@Pixion said in #19 and #20:
As stated above, this should be analyzed using players local time, on both sides separately as players may be in different time zones.
I do wonder wether the results are statistically significant. If you look at the y axes of these plots, the range where the data varries is very small compared to the average value.
I totally agree. I also was very confused when i read that the analysis is performed with UTC, making the study practically useless.
I think the problem is, that the lichess database only provides metadata to the game not to the players itself. So figuring out the local time for both players is not so straight forward in Data-Analysis. It might require a more sophisticated analysis, using a webscraper to get the country of each participating player and then convert UTC to the local time.
Nevertheless i think the idea is very neat and interesting.
@Pixion said in #19 and #20:
> As stated above, this should be analyzed using players local time, on both sides separately as players may be in different time zones.
>
> I do wonder wether the results are statistically significant. If you look at the y axes of these plots, the range where the data varries is very small compared to the average value.
I totally agree. I also was very confused when i read that the analysis is performed with UTC, making the study practically useless.
I think the problem is, that the lichess database only provides metadata to the game not to the players itself. So figuring out the local time for both players is not so straight forward in Data-Analysis. It might require a more sophisticated analysis, using a webscraper to get the country of each participating player and then convert UTC to the local time.
Nevertheless i think the idea is very neat and interesting.
Interesting analysis. I have a question: PGN files only include UTC timestamps, yet players can be located anywhere in the world. This means an evening game in UTC could correspond to morning in another region. Is there any way to infer a player's local time—or even their geographical location—from the PGN metadata alone? Without that context, it’s difficult to know where or when the game was played locally. I haven’t read the other responses yet, and it looks like others may have raised similar points, so feel free to disregard this if it’s already been addressed.
Even though the player's actual location isn't known—since PGN files record time only in UTC—I can clearly see from the graph that certain UTC time slots are associated with a higher frequency of blunders. So, as a player, I could strategically choose to play during those hours to increase my chances of being paired with opponents who are more likely to make mistakes.
Interesting analysis. I have a question: PGN files only include UTC timestamps, yet players can be located anywhere in the world. This means an evening game in UTC could correspond to morning in another region. Is there any way to infer a player's local time—or even their geographical location—from the PGN metadata alone? Without that context, it’s difficult to know where or when the game was played locally. I haven’t read the other responses yet, and it looks like others may have raised similar points, so feel free to disregard this if it’s already been addressed.
Even though the player's actual location isn't known—since PGN files record time only in UTC—I can clearly see from the graph that certain UTC time slots are associated with a higher frequency of blunders. So, as a player, I could strategically choose to play during those hours to increase my chances of being paired with opponents who are more likely to make mistakes.
<Comment deleted by user>
<Comment deleted by user>
