For two years, markets ran on a simple algorithm: bad economic data meant the Fed would cut rates, and rate cuts meant stocks go up. Bad news was good news. The worse the jobs report, the more the S&P rallied. Inversion Theory described this mechanism precisely — the signal manufactures its counter-signal.
This report investigates whether that mechanism has broken. Not whether it's weakening or rotating — whether it's dead. The user's hypothesis: the contradiction between signal and response has been shrinking over time, and eventually collapses into "double inversion" — bad news just means bad news, and the framework that described the old regime has inverted on itself.
The data says this isn't a hypothesis. It already happened. The date was approximately March 6, 2026.
The SPY/TLT 10-day rolling correlation is the cleanest measurable proxy for market regime. When stocks and bonds move in opposite directions (negative correlation), the world is "normal" — bad news sends money from stocks into the safety of Treasuries. When they move together (positive correlation), something has broken — either both are rallying on liquidity (risk-on) or both are falling because there's nowhere to hide (stagflation).
Three months of data. Three distinct regimes. The oscillation that was supposed to be contracting instead expanded — until it suddenly didn't.
Stocks and bonds moved together — both falling on bad data, both failing to rally on good data. The Q4 GDP report on Jan 30 (+0.7% annualized, weakest since 2022) hit stocks for -0.30% and bonds for -0.56% simultaneously. The market was pricing stagflation: growth is dying but inflation won't let the Fed respond. Inversion Theory was already struggling here — the forced response (rate cuts) was blocked by the second constraint (inflation at 3.1% core PCE).
The correlation violently flipped negative. For three weeks, the old playbook seemed to work again — bad news sent money into Treasuries (TLT +1.33% on Jan CPI day while SPY fell -1.54%). This looked like the Inversion Theory mechanism reasserting: crisis → flight to safety → eventually, enough pain forces the Fed to act. But there's a critical detail: stocks never rallied. The bonds-as-safe-haven trade worked, but the follow-through — "therefore the Fed will cut, therefore stocks go up" — never materialized. The transmission was broken at the second link in the chain.
Three consecutive macro data releases. Three consecutive "stocks AND bonds both fall" results. The safe haven stopped working. Not because the data was ambiguous — the February jobs report showed -92,000 nonfarm payrolls, the most unambiguously bad reading in three years. TLT didn't rally. SPY didn't rally on rate cut hopes. Both fell. Inversion Theory inverted on itself. The mechanism that converts bad news into good outcomes has been consumed by the very inflation that was supposed to trigger it.
Every major macro data release in Q1 2026, with the SPY and TLT reaction. The regime column tells you whether we were in "bad news = good news" world (Normal Crisis) or "bad news = bad news" world (Stagflation).
| Date | Event | SPY | TLT | Regime |
|---|---|---|---|---|
| Jan 14 | Dec CPI | -0.49% | +0.58% | NORMAL CRISIS |
| Jan 15 | Dec PPI | +0.27% | -0.02% | ROTATION |
| Jan 29 | FOMC Decision | -0.20% | +0.02% | NORMAL CRISIS |
| Jan 30 | Q4 GDP (+0.7%) | -0.30% | -0.56% | STAGFLATION |
| Feb 12 | Jan CPI (hot) | -1.54% | +1.33% | NORMAL CRISIS |
| Feb 13 | Jan PPI | +0.07% | +0.55% | RISK-ON |
| Feb 27 | Q4 GDP (2nd est) | -0.48% | +0.61% | NORMAL CRISIS — last one |
| Mar 6 | Feb Jobs (-92K) | -1.31% | -0.37% | STAGFLATION |
| Mar 11 | Feb CPI | -0.13% | -1.29% | STAGFLATION |
| Mar 13 | Feb PPI | -0.57% | -0.49% | STAGFLATION |
The red line in the table is the boundary. Above it: mixed regimes, with Normal Crisis (bonds as safe haven) appearing on 4 of 7 data days. Below it: three straight Stagflation readings. No Normal Crisis response since Feb 27. The safe haven trade — the first link in the Inversion Theory chain — stopped working.
Inversion Theory describes a three-link chain: (1) bad data → (2) market expects policy response → (3) response reprices risk assets upward. The chain broke at link 2. Here's why:
Prediction markets price 34% recession probability by end of 2026. The Fed funds rate sits at 3.50-3.75%. Inflation is at 3.0% core PCE — a full percentage point above target. The Fed can't cut without abandoning its inflation mandate, and the market knows it. Rate cut expectations for March 18 FOMC: effectively zero. Even June is uncertain.
In Inversion Theory terms: the "forced response" (rate cuts) requires a precondition (inflation near target) that doesn't exist. The response isn't being played from a shrinking deck of cards — the card has been removed from the deck entirely by inflation. You can't force someone to play a card they don't hold.
The February jobs report showed -92,000 nonfarm payrolls. In 2023 or 2024, this would have been rocket fuel for rate cut hopes. Instead, SPY fell -1.31% and TLT fell -0.37%. Why? Because wage growth is still 4.5% YoY. Employers who survived the 2021-2023 labor shortage are hoarding workers — paying them more to retain them even as demand cools. This creates the worst possible combination for Inversion Theory: employment data bad enough to scare growth but wage data hot enough to prevent the Fed response that would make bad news good news.
S&P 500 futures speculators have been covering shorts aggressively: net short position improved from -477K (Feb 24) to -358K (Mar 10) — a +119K contract swing in two weeks. They're buying back shorts not because they're bullish, but because being short has become uncomfortable (VIX at 27.2, implying 1.7% daily moves). This is mechanical covering, not conviction. In Inversion Theory: forced buyers showing up out of role, not conviction. But unlike the usual script where this forced buying creates the rally that validates the framework — SPY still fell from $686 to $662 during this covering. The buying isn't working.
The user's hypothesis was specific: the contradiction between signal and response has been shrinking over time — a gradual compression that eventually reaches zero (double inversion). The data tells a more violent story.
The oscillation amplitude didn't shrink. It expanded: the 10-day correlation swung from +0.826 to -0.801 (a 1.627-point swing) in just 27 trading days. This isn't contraction. This is a system being driven to resonance — like a bridge swaying wider and wider before it snaps.
Then it snapped. The correlation didn't gently converge to zero. It whipped through three massive oscillations and then settled into a narrow band near zero (+0.09 to +0.21 from Mar 9-13). The current +0.130 reading isn't "gradually approaching the contraction point." It's the wreckage left after the oscillation broke the mechanism.
The 20-day rolling correlation tells the same story on a slower timeframe: peaked at +0.657 (Jan 26), bottomed at -0.534 (Feb 27), and has been climbing steadily back toward zero, crossing positive on Mar 13 at +0.093. The 20-day window smooths out noise and reveals the structural shift: the positive correlation regime (stocks and bonds falling together) is reasserting, but gently. Not the violent +0.83 of January. A muted +0.09. The market isn't oscillating anymore. It's settling into a new equilibrium where nothing works as a hedge.
How would we know if the double inversion is complete vs. just a temporary regime? Four conditions:
If the dot plot shows fewer cuts than expected AND SPY falls AND TLT falls → double inversion confirmed. If SPY rallies on "at least they didn't hike" logic → Inversion Theory lives. If TLT rallies while SPY falls → we're back to Normal Crisis. This is the decisive test.
Currently 34%. If recession odds rise AND stocks keep falling → the market has stopped believing the Fed will rescue it. Double inversion. If recession odds rise AND stocks rally → the old "bad enough to force a cut" logic is back.
If correlation remains above zero for 20+ trading days, the old regime is dead. The 2023-2024 norm was correlation around -0.3 to -0.5. Sustained positive correlation means the market has structurally repriced: bonds don't protect you anymore.
VIX at 27.19 is historically where dip-buyers show up. If SPY doesn't rally 2%+ within 5 trading days of VIX >25, the reflexive loop has broken.
If the double inversion is real — if "bad news = good news" is dead — what framework describes the new regime?
Every policy tool has been used. Rate cuts are blocked by inflation. Fiscal stimulus is blocked by $36T debt and political gridlock. QE is blocked by a balance sheet already at $6.7T. Tariff revenue is blocked by retaliatory trade war. The system has no remaining forced responses to play. In game theory terms, the deck is empty. What happens when there are no more cards? The game just... plays out. Gravity takes over. Stocks fall to where earnings justify them. Bonds fall to where inflation risk premiums justify them. Nothing forces a reversal because nobody has anything left to force with.
The economy isn't one thing. Top 40% of consumers are still spending (Costco +14.0% 6mo). Bottom 40% are breaking (PTON -43.3%, Dollar General -10.4%). Inversion Theory assumes a single economy that can be pushed to an extreme and then inverts. A K-shaped economy doesn't invert — the top half and bottom half move in opposite directions indefinitely. There's no single extreme to trigger a reversal because the average never reaches an extreme. The mean stays moderate while the variance explodes.
Maybe the entire signal-response framework is wrong. Maybe markets aren't driven by policy responses at all right now. Maybe they're driven by something simpler: mean reversion on a massive speculative cycle. The AI trade pushed SPY from $400 to $693. That's a 73% rally in two years, concentrated in 7 stocks. What if the current decline isn't about Fed responses, tariffs, or economic data at all? What if it's just the market de-rating from 22x forward earnings to 18x forward earnings because 22x was always absurd for a 3% growth economy?
Under this thesis, Inversion Theory was never wrong — it just wasn't the dominant force. The dominant force was a speculative bubble inflating and now deflating, and all the policy/data/response analysis was noise on top of that signal.
The double inversion is real but it's not the slow contraction the hypothesis predicted. It was a violent oscillation that expanded until it broke the mechanism. The SPY/TLT correlation swung from +0.83 to -0.80 and back toward zero in six weeks — a system driven to resonance, not gently converging.
The approximate date of death for "bad news = good news": March 6, 2026 — the day a -92,000 jobs print failed to rally either stocks or bonds. Every macro data release since has produced the same result: both assets fall.
What replaces Inversion Theory? Possibly nothing. The period of reliable policy-response trading may simply be over, replaced by a market that prices fundamentals without the distortion of expected intervention. After fifteen years of the Fed Put, that would itself be the most extreme inversion of all — the death of the expectation of rescue being the ultimate enantiodromic turn.
March 18 FOMC is the test. If Powell's dot plot produces "both fall" for the fourth consecutive macro event, the double inversion is confirmed. If it produces "SPY rallies on dovish surprise," Inversion Theory was only sleeping.