Wayback machine confirms Fed revisionist horseshit

I've been following the St. Louis Federal Reserve Bank series called "Smoothed U.S. Recession Probabilities (RECPROUSM156N)" because I thought it was dubious. Back on March 19, I noted that it was completely ignoring the economic realities of the pandemic and blissfully reported an almost non-existent chance of recession.

If it was an infrequently updated chart that used a 3-month rolling average of data, I could understand how it would only slowly reflect economic changes. The Chicago Fed National Activity Index works like that. I mused back on March 19:

In other words, why publish a useless, untimely chart? Maybe so they can update it after the fact to show this model was somehow predictive. Fail.

So I checked the Wayback machine, a site that captures snapshots of other web sites over time to prove this series was revised after the fact. Revisionist lies are not a way to prove a model is useful. All it proves is that the model is useless and that the Fed makes up data to somehow show it is relevant. It's worse than a broken model, it's just made up.

Actual recession chances on 2/17/2020: 2.06% Revisionist recession chances: 32.55%

Actual recession chances on 3/17/2020: 2.02% Revisionist recession chances: 100%

Actual recession chances on 4/19/2020: 0.66% (Wow, things got much better from March to April) Revisionist recession chances: 100%

The next snapshot the wayback machine had was on 5/2/2020 when all the revisions had been done so I can't pinpoint the exact day someone hastily edited the data to produce an obvious economic storm on the way. If you look now, it appears the model saw this coming back in December, while the wayback machine proves it was predicting a 0.66% chance of recession as late as 4/19/2020.