Time-series forecasting is becoming a lost art

> …given all the success we've had with LLMs, it's weird that we can't use the exact same tricks with time-series data.

Forecasting works quite well when the series is forecastable. The fact that it often doesn't work — e.g. there really is no effective predictor of where the S&P 500 will be 3 months from now — might just be because the series really is effectively unpredictable (combinatorial explosion, chaotic complex systems, and all that).

One thing to note with LLMs is that they do things (e.g. generate text) that we know are possible. We know this because we consistently see humans successfully doing it, this comment being one such example (hopefully?).

We _don't_ see humans consistently predicting the stock market nor other notoriously difficult to predict time series. That's not to say that just because humans can't do it, it's not possible for an AI, but I'd consider it another clue that it may not be.

> …given all the success we've had with LLMs, it's weird that we can't use the exact same tricks with time-series data.

Forecasting works, to a certain degree, quite well when there is actually something to forecast. The fact that it often doesn't work (e.g. there really is no effective predictor of where the S&P 500 will be 3 months from now) might just be because the series really is effectively unpredictable (combinatorial explosion, chaotic complex systems, and all that).

> …given all the success we've had with LLMs, it's weird that we can't use the exact same tricks with time-series data.

Forecasting works quite well when the series is forecastable. The fact that it often doesn't work (e.g. there really is no effective predictor of where the S&P 500 will be 3 months from now) might just be because the series really is effectively unpredictable (combinatorial explosion, chaotic complex systems, and all that).

One thing to note with LLMs is that they do things (e.g. generate text) that we know are possible. We know this because we consistently see humans successfully doing it, this comment being one such example (hopefully?).

We _don't_ see humans consistently predicting the stock market nor other notoriously difficult to predict time series. That's not to say that just because humans can't do it, it's not possible for an AI, but I'd consider it another clue that it may not be.

Is advertising spam, for some random tool now with added AI.