This post, and the thread that follows, from Matt Busigin is worth checking out.I've tried to summarise the findings, probably poorly.Perhaps the most compelling finding is that Phillips Curve residuals may serve as a recession early-warning system. Spikes in residual volatility often occur 12 to 18 months before a downturn. “The ‘broken’ model is actually an early warning system.”Residual patterns identified pre-recession stress long before conventional models flash red. And unlike lagging indicators like GDP or even employment, these failures appear in real time, offering a potential lead.The Bigger Idea: Stop Ignoring the ErrorsThe meta-lesson of Busigin’s work is clear: in complex systems like the economy, the errors aren’t noise — they’re the signal. Economists have spent decades trying to build models that don’t break. But in doing so, they’ve thrown away the most valuable information — the very moments when the model does break down.“It’s like throwing away outliers in your backtest. Congrats, you just deleted all the regime changes.”By embracing the failures — not dismissing them — economists and policymakers may finally be able to build tools that work better in real time, especially during turning points.---If you've read down this far you've reached my real point. While the economics is interesting, what is more interesting to me is how this might be applied to the model's we (I) use in trading. “It’s like throwing away outliers in your backtest. Congrats, you just deleted all the regime changes.”By embracing the failures — not dismissing them — economists and policymakers traders may finally be able to build tools that work better in real time, especially during turning points.Food for thought. This article was written by Eamonn Sheridan at investinglive.com.