A guide​ to the 2019 Update, so far the most data driven one till now

This site uses cookies. By continuing to browse this site, you are agreeing to our Cookie Policy.

The latest issue of the 9th Scroll is here! You can read all about it in the news.

Our beta phase is finally over. Download The Ninth Age: Fantasy Battles, 2nd Edition now!

  • First I want to appologize, for being late with this article. I originally wanted it to be ready when the first part of the update hits. Well, that did not work as intended. The good thing is, that now I should know what interests you most about the 2019 update.

    So what info will you get in this blog article?
    • The goals of this update and how the update was prepared and done
    • The future of the update and what you can do to improve it

    • The goals of this update and how the update was prepared and done

      The update had several different goals, I'm going to walk you through each goal seperateyly. Basically the update targeted 3 things treated with equal importance:

    • Improving external balance
    • Improving internal Balance
    • Bringing in line to strong designs for which pricing could not be used as a solution as it would cause to much collatteral damage

    • Improving external balance

      For that of course one needs to know where an army stands. As written in this Announcement we did collect data from multiple sources. Primary datasources were surveys (to both open community and a group of top tournamentplayers) and tournament data (single and team seperated).Than the ranking those datasets gave tot he armies were compared. If they didplace an army in the same area (= only did differentate about 1 -2 places), Thescources were calculated together giving the single data the highest value.After that with different weightenings oft he datascources and other methodesthe rest oft he armies were placed.That of course means, that none oft he scources will rank the armies total identical with the final tier list. If for example single data would put Army Asomewhere in the lower middle oft he pack, but External Experts, Community andTeam tournament data puts it on the last spot, than 3 scources vs 1 scourcemake it likely that the one scource might show the army potentially as a bitstronger than it really is. (The width oft he corridors the armyperformancedata puts the armies, makes that also a possibility). But more on the topic of theexternal performace tourney data at a later time (=around when the final updatehits or around december, which ever is nearer to me finishing that report.Basically to improve the external balance things should get cheaper and others more expensive. Tie 1 should get more rises in total on current builds (even after taking internal balance into account), than tier 5 (where rises on things would beinternal motivated and be much less, than if the same army would be Tier 2-4.)

    • Improving internal Balance

      You hopefully all have seen and contributed to your armies survey and the armylist analyse threads in the armyboards. A link to those threads can be found here. In those threads armylists from both, single and team tournaments of various size were entered in TA’s files and then automatically analyzed. This analysis is for you to see in each thread in one of the first three posts.

      Let me give you an example of how the internal balance adjustion works so far. Through data analysis, EE Reports and the Community surveys every entry gets a grade between 1 (hardcore seriously too point efficient) and 6 (hardcore serious too point inefficient). Then those grades are combined into a final grade. For that the different info scources have been weighted accordingly. Afterwards according to the final grade a factor is used to either increase or decrease the price.

      Let us get more concrete on the armylist data site, as about that I can explain more than about the other factors.

      We decided to use the taken "once or more often per army" dataset, as the basis for our analyses. Basically we belive it models the unit usage more fairly for units which can be taken in different quantities and/or for different utilities. Of course there are things where the "total taken" dataset might produce more exact results. (Yes, you guess it right, next year we will very likely have found a way to combine the usage of both datasets in a way to get the best out of both).

      From that we got how often something has been used at least once and how often from that category (Core, Characters, Noncore + Noncharacters, faction specific items, unit/character- options) would have been expected to be taken on average. There are several possible models to use for that each with its own pro's and con's and ultimately influences on the result. For example one can calculate the average expected shield usage for a character with the option for shield and great weapon as shield or no shield, which results in dividing the number of how often that character has been taken at least once by 2 or one could say shield, great weapon or no shield and divide by 3. This is an area where we will review our macros if we still believe to have taken the right decision.

      Now, how often things which were taken "at least once" were compared with that average. Things being used 190%+ are totally hardcore more cost efficient than what we want them to be. So they are very likely to get a bigger price increase than thing between 170% and 190% of the average which should get a price increase, too. 100-170% of the average usage and 30-100% of it are considered to still be okay. So 30-10% of the average would be underused enough to warrant a small price decrease. 10-0% should warrant a bigger decrease, too.

      From that onwards, the grades from the surveys were mixed in (tourney data was weightend higher than those individually but not higher than both together.) and then the factors for the algorithms determined by the grades were used.

      Let me at this part of the text thank everyone who entered data into our files, withou you the armylist analysis would not have been possible, you all did tremendous work.

    • Bringing in line to strong designs for which pricing could not be used as a solution as it would cause to much collatteral damage

      Here the KoE Murder Duke (basically Lord on Pegasus with might and divine Judgement and additional equipement) and the UD Hereditary were identified among other things as potential candidates. After a few checks those two were the only the things which as far asI remember stayed on the list.

      Then RT started to develope solutions. As more brains sometimes have more ideas, and as it was important for me that the dedicated ACS and the communities of the armies in question had a sort of forwarning, I did ask the ACS to initiate brainstorming on how to "repair" those things.

      RT then did collect ideas from those threads and did put them and their own ideas to the vote. What was part of the update basically was what won.

    • The future of the update = what you can do

      Above I already hinted that we will improve dataset usage, macros and will build up a way to link performance to lists and so weighten good performing lists more than bad performing ones. But what will (or should your) ACS do and how can you help them?

      Basically it helps if your ACS have the price changes for the lists used at ETC. (That already exists for all armies)

      Then the most pressing questions from the community will be collected and your ACS will bring back RTs answers. Please remember, that they are the messenger and arguing them doesn’t change the message. Attacking them will not be tolerated.

      If principle problems in your opinion were created or not solved by the update and they are solvable by price, than voice them. (Remember there won’t be any redesignes which aren't nerfs and even those the bar is quite high.)

      Collect and rank in a thread entries from your army, where in your opinion (more) price decreases are needed. This could be something that was changed already but the changes were not enough, or things that weren't changed. But be reasonable.

      Collect and rank entries from your army, where in your opinion (more) price increases are needed. This could be somethings that was changed already but the changes were not enough, or things that weren't changed.

      Collect and rank cases where new prices have opened previously unavailable potentially problematic combos. E.g. one more war machine, one more monster, a strong combination of special items.

      Collect and rank cases where new prices have blocked previously available combos. E.g. 2/3/4/5 specific war machines or monsters no longer fit in a category, or an often used combination of special items.

      Armed with those things your ACS will transport your feedback back to the project. Which will than work on the final version of this update.
    Thank you for reading so far and for helping us to improve

    Advisary Board Member

    Workfields: Tournament Analysis, Army Community Support, Playtesting, Community Engagement, Translation/ United Nations Blog: Inside TA. Admin of the biggest german Tabletop Board: tabletopwelt.de We want you to join the project and apply to staff.

    342 times read