Kroniq Research Series · Real-Time
Regime Radar Dashboard
Cross-asset HMM K=5 · 9-feature model · Walk-forward validated 2021–2024 · Updates daily at market close
Live Regime Signal
Regime
—
Loading…
Confidence
—
Posterior probability
SPY Allocation
—
Current recommended
As Of
—
Latest aligned feature date
Model
HMM K=5
Train end: 2020-12-31
All State Posteriors
Loading…
Recent Regime History — OOS 2021–2024 (252 trading days)
WALK-FORWARD REGIME CLASSIFICATION
QUARTERLY RETRAINING · 16 RETRAIN STEPS · 2BPS TRANSACTION COSTS
LOW-VOL
BULL
NEUTRAL
MACRO
CRISIS
LOADING REGIME DATA…
Key Finding — 141-Day COVID Early Warning
TWO-STAGE EARLY WARNING SYSTEM — MARCH 2020 COVID CRASH
RETROSPECTIVE FINDING · STATIC MODEL TRAINED 2015-2020 · APPLIED TO FULL FEATURE HISTORY
141 DAYS EARLY WARNING
Oct 1 2019 → Feb 19 2020
Stage 1 — Oct 1 2019
MACRO SIGNAL
HY credit spreads widening. Equity allocation reduced to 30% SPY. SPY still making new highs.
Feb 19 2020 · Day 0
SPY ALL-TIME HIGH
Equity market sees nothing. Kroniq has been at 30% SPY for 141 days. Credit already stressed.
Stage 2 — Feb 28 2020
CRISIS STATE
VIX spikes. Cross-asset stress confirmed. Allocation moves to 0% SPY. 79% of drawdown still ahead.
Mar 23 2020 · Trough
DRAWDOWN ABSORBED
SPY -33.92% peak-to-trough. Kroniq at 0% equity throughout. Max drawdown: -8.37% vs -33.72%.
Retrospective finding. The static model (trained 2015-2020) applied to the full feature history first flagged Macro stress on October 1, 2019 — 141 calendar days before the SPY February 19, 2020 peak. The 7-feature baseline (no credit spreads) detected stress only 23 days before the peak. Credit spreads extended the early warning by 118 days. The 141-day lead time is event-specific and should not be assumed to generalise as a consistent predictive horizon — see SSRN working paper for full methodology and limitations.
Validated Performance — OOS 2021–2024
| Strategy | Sharpe | Max DD | CAGR |
|---|---|---|---|
| Kroniq Walk-Forward | 0.881 | -17.46% | 9.97% |
| Kroniq Static OOS | 1.107 | -8.37% | 7.94% |
| SPY Buy-and-Hold | 0.859 | -33.72% | ~14% |
Regime Taxonomy
STATE
CONDITIONS
ALLOC
PERSIST
LOW-VOL
VIX < 15, tight credit
100% SPY
95.5%
BULL
Highest return state
100% SPY
92.7%
NEUTRAL
Transitional, residual
70% SPY
92.5%
MACRO
Chronic stress, VIX > 18
30% SPY
81.7%
CRISIS
VIX > 30, OAS > 6.5%
0% SPY
73.0%
Live API Response
GET api.kroniq.finance/regime
200 OK
{ loading… }
Key Research Findings
| COVID early warning | 141 days |
| vs 7-feature baseline | +118 days |
| BIC optimal K | K=5 |
| K sweep range | K=3 through K=9 |
| OOS period | 1,005 days |
| Retraining steps | 16 quarterly |
| Transaction costs | 2bps per rebalance |
| SSRN working paper | Polavarapu 2026 → |