I created an algorithm for forecasting time series (mix of ML methods). Now I need some data so I can compare my results with others and assess accuracy. Unfortunately, I can't find anything like the MNIST dataset for digit recognition task (ie. table with the best results and used methods: example). There are many articles about the topic but they usually use a different data (or stock data with no hope for any reasonable results). Do you have any suggestions how find some?
The International Institute of Forecasters has some datasets (under "Resources"). One very commonly used dataset is the one from the M3 competition.
The Forecasting Principles website (affiliated with the IIF) has some datasets.
Here is the Time Series Data Library created by Rob Hyndman, which @IrishStat mentioned.
You can also look at the datasets for Rob Hyndman's online forecasting textbook.
I think the most commonly used dataset would be the one from the M3 competition. If you plan on publishing in the forecasting field, using this dataset will yield a certain recognition effect among referees. The original publication on the M3 competition is this one.
The demo version of AUTOBOX has a large number of test sets (http://www.autobox.com) that could be useful to you. Rob Hyndman has amassed a large library of time series data that could also be useful.
The federal reserve allows you to download 15,000 economic indicator data like housing starts, etc.