Skip to content

Curious to see a brief comparison eventually of how these MMR tests went

dugman dugman
in General Discussions

I’d be really curious to see a really brief comparison eventually on how the various MMR tests went and roughly what the different settings were they tested. It’d be interesting to hear kind of what sort of MMR variance and kill rate variances they saw in the various experiments and what they were tuning in them.

Just my own person academic curiosity really, MMR systems are kind of fascinating.

Tagged: BHVR

Comments

  • Mandy Mandy

    Hopefully once the tests are fully over we'll be able to share further information on what exactly was tested.

  • dugman dugman

    @Mandy said:

    Hopefully once the tests are fully over we'll be able to share further information on what exactly was tested.

    That'd be awesome, thanks! 🙂

    P.S. As an aside not much to say on a personal level about the tests so far. I only play one or two matches a night so it's kind of hard to gauge the difference for me.

  • EntitySpawn EntitySpawn

    @dugman said:

    I’d be really curious to see a really brief comparison eventually on how the various MMR tests went and roughly what the different settings were they tested. It’d be interesting to hear kind of what sort of MMR variance and kill rate variances they saw in the various experiments and what they were tuning in them.

    Just my own person academic curiosity really, MMR systems are kind of fascinating.

    @Mandy said:

    Hopefully once the tests are fully over we'll be able to share further information on what exactly was tested.

    Doubtful but its be really nice if it was split into servers and maybe two 12 hour segments (4pm-4am &4am-4pm)

  • StarLost StarLost

    @Mandy said:

    Hopefully once the tests are fully over we'll be able to share further information on what exactly was tested.

    I'm super curious.

    I will say that I have noticed a significant difference from day to day, but I definitely prefer waiting a bit longer for more even matches.

    That said, if this entire experiment turns out to be a massive placebo, I'll have to buy you all a virtual beer. Because it can be so, so subjective at times.

  • Valik Valik

    Too much cloak and daggers.

    Best we get from the devs are some ambiguous digits and a chart that has a vague connection to performance. As if they had 5,000 numbers to choose from and they just gave us one.

    They should come out with a comprehensive digest.

    Almost like WE are in the board meeting, watching the statisticians explain to the executives the discoveries made in the test with various metrics, loads of numbers, and tons of compounded truths.


    Even if the picture painted is larger than life, at least the community would understand what's going on and be able to draw their own conclusions.

    If you hide behind a streamlined figure - you run the risk of either looking ignorant or arrogant, and that's never a good feeling for a community.

Categories