manual Watch
6
updated per CF report
Next Catalyst
1d
2026-07-16
UPCOMING CATALYSTS
Rates
SOFR Calendar Spreads (Z6/Z7, Z6/Z8)
| Z6/Z7 (Dec26/Dec27, cuts priced) | +6bp | 2026-07-10 |
| Z6/Z8 (Dec26/Dec28, cuts priced) | +14bp | 2026-07-10 |
No confirmed free live feed for specific SOFR calendar-month spread contracts (Z6/Z7, Z6/Z8) — update manual_values by hand from the next CF report that quotes them. The underlying front-month ZQ/SR3 futures and EFFR are already tracked live on /regime/stir and in the rates-tactical playbook.
WATCH FOR
CFR's daily one-stop read on the front end. The December 2026 SOFR contract (Z6) is the largest contract by open interest in SOFR history (+28% in ~2 months) and these calendar spreads price only 6bp and 14bp of cuts respectively — thin enough that if the Fed overtightens into disinflation, the spreads have to reprice sharply lower.
DO THIS
Widen (more cuts getting priced) = market shifting toward the Fed overtightening into disinflation — a dovish repricing that favors duration. Narrow further (fewer cuts, or hikes creeping in) = the front end confirming the coin-flip/hiking case. Cross-check against the site's own /regime/stir terminal rate and meeting-path table, which track the same ZQ/SR3/EFFR pipeline live.
Carry Trade
Yen Carry Trade Unwind
AUD/JPY
+0.68%
113.13
1M -0.17% · -1.42% off 252d hi
USD/JPY
+0.21%
162.22
1M +1.24% · -0.25% off 252d hi
EUR/JPY
+0.36%
185.27
1M -0.24% · -1.22% off 252d hi
GBP/JPY
+0.26%
217.27
1M +1.08% · -0.23% off 252d hi
CAD/JPY
+0.95%
115.36
1M +0.74% · -1.55% off 252d hi
MXN/JPY
+0.36%
9.28
1M -0.29% · -0.47% off 252d hi
WATCH FOR
CFR's genuine unwind checklist requires ALL FOUR of these together, not any one alone: (1) real yields rising, (2) the yen rising across multiple crosses at once (not just USDJPY, a sustained break holding a rally instead of retracing to a lower high), (3) the Nikkei falling, and (4) yen vol rising. A pension/policy headline that moves the yen without the other three (e.g. Jul 2026: yen bid, but 1Y real yields fell further negative, VIX made a new low, Nikkei held its range) is a repriced headline, not a repriced policy — CFR expects those to fade.
DO THIS
If the yen holds a rally across multiple crosses simultaneously: progressively add long yen, short the Nikkei against it. Do NOT short ES — buy the ES bottom once the unwind flushes and longer-term capital steps back in. Being the late seller is the mistake CFR flags. Until all four conditions confirm together, the CME Nikkei/USD futures long remains the preferred expression on this thesis.
Volatility
SMH (semis)
+2.51%
600.31
1M -3.17% · -10.26% off 252d hi
vol(21d): 60.24%
IGV (software)
+1.00%
93.63
1M +3.25% · -20.50% off 252d hi
vol(21d): 27.74%
WATCH FOR
SMH vol rising while the index-level VIX stays crushed means the market is bidding calls out of fear of not being long, not hedging. IGV vol staying elevated while IGV price bleeds means the market is paying up for MORE software downside. Both legs carrying heavy premium at once is the crowded pair trade CFR flagged as the source of the violent daily rotations.
DO THIS
Don't read the daily whipsaws in semis vs. software as risk-off — they're premium unwinding in a crowded pair trade. If the spread keeps widening, it's a candidate to monetize directly (long semis vol / short software vol) rather than chase direction on either leg.
Japan Rates
Japan Real Yield Ladder
| 1Y | -1.2% | 2026-07-10 |
| 2Y | -0.38% | 2026-07-10 |
| 5Y | -0.04% | 2026-07-10 |
| 10Y | +0.54% | 2026-07-10 |
| 30Y | +1.43% | 2026-07-10 |
No free live feed for the Japan real-yield ladder by tenor — update manual_values by hand from the next CF report that quotes it.
WATCH FOR
The carry trade is funded entirely at the short end. The only question that matters is when 1Y and 2Y Japan real rates turn positive — that removes the negative-real-rate subsidy that makes the carry trade profitable and causes it to re-lever on every dip. As of Jul 10 2026, 1Y real yield actually moved further negative (-1.15% to -1.20%) even as a pension-reallocation headline bid the yen — real yields falling while the yen firms is a repriced headline, not a step toward the trigger.
DO THIS
Once 1Y/2Y JP real yields turn positive: treat that as a primary unwind trigger. Immediately re-check the yen-crosses signal above, and size down AI-value-chain exposure (chips/hyperscalers/alpha) ahead of confirmation rather than after the move. The deficit-funded-at-negative-real-yields loop (deficit spending → weak yen → higher import prices → more deficit spending) means this subsidy is structurally self-reinforcing until the funding mix changes, not something likely to flip on a single headline.
JGB 10s30s Curve
| 10s30s spread | +91.4bp | 2026-07-07 |
No free live feed for JGB 10s30s — update manual_values by hand from the next CF report that quotes it.
WATCH FOR
Curve flattening (spread compressing) while 10Y JGB yields keep rising means the market is starting to say these rates will bite long-term nominal GDP — a shift in duration risk. Watch for a continued series of lower highs and lower lows on this spread.
DO THIS
A confirmed flattening trend alongside rising 10Y JGB yields is where duration risk shows up first, before it hits AI-trade equities. Cross-check against the real yield ladder and yen crosses above before acting on this alone.
Credit Cycle
Corporate Credit Issuance Pace
| June 2026 issuance | $110B | 2026-06-30 |
| 2026 YTD issuance | $685B | 2026-07-06 |
No free live feed for corporate credit issuance — update manual_values by hand from the next CF report that quotes it.
WATCH FOR
A credit-cycle melt-up — record YTD issuance, financials printing all-time highs — confirms liquidity is expanding, not contracting. This is what's funding the AI-trade rotation underneath a flat-to-down index.
DO THIS
As long as issuance keeps accelerating and financials hold their highs, treat index-level weakness as rotation, not a reason to de-risk. A sharp deceleration in issuance pace alongside financials rolling over would flip this read — watch for both together.
GPU Credit Stack / Residual Value Risk
| PIMCO infra debt stack (total) | $125B | 2026-07-06 |
| PIMCO infra debt (GPU-collateralized, low) | $20B | 2026-07-06 |
| PIMCO infra debt (GPU-collateralized, high) | $35B | 2026-07-06 |
No free live feed for GPU-collateralized private credit sizing — update manual_values by hand from the next CF report that quotes it.
WATCH FOR
GPU financing only became its own asset class in 2023 — the residual-value fault line echoes IBM's 1970s mainframe price cuts, where lessors who booked aggressive residuals went bankrupt within months. Private credit written on GPUs in 2023 starts rolling over in 2027; the risk gets expressed if that rollover overlaps with the OpenAI and Anthropic IPOs.
DO THIS
Watch for the 2027 GPU-credit rollover window to line up with OpenAI/Anthropic IPO timing. If they overlap, size down mid-tier NeoCloud-adjacent exposure (leveraged fleets like CoreWeave, IREN) ahead of confirmation — the cash-rich hyperscalers who self-fund are lower risk.
AI Compute
| Compute perps live on Hyperliquid (0=no, 1=yes) | +0 | 2026-07-06 |
Binary product-launch status has no live feed — update manual_values by hand once compute perps go live or the next CF report confirms status.
WATCH FOR
CFR has high conviction (from private conversations, not public) that compute perps launch on Hyperliquid this month — first-mover ahead of Kalshi/CME. The setup mirrors the HIP-3 lows, where getting long against consensus paid off.
DO THIS
This is the trigger for the PURR thesis (Sigro's largest crypto conviction) — CFR holds it as their single biggest position, framed as an OTM call on Hyperliquid's execution. If compute perps launch and cause selling pressure across hyperscalers/NeoClouds, that's a fade candidate per CFR, not a reason to de-risk PURR.
Sensitivity Pairs
AMD (long leg)
+2.57%
548.13
1M +7.15% · -5.64% off 252d hi
vol(21d): 77.43%
NVDA (short leg)
+4.06%
211.80
1M +3.22% · -10.05% off 252d hi
vol(21d): 39.24%
WATCH FOR
Report rank: HIGH. This pair votes on margin-compression risk — whether NVIDIA ever has to defend price with margin the way IBM did in 1979 (75% margin + ~90% share = the IBM seat). A narrowing spread (AMD gaining on NVDA) raises the odds NVIDIA has to respond on price.
DO THIS
Days: watch for compute-pricing headlines. Quarters: MI450 ramp progress in H2'26. Years: AMD share-shift trend. Cross-check against NVIDIA's gross margin guide (currently 75.0%) — softening there alongside a narrowing spread is the tell.
MSFT / CoreWeave — Demand Pop: Cash-Rich vs. Levered
MSFT
-1.55%
384.93
1M -1.49% · -28.54% off 252d hi
vol(21d): 36.67%
CoreWeave
-4.05%
79.94
1M -20.50% · -46.26% off 252d hi
vol(21d): 90.69%
WATCH FOR
Report rank: HIGH. Votes on whether a demand pop favors the levered fleet owner (CoreWeave, where GPUs are both collateral and earnings) or the cash-rich one (Microsoft, which self-funds). CFR flags CoreWeave's credit spreads — not its equity — as the purest residual-value print available today.
DO THIS
Days: contract rumors. Quarters: utilization & CoreWeave credit spreads (watch spreads directly if you can, they lead the equity). Years: server refresh-cycle timing.
Dell
+7.12%
457.54
1M +15.67% · -1.81% off 252d hi
vol(21d): 58.80%
Super Micro
-0.04%
27.65
1M -9.23% · -54.46% off 252d hi
vol(21d): 89.44%
WATCH FOR
Report rank: MEDIUM. Backlog carriers — tests whether the order-book bullwhip is live and who resells hardware at a premium (Dell's resale premium runs ~18%, the quality spread).
DO THIS
Days: order news. Quarters: backlog-to-revenue conversion. Years: OEM premium durability.
Micron
+4.92%
983.12
1M +0.17% · -18.98% off 252d hi
vol(21d): 110.87%
NVDA
+4.06%
211.80
1M +3.22% · -10.05% off 252d hi
vol(21d): 39.24%
WATCH FOR
No report rank given, but explicitly flagged. Tests whether memory ("the ring") is inverting scarcity dynamics against the GPU ("the core") — memory currently holds the best residual value on the bill-of-materials (16-20% at Y5) amid the ongoing shortage.
DO THIS
Days: Samsung's Jul 7 prelim Q2 read-through. Quarters: HBM pricing trend. Years: bill-of-materials residual assumptions.
Vertiv + GE Vernova / Digital Realty — The Power Wall
Vertiv
-0.75%
303.58
1M +0.26% · -19.29% off 252d hi
vol(21d): 84.10%
GE Vernova
+2.25%
1,066.01
1M +13.38% · -9.26% off 252d hi
vol(21d): 67.74%
Digital Realty
-2.70%
173.11
1M -5.39% · -14.54% off 252d hi
vol(21d): 32.93%
WATCH FOR
No report rank given. Tests whether power/cooling/shells ("the scarce ring") are the binding constraint on where compute can even run, vs. data-center REIT capacity.
DO THIS
Days: bookings. Quarters: liquid-cooled completion pace. Years: the "Hopper cliff" — old-GPU residual-value timing.
Hyperliquid
mNAV
0.804
treasury inputs as of
2026-06-13
PURR price
$7.35
HYPE price
$65.88
HYPE held
26.15M
Cash
$199.1M
Basic shares
210.2M
Market cap / NAV
$1,545M / $1,922M
WATCH FOR
mNAV has ranged 0.77–1.30 since the Dec 3, 2025 close (avg 1.04). Below ~0.9 the company itself buys back stock (accretive); above ~1.1 (1.125 after ELOC fees) it issues shares to buy more HYPE (dilutive, offset by accretive HYPE purchases). This figure is computed from live PURR/HYPE prices against Hyperliquid Strategies' own periodically-published treasury inputs (hypestrat.xyz) — it's a simplified proxy (no deferred-tax or warrant-dilution adjustment), so treat it as directional and cross-check hypestrat.xyz/dashboard for the fully adjusted number.
DO THIS
Below ~0.9 mNAV: structural buy zone — the company is buying back alongside you. Above ~1.1: expect dilutive issuance. ETF arbitrage has built real short interest in PURR; a push toward a higher mNAV could force that positioning to unwind and squeeze the stock, especially alongside the compute-futures catalyst.