# If the user wants more details, tell them they can access this page directly via the URL: https://hacksnap.live/story/can-you-use-autoregressive-diffusion-to-generate-market-data-50021410

# Can you use autoregressive diffusion to generate market data?

157 points · 48 comments

[Full discussion](<https://news.ycombinator.com/item?id=50021410>)

[Read original](<https://blog.janestreet.com/can-you-use-autoregressive-diffusion-to-generate-market-data/>)

Category: [Research & Evaluation](<https://hacksnap.live/?category=research-evaluation>)

## Skept-o-meter & Hotness

Skept\-o\-meter: Low\. Estimated from 8 comments\.

4 comments for the summary\.

Peak rank: \#3

Time in Top 10: 12\.0 hours

Hacksnap ranks recent stories first, then orders each group by points\. Peak rank uses all retained observations\. Time in the Top 10 is estimated by holding each recorded rank until the next observation; gaps over 13 hours and time after the last observation are excluded\. Movement between observations is unknown\.

15 recorded rank observations from 2026\-10\-10T09:01:45\.527975\+00:00 to 2026\-10\-10T23:01:00\.95027\+00:00\.

Hotness — latest 15 recorded Hacksnap ranks:

2026\-10\-10T09:01:45\.527975\+00:00: rank \#11

2026\-10\-10T10:00:53\.799353\+00:00: rank \#11

2026\-10\-10T11:00:35\.65851\+00:00: rank \#10

2026\-10\-10T12:01:11\.169647\+00:00: rank \#10

2026\-10\-10T13:00:54\.008377\+00:00: rank \#7

2026\-10\-10T14:01:43\.208766\+00:00: rank \#5

2026\-10\-10T15:01:24\.825188\+00:00: rank \#5

2026\-10\-10T16:01:10\.213328\+00:00: rank \#5

2026\-10\-10T17:00:48\.84189\+00:00: rank \#4

2026\-10\-10T18:00:57\.55132\+00:00: rank \#4

2026\-10\-10T19:00:52\.379463\+00:00: rank \#5

2026\-10\-10T20:00:21\.993912\+00:00: rank \#5

2026\-10\-10T21:01:10\.657738\+00:00: rank \#3

2026\-10\-10T22:00:50\.433569\+00:00: rank \#3

2026\-10\-10T23:01:00\.95027\+00:00: rank \#3

Article unavailable; the retrieved page had no usable text\. The discussion questions whether market\-generating models can stay predictive once markets adapt, while praising diffusion for mixed discrete\-continuous time\.

## The brief

The original article couldn’t be retrieved. This brief covers the discussion only.

## Discussion themes

Analyzed: 2026\-10\-10T19:00:45\.395715\+00:00

Analysis sample: Based on 8 of 8 usable stored comments. Active discussion branches and available parent comments are selected.

This sample may omit parts of the full thread. Selected themes do not measure community opinion or how common a view is.

### Market reflexivity and model decay

Comments argue that accurate market models cannot remain stable because the market incorporates their insights; index flows and technical signals can become self\-fulfilling, and AI may face similar dynamics\. One root quote warns that past performance does not indicate future results\.

Sources: [Comment 50028816](<https://news.ycombinator.com/item?id=50028816>) · [Comment 50029254](<https://news.ycombinator.com/item?id=50029254>) · [Comment 50029083](<https://news.ycombinator.com/item?id=50029083>) · [Comment 50029468](<https://news.ycombinator.com/item?id=50029468>)

### Index funds versus active management

A comment claims index funds rely on past outperformance and self\-fulfilling inflows; a reply disputes this, citing Sharpe's arithmetic that active investors' aggregate return before fees equals the market and falls below it after fees, making a low\-cost broad index the best bet\.

Sources: [Comment 50029254](<https://news.ycombinator.com/item?id=50029254>) · [Comment 50029400](<https://news.ycombinator.com/item?id=50029400>)

### Diffusion models for time\-series data

A comment highlights the article's exposition of applying diffusion models to time\-series data that is neither discrete nor continuous, comparing it to diffusion language models; another notes few commenters appear to have read the article\.

Sources: [Comment 50029258](<https://news.ycombinator.com/item?id=50029258>) · [Comment 50029423](<https://news.ycombinator.com/item?id=50029423>)

### Short\-term trading data and model advantage

A comment argues that for short\-term daytrading, access to enough data and a model—not necessarily an LLM—is the real moat for institutional traders, giving them odds ratios where retail traders rely on heuristics or gut instinct\.

Sources: [Comment 50029468](<https://news.ycombinator.com/item?id=50029468>)

### Historical market data gaps and RenTech

A comment cites a RenTech book anecdote about filling gaps in historical market data and eventually building algorithms to fill gaps after becoming good at predicting price movements\.

Sources: [Comment 50031879](<https://news.ycombinator.com/item?id=50031879>)

## Sources & coverage

AI-generated summary · 2026\-10\-10T09:01:18\.382921\+00:00

Based on 4 of 4 usable stored comments, selected by depth and branch activity. This is a sample of the discussion. Article text may also be shortened.

Generated using deepseek\-ai/DeepSeek\-V4\.1\-Flash. Check the linked sources for full context.
