.png)
Every polar trend on this website follows this strict validation criteria.
While we do scrape the entire web, the inital signals come to us via reddit.
Our bots are scraping (almost) every corner of the platform 24/7.
When a certain subject is talked about in comments or posts more often than normal, we consider it to be a possible polar trend.
The topic must have been mentioned at least 250% more frequently in a 5 day window, in order for it to be picked up by our scraping software.
About 90% of the topics we pick up using this creiteria, end up being total junk. That's why we have step 2
Topics that meet this criteria, are then manually verified by real humans.
It's not just a surface-level check.....
We look into:
- CDOC (see below)
- Similar existing topics that are trending
- What benefit does the topic provide / what problem does it solve?
(CDOC) = Early signs of growth that occur beneath the surface.
Underground signals are what people don't see, such as an increase in word-of-mouth discussion about a topic.
Above-the-surface signals are what people do see, such as visible hype in the form of content creators posting about the topic, increased comments, and broader online discussion.
%20(2).png)
Why we stopped publishing our predictions, then started again suddenly in early Sep.
The Polar Trends uploaded in the first few days of September were actually detected months earlier. However, at the time, our detection model was not accurate enough to confidently determine which predictions had genuine potential to become future trends. Rather than publish predictions we weren't confident in, we paused publishing while we worked on improving the system.
Over the past few years, we have continued refining our trend-detection scraper and filtering system. Our model is now significantly more accurate and has been trained on substantially more data, allowing it to filter out trends with a low likelihood of becoming genuine future trends.
This is why you may have noticed a gap in our predictions followed by a large number being uploaded within a short period. The predictions weren't newly detected, we had been holding them back while we improved the accuracy of our detection and filtering system.
Now that our system is performing much more reliably, we will be sharing Polar Trend predictions more consistently and earlier, with greater confidence in their potential to become genuine trends.
If you'd like to help us improve our trend-detection scraper and filtering system, feel free to contact us and help shape the future of PolarTrends.
All trends are detected using the same data-driven rules.
No brands, companies, creators, or agencies can pay to feature or promote trends on Polartrend. Everything you see on this site is backed by real organic engagement on Reddit and other platforms, with evidence provided for each polar trend.
All trends are surfaced using a mix of predefined, data-driven criteria based on Reddit activity as well as manual curation and subjective judgment.
Every topic is evaluated using the same rules, regardless of industry, popularity, or commercial potential.
.png)
.png)
