When are the recorded games played?
Measured across the recorded pool on patch 16.14: 41,111 recorded ranked games in 24 buckets.
Filters
Another dataset, or this one narrowed by lane. Every choice is a plain link.
dataset
The answer
This is a correlation and nothing more. It is not the best time to play. The players change by hour, not only the player: a shift in win rate by hour describes who is queueing, not what queueing at that hour does to you.
14:00 UTC holds more recorded games than any other hour
The ranking
Share of games in percent, the bar drawn from 3 to 6. Win rate in percent, the bar drawn from 49.2 to 50.8. Order: fixed bucket order. Under 30 recorded games a measure is null and the row still reports the games so far. A share is a count and carries no floor. Values pulled toward the pool are marked Pooled.
- 00:00 UTC
- 3.858%
- 01:00 UTC
- 3.491%
- 02:00 UTC
- 3.412%
- 03:00 UTC
- 3.259%
- 04:00 UTC
- 3.201%
- 05:00 UTC
- 3.405%
- 06:00 UTC
- 3.449%
- 07:00 UTC
- 3.491%
- 08:00 UTC
- 3.712%
- 09:00 UTC
- 3.768%
- 10:00 UTC
- 3.899%
- 11:00 UTC
- 4.033%
- 12:00 UTC
- 4.364%
- 13:00 UTC
- 4.624%
- 14:00 UTC
- 5.220%
- 15:00 UTC
- 5.001%
- 16:00 UTC
- 5.132%
- 17:00 UTC
- 4.991%
- 18:00 UTC
- 4.848%
- 19:00 UTC
- 4.770%
- 20:00 UTC
- 4.641%
- 21:00 UTC
- 4.590%
- 22:00 UTC
- 4.639%
- 23:00 UTC
- 4.201%
The spread
One point per row, best first. The table under the chart lists every value.
Data as a table
| Point | Value |
|---|---|
| 00:00 UTC | 3.858 % |
| 01:00 UTC | 3.491 % |
| 02:00 UTC | 3.412 % |
| 03:00 UTC | 3.259 % |
| 04:00 UTC | 3.201 % |
| 05:00 UTC | 3.405 % |
| 06:00 UTC | 3.449 % |
| 07:00 UTC | 3.491 % |
| 08:00 UTC | 3.712 % |
| 09:00 UTC | 3.768 % |
| 10:00 UTC | 3.899 % |
| 11:00 UTC | 4.033 % |
| 12:00 UTC | 4.364 % |
| 13:00 UTC | 4.624 % |
| 14:00 UTC | 5.22 % |
| 15:00 UTC | 5.001 % |
| 16:00 UTC | 5.132 % |
| 17:00 UTC | 4.991 % |
| 18:00 UTC | 4.848 % |
| 19:00 UTC | 4.77 % |
| 20:00 UTC | 4.641 % |
| 21:00 UTC | 4.59 % |
| 22:00 UTC | 4.639 % |
| 23:00 UTC | 4.201 % |
About these numbers
The payload's own caveats, word for word.
- Only games with an assigned lane are counted, so ARAM is out of every row.
- Hours are UTC. The pool spans three platforms in different time zones, so one row mixes an evening on one platform with a morning on another.
- This is a correlation and nothing more. It is not the best time to play. The players change by hour, not only the player: a shift in win rate by hour describes who is queueing, not what queueing at that hour does to you.
- With no champion filter the win rate is 50 in every hour by construction, because both teams play the same game.
Ranked games our own crawler recorded. It is our measured pool, not a sample of every game of League. The crawler seeds from the Diamond ladder on those three platforms and walks outward from the games it finds, so the pool is high MMR by construction.
These rows were measured in axiom.participants: 41,111 of the 41,126 games stored for patch 16.14 (100% of them) , recorded to .
How these numbers are worked out
Win rates are pulled toward the pool win rate with 2000 pseudo-games. Averages are pulled toward the pool average with 100 pseudo-games. Rows are ranked after that, never before, so a lucky small cell cannot top the table. More on every floor and every constant on the methodology pages.