Hacker News Reader

Comment by btbuildem | Hacker News Reader

There were some developments using LLMs in the timeseries domain which caught my attention.

I toyed with the Chronos forecasting toolkit [1], and the results were predictably off by wild margins [2]

What really caught my eye though was the "feel" of the predicted timeseries -- this is the first time I've seen synthetic timeseries that look like the real thing. Stock charts have a certain quality to them, once you've been looking at them long enough, you can tell more often than not whether some unlabeled data is a stock price timeseries or not. It seems the chronos LLM was able to pick up on that "nature" of the price movement, and replicate it in its forecasts. Impressive!

1: https://github.com/amazon-science/chronos-forecasting

2: https://imgur.com/a/hTRQ38d

View on HN

I used to work in financial software, and when writing the charting UIs, I'd wire them up to a randomwalk to generate fake time series data. It was a relatively common occurrence for a VP or the company CEO to walk by, look at my screen, and say "What stock is that? Looks interesting."

Unpopular opinion backed up by experience: a randomwalk is the most effective model for generating timeseries that have the "feel" of real stock charts.

Or it looked interesting because it did not look normal.

It looked interesting because it was going up. The random-walks that trended sideways or downwards did not look interesting.

This is true, I have tested this with multiple veterans and none could tell them apart

I have tested this with multiple veterans and none could tell them apart - but they had a high conviction on which random walks were a good buy and which were compelling shorts.

And there is your arbitrage opportunity. If you can model how analysts will react to a particular timeseries, even if it was random until that point, you have some information about the future. It'd be a good question to figure out if there is a consensus or majority about how to interpret patterns among the people making decisions or writing quant algos, that's something one could use.

That's an interesting take! You show them meaningless data, in order to extract their overall market sentiment, and use _that_ to inform your investing strategy?

This means that you don’t even need to ask the analysts to see how they react because any bias worth trading on can be predicted from the time series itself.

That’s my experience as well. A random walk looks just like market data. You could even perform technical analysis on it, finding support, resistance, trendlines, etc. It really makes you realize why technical analysis doesn’t work.

> Unpopular opinion backed up by experience: a randomwalk is the most effective model for generating timeseries that have the "feel" of real stock charts.

That's not an unpopular opinion. The BSM model is based on the assumption that stock prices are stochastic i.e. random walks. Monte Carlo simulations and binomial trees are the two common methods of deriving a solution to the BSM model.

Since volatility clustering does exist in returns, a GARCH model should produced more realistic-looking returns than a pure random walk.

Yes, but it is also possible to generate "parameterised" random walks that have some predictability and are visually indistinguishable from "pure" random walks.

Or two series that are dependent, but individually look like random walks.

You can tell a stock time series by certain characteristics:

1) There are more jumps down than up. (Maybe not in Pharma, but in general). If there's a gap up, chances are it's on earnings day.

2) Upward movements tend to be accompanied by lower volatility, and downwards by higher.

3) There's a lot of nothing-happened days, and a lot more large jumps than you'd expect in a random walk.

I've also spent a bunch of time generating random walks, and it's true that some look realistic, but they often fall into this trap that stock returns are not normally distributed.

I also wrote a number of random trading backtests, and it's frightening how few times you need to click the "recalculate" button to get a thing that looks like a money printing machine.

Perhaps it would be easy to code a pseudo trading sequence given a model of the psychological state of the agents in the trading system

It's not easy

I'd love to see some examples, if you have old screenshots laying around!

Your take conflicts with my toy hypothesis, and I wouldn't mind being proven wrong if it saves me time and effort.

I wonder if the folks who were fooled by your screens were fooled by the random data itself, or the fact that it was presented within all the familiar chrome and doodads that people associate with stock price visualization.

As always, when running time series predictions on financial datasets, one need to use daily return (including dividends, corporate actions, etc.) rather than end of day price.

Simply outputting the last value (as more or less shown in these charts) is a pretty good end of day price predictor!

Reply on news.ycombinator.com