Hacker News Reader

Comment by hhmc | Hacker News Reader

An inverted yield curve has accurately predicted 9 out of the last 5 recessions ;)

You can even see it on the chart in the linked article, in 2005/06 — that time, it was different.

Fed Fund Rate vs 30YT since 1970(only cherry picking this year because the data on fred is poor before this period, and end of gold standard is significant) is a perfect 8/8[0].

[0]: https://fred.stlouisfed.org/graph/?g=ENDs

And how many degrees of freedom went into the choice of metrics to compare? Just because something went 8/8 doesn't mean it will predict the future crashes.

I'd argue that minimum time frame -> maximum time frame is a more fundamental metric than any other individual yield curve.

Which curve am I looking at and for what? The blue line dips a lot of times, but only sometimes does a dip coincide with a recession.

Not the OP, but I'd guess they meant "blue line greater than red line" .

Exactly. When the blue line greater than red line, shortly thereafter a recession occurs. The blue line is set directly by the Fed.

No one knows exactly why this is.

That's not that bad! Still seems potentially very useful.

9/5 is terrible accuracy.

In the investment industry, 9/5 is excellent.

How do you mean? Looks like an equivalent of a coin flip to me.

I'm not super knowledgeable in this area, but one possibility is that the "9 of the 5" only gives the number of times that it predicted that there would be a recession. There could be many more times that it accurately predicted that there would not be a recession.

If you flipped a coin every day in order to determine whether or not a recession would start, you'd be wrong 99% of the time.

How do you figure? Wouldn't you be wrong 50% of the time?

it's approximately 50/50. You're almost just as good with a coinflip.

A coinflip would be a 3650/5 or so.

A biased coinflip that gives 5 signals would be 5/0.

So you are saying that a plausible mechanism leads to a ~50% prediction accuracy?

I'm curious how this stacks up to the accuracy of other prediction methods? Off the cuff, this one doesn't seem too bad. esp given the stochastic nature of the variable being predicted.

"Accuracy" in statistical terms is likely even higher because that would include the predictions of "no upcoming recession".

This "9 out of the last 5" quote is the deadest of beaten horses on HN. It shows up at least once a week here, and it is written with such smugness and unoriginality that it makes even dad jokes funny by comparison.

Let's cut to the chase and make a bot that posts "9 out of the last 5 recessions, hyuk!" on every post remotely related to finance and be done with it. It is the furthest thing from a substantive comment; it is a recycled low-effort joke in the style of reddit.

9 out of 5 readers cringe at this type of comment.

(I’ve never heard of the expression before so I found it amusing)

I agree that it's amusing the first time one reads it. But it is posted here a lot and has become very trite; it probably competes with Hanlon's razor in terms of frequency. After just a cursory search, here are some other examples:

https://news.ycombinator.com/item?id=20372163

https://news.ycombinator.com/item?id=19995167

https://news.ycombinator.com/item?id=22352073

https://news.ycombinator.com/item?id=2127845

https://news.ycombinator.com/item?id=21023682

https://news.ycombinator.com/item?id=14417296

https://news.ycombinator.com/item?id=20641031

https://news.ycombinator.com/item?id=17266452

https://news.ycombinator.com/item?id=9476984

https://news.ycombinator.com/item?id=20656629

https://news.ycombinator.com/item?id=1628716

My interpretation is that it's not a joke, but instead a way to convey the heuristic can have false positives.

Reply on news.ycombinator.com