More detailed replay statistics (mass killed etc.)

1

Hey!

As I am trying to climb the multiplayer ladder back to 1000 from my current 700, I started to realise that the issue might not be only with my meager APM but also with bad decision making (e.g. investing mass too early while not doing enough damage).

I am interested in detailed replay analysis, e.g. instead of just total mass graphs I would like to see:

  • mass reclaimed over time
  • mass spent on mexes, fabricators and related upgrades
  • mass spent on energy production
  • mass spent on factories and engineers
  • mass spent on units and combat upgrades
  • mass of enemy units killed by player
  • mass of enemy structures killed by player
    etc.

Maybe there is a mod for this or some replay processing library that would allow quickly putting up some code (in Python?) for generating this detailed analysis?

0

If we could see these kind of stats that would be pretty nice. You can kind of get an idea of them just from watching the replay with supreme scoreboard but this would be a bit more detailed.

1

Not possible from the replay stream.
The replay stream only contains the game orders. All of the actual gamestats are only available while the game runs through the simulation.

0

@Brutus5000 said in More detailed replay statistics (mass killed etc.):

Not possible from the replay stream.
The replay stream only contains the game orders. All of the actual gamestats are only available while the game runs through the simulation.

Well supreme score board potentially can spam those statistics to chat or post invisible 'simulation' orders so there will be a log of them inside replay.

0

@techmind_ thanks for suggestion! It appears that I can simply modify local files for existing mods to write some debug information from LUA code, and later build some charts in HTML based on those log files (I am not sure, however, if it is possible to access some external logs or not).

It would be cool to see these stats in-game but viewing them afterwards would work too.

@Brutus5000 does this look like a reasonable way to approach this task? Or maybe I am missing something more simple?

0

Sounds cool, give it a try