Why this works — and where it has been wrong

The case for watching credit is not clever modelling; it is order of payment. Lenders get paid before shareholders, so lenders reprice risk first. In every major drawdown of the past quarter-century, at least one of these signals moved before equity markets accepted the problem. What varied enormously was the timing.

Four episodes, honestly told

2000–2002 — dot-com unwind

What the signals caught: The 10Y–2Y curve was inverted through 2000 and began un-inverting sharply in December 2000. High-yield spreads, already elevated near 450–500 bps, pushed toward 700 bps through 2001 and peaked around 1,000 bps in the 2002 telecom default wave.

The timing lag: Credit warned months ahead of the worst equity losses, but the drawdown ground on for two more years. Anyone treating the first signal as a bottom was early by a very long way.

2007–2009 — global financial crisis

What the signals caught: Spreads began widening in mid-2007 while equity indices were still near highs. Bank shares underperformed sharply. Volatility term structure flipped into backwardation repeatedly through 2008, and high-yield spreads eventually blew past 1,500 bps.

The timing lag: The credit warning led the equity peak by roughly half a year; the equity bottom came a further year after that. The signal identified the regime, not the turning point.

February–March 2020 — Covid shock

What the signals caught: Two-year Treasury yields collapsed as emergency rate cuts were priced in, steepening the curve in classic bull-steepener fashion. High-yield spreads went from roughly 350 bps to over 1,000 bps in weeks, and VIX moved far above VIX3M.

The timing lag: This was fast — days and weeks, not months. Anything slower than daily monitoring would have missed most of the move, which is exactly why the volatility ratio is on the dashboard.

August 2024 — volatility shock

What the signals caught: Volatility term structure flipped violently into backwardation and small caps and regional banks were hit hard, while high-yield spreads barely moved.

The timing lag: A useful negative example: the fast signals fired, the credit signals did not, and the episode resolved within weeks. Two of six flashing is not the same as five of six.

The most important caveat: lead times are not fixed

Across these episodes the gap between the first credit warning and the equity low ranged from a few weeks to well over a year. So the score is not a countdown clock. A move to "risk off" says the environment now resembles those periods; it does not say when — or whether — prices follow. Some warnings faded without a crash, and those false alarms are part of the record too.

Wildcards these signals cannot see

If that sounds like a lot of hedging, good — that is the honest version. The value here is consistency: the same six thresholds checked every day, with every change recorded so the record is public rather than remembered selectively. Compare it with the live reading and judge for yourself.