You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
IMP uses STRATZ AI, a suite of inhouse neural networks, to calculate performance factors. These are extremely specific, comparing data specific to each player’s hero, lane, role, skill bracket, and match duration. IMP is calculated every minute, always starts at 0, and uses a rolling average to calculate each overall match IMP score.
IMP models are retrained regularly to ensure that performance scores are always up to date with in-game trends and patch updates. The IMP models also evolve over time as part of our commitment to providing the best possible performance analysis possible.
There are currently 27 metrics used to measure IMP. The neural networks determine how important each value is in contributing to win rate probability, and higher values are not necessarily better.
IMP uses STRATZ AI, a suite of inhouse neural networks, to calculate performance factors. These are extremely specific, comparing data specific to each player’s hero, lane, role, skill bracket, and match duration. IMP is calculated every minute, always starts at 0, and uses a rolling average to calculate each overall match IMP score.
IMP models are retrained regularly to ensure that performance scores are always up to date with in-game trends and patch updates. The IMP models also evolve over time as part of our commitment to providing the best possible performance analysis possible.
There are currently 27 metrics used to measure IMP. The neural networks determine how important each value is in contributing to win rate probability, and higher values are not necessarily better.
These metrics include:
Check out this blog post for a [dated] in-depth look at IMP: https://medium.com/stratz/stratz-ai-the-imp-ire-strikes-back-a7a3ca2b9eba
The text was updated successfully, but these errors were encountered: