Regime Transitions & Leading Indicators

Regime Transitions & Leading Indicators

Most regime profit and loss is made or lost in the transitions, not the steady states. The state itself is the easy part — once a regime is established, every framework agrees and the playbook is clear. The hard money is in the days where one regime is dying and the next is being born, and the model does not yet know which one.

5.3.1Why Transitions Matter More Than Steady States

A Bull regime that has been Bull for ninety days does not contain much information that you do not already have. Your sizing is set, your watchlist is filtered, your hedge ratio is light, and the next ninety days will most likely look like the prior ninety. The information density of any single day inside an established regime is low, which is why so much of trading inside a steady regime is about discipline rather than alertness.

Transitions are the opposite. The day a Bull regime dies and the Neutral or Crisis regime is born is the highest-information-density day in the calendar. Your sizing is wrong, your watchlist is filtered for the dying regime, your hedge ratio is too light, and every dollar of slippage you take changing the framework over the next five trading days will dwarf the alpha you collected during the steady state. The traders who study regime transitions are the ones whose annual returns are not given back in two-week drawdowns when the wind shifts.

The asymmetry is profound: the cost of being late on a transition is measured in percent of account; the cost of being early is measured in basis points of hedging premium. A trader who pays 0.5% per quarter on out-of-the-money puts as a regime-transition hedge will look foolish for years and brilliant for the one quarter it matters. The 2022 transition off the 2020-2021 Bull regime cost unhedged systematic equity portfolios twenty to forty percent. The cost of having been hedged through the entire prior Bull was a few percent of cumulative carry.

Steady regimes test your discipline. Transitions test your framework. Most trader edge erosion happens in the transitions.

5.3.2The Lag of Detection

An HMM regime model recognizes a regime shift only after enough new evidence has accumulated to overwhelm the prior. For a well-calibrated three-state model, the typical detection lag from the actual regime shift to the posterior crossing the actionability threshold is 2 to 5 trading days. The model is not broken when it lags; the model is correctly being conservative about flipping its prior on a single day’s evidence.

The implication is that if you wait for the regime model to confirm the transition, you have already paid 2 to 5 days of slippage relative to a leading-indicator framework. In a Crisis transition, those 2 to 5 days can be 6–15% on SPY. The regime model is the right framework for ongoing sizing inside an established regime; it is the wrong framework for being early to a transition. You need a leading-indicator overlay specifically designed to fire days before the regime model’s posterior shifts.

Leading indicators have their own failure mode — they fire on transitions that do not happen. The cost of acting on every leading-indicator signal is high. The discipline is to use the leading indicators to pre-position, not to wholesale change the framework. A leading indicator that fires gets you hedged at half the size of your full Crisis playbook; the regime model confirming a few days later upgrades you to the full playbook.

5.3.3Leading Indicators by Asset Class

The single best predictor of equity regime transition is not from equities. The cleanest leading indicators come from the higher-frequency, higher-information markets adjacent to equity. Each has a characteristic lead and a characteristic false-positive rate. The discipline is to know which one is firing and what its track record is.

VIX spike beyond +2σ. The cleanest of the leading indicators. A daily close in VIX more than two standard deviations above its 90-day mean has historically preceded an equity regime transition by an average of 1 to 3 trading days. False positive rate around 35% — the spike fires on transient stress as well as regime change — but combined with the next two indicators, the joint probability of transition rises sharply.

Breadth deterioration. The percentage of S&P 500 names trading above their 50-day moving average. When this falls below 40% while the index itself is still within 3% of its high, the divergence is a classic transition warning. Lead time is longer (5 to 10 days) and the false positive rate is lower (around 25%). The signal is mid-cycle: it fires during a transition more often than at its onset, but the magnitude is large.

Credit spreads widening. The HYG-LQD spread (high-yield to investment-grade ratio) is the bond market’s vote on credit risk. When this widens by more than 1σ over a five-day window while equities are flat or up, the bond market is pricing in a deterioration that equities have not yet caught. Lead time is 5 to 15 days, and the false positive rate is the lowest of the three (around 20%) because credit markets price for capital structure rather than sentiment.

5.3.4The Jan 6 Crisis Transition

The cleanest case study in the recent tape is the early January 2026 transition into the brief Crisis regime that lasted roughly two weeks. The leading-indicator stack fired sequentially in a textbook pattern. On Day -7, the HYG/LQD spread widened by 1.4σ while SPY was within 1.5% of its all-time high. On Day -4, breadth fell from 52% to 41% above 50DMA in a single session. On Day -2, VIX spiked from 14 to 19, a +2.6σ move. The regime posterior on Day 0 was still 0.74 Bull / 0.22 Neutral / 0.04 Crisis. Two days later, on Day +2, the posterior had moved to 0.31 Bull / 0.48 Neutral / 0.21 Crisis. By Day +4 it was 0.08 Bull / 0.34 Neutral / 0.58 Crisis.

The trader who acted on the leading indicators between Day -7 and Day -2 had cut sizing by half and added downside hedges before the regime model crossed the threshold. The trader who waited for the posterior to flip to majority Crisis was sizing down on Day +4, after SPY had already given back roughly 7%. The cost of being late was visible in the equity curve within the same week.

The transition did not last. By Day +14, the posterior had recovered to 0.62 Bull and the Crisis state was discounted to single digits. The disciplined response was symmetric: leading indicators turning back — HYG/LQD compressing, breadth recovering, VIX falling — were the basis for re-adding risk before the regime model fully reverted. The trader who waited for the posterior to confirm Bull again paid the slippage on the recovery as well as the slippage on the Crisis.

The model lags by design. The leading indicators are the framework’s nervous system. Trade the system, not just its slowest sensor.

5.3.5Building an Early-Warning Composite

Individual leading indicators have too high a false-positive rate to size against on their own. The composite approach is to require concordance across at least two indicators from different asset classes before taking the pre-positioning action. The AZTMM Early-Warning composite scores each indicator 0 or 1 (fired or not) and triggers at a count of 2 or higher across at least two asset-class buckets (Vol, Equity Internals, Credit, Rates, FX).

The combinatorics matter. A single VIX spike fires 35% of the time on transient stress; combined with HYG/LQD widening, the joint false-positive rate falls to roughly 12%. With three indicators concordant across three asset classes, the false-positive rate falls below 5%. At that point, the composite is more accurate than the regime posterior on a same-day basis — because it leads.

The composite output is a number from 0 to 6 (one per indicator in the table above). The discipline is: 0–1 firing means hold framework; 2–3 firing means pre-position to half; 4+ firing means accelerate the playbook ahead of the model. Re-check daily, and unwind symmetrically when the firings clear. Track the composite alongside the regime posterior in your daily routine; the spread between them is the most actionable transition signal you have.

5.3.6Common Mistakes

• Treating the regime model as the whole framework. The leading-indicator overlay exists because the model lags by design.
• Acting on a single leading indicator. Individual false-positive rates are 20–40%; require concordance.
• Failing to unwind symmetrically. The discipline that pre-positions on transition entry must also un-position on transition exit.
• Confusing transient stress with regime change. A VIX spike with breadth and credit unchanged is stress, not transition.
• Being too cute on timing the transition top or bottom. The composite gets you in early; do not try to add scalping precision on top.
• Anchoring to the prior regime. Once the composite is at 4 firings, the prior regime is over. Acting like it is not is denial.