Big data based Provisional Cost for fair and varying deck styles

[ClimbHigh]

We're currently exploring the option to have only two copies of each bronze cards. The cost is based on the power level of the cards ability.

One reason current Gwent few different is its controlled level of randomness, you build a deck you know how the deck work in most of your games. 3 copy of bronze in a 25 card deck feel right. I like this very much instead of just 2 copies.

The one major cons of this is, since the lack of randomness, the cards seen play is just a fraction of the total card pool, because the meta will not play cards that is even slightly less valuable, making diversity very difficult. Reducing card draw consistency (like in a Ciri Nova deck, even with 2 Recon) is one way to see more cards in play. This is the only advantage I see why 2 bronze is good (of course there is other reason CPDR or hive mind pls enlighten me).

Thus I suggest a mechanism for HomeComing:

1. Power level of a card = points differential during its lifetime

• Cost equals to point differential after unit is deployed on board until its deploy chain completed. If Card Advantage card, also count the power differential of the extra card drawn. If engine, point differential during each turns/round is also counted.

Odrin’s power level = 8 + the 1 point boast every turns, all accumulated​

Blueboy Lugos = 9 - 3 + the 1 point damage per opponent units every turns, all accumulated​

Artifact compression’ s power level = [points of opponent removed - 2]​

Tormented Mage 's power level = 2 + 8 (value of thunder this time)​

2. (Provision) Cost of a card = Average power level of this card in all Gwent matches past season (hence Big data); calculated separately for a specific leader, separately for a specific game mode​

Half elf Hunter's cost = 12, no matter used by any of the current leader​

Scale Shield will provide different power level when used by a different leader​

• Some cards’ effectiveness depends on the leader, some don't
• once every new season, gamers can try find new combo due to cards cost reduction induced by past season

3.Winner’s cost bias multiplier​

• If a match is won, the provision cost of a Half elf hunter will registered as 12 x (a winner bias multiplier) instead of 12, during CDPR’s calculation of 'average power level for all Gwent matches past season’
• If CDPR set this bias high, cards used in deck that always lost will become more cost effective next season, encouraging more evenly distributed utilization of all cards made.

4. Variance to the final cost, and Sort from highest final cost in UI ​

• Intentionally add -5% to +5% to final card cost after Winner’s bias applied. This make deck building more organic that player experience is needed to tell if a card is a real bargain or not.
• The UI sorts from the highest final cost by default, with auto refresh, allowing player easily see approx. what was most powerful card in each Gold/Silver/Bronze category. The cost (in numbers) aren’t published. Don’t want to make this min-max game. UI does show which card has a significant cost drop compared to last season.
• When player spends points building in deck builder, cards they no longer afford is auto dimmed.

5. Varied deck styles​

​

• For a specific game mode, there can be 1 or 3 separate point pools for G/S/B. The default G/S/B number of card need not be 4/6/15~16. The default no. of duplicates for G/S/B needs not be 1/1/2~3. Minimum hand needs not be 25 cards. All these can be changed by CDPR and their numbers shown in deck builder UI.

6. Higher Consistency Cost for each extra copy of same G/S/B​

• Putting an extra 3rd Beast master into the deck into 2 duplicate bronze game mode, this beastmaster will cost more, 12 x (a consistency multiplier) instead.
• This multiplier exponentially increases for additional extra copy in deck (4th, 5th …)

7. Inconsistency Cost reduction multiplier for decks with 26+ cards​

• Significant cost discount will be given for the 27th card and onwards.
• Eventually the multiplier will enable making for example a 52 card deck with roughly 8G 12S 32B, even with somewhat higher quality cards to compensate for the increased inconsistency.

8. (Optional, debatable), player with comparatively lower deck cost got the red coin​

What this accomplish or enable:

• Automatic, fair and organic balancing of all cards by cost; no card left behind
• High winner cost bias multiplier setting can encourage less good card sees play
• Low consistency multiplier setting can encourage deviation from the standard 4G 6S 2X8B decks
• High inconsistency cost reduction multiplier setting can encourage 26+ cards deck
• CDPR’s alternation of various multipliers & provision points for G/S/B enable much varied game modes (e.g. multiple fraction constructed, hard mode Thronebreaker), easily changed for between season if desired.

 

[ClimbHigh]

does it mean no one like this suggestion, if so let me know the reason

 

[sfruzz]

I think the mechanism you propose here is too convoluted. Your previous suggestion of having the “provision” cost tied to the copy of cards, is easier to explain to the playerbase and easier to implement.

Automatic, fair and organic balancing of all cards by cost; no card left behind

As the provision cost is not updated “on the flight”, you would still have unbalanced cards. Cards that have a high value / provision ratio in that specific season.

In short, I fear you would turn Gwent into a large optimization problem. Something like the “Traveling Sales…”, pardon “Traveling Witcher Problem”. While your suggestion is well written and explained I don’t think it would be beneficial for the majority of the players.

 

[Morh_Shion]

I think the mechanism you propose here is too convoluted. Your previous suggestion of having the “provision” cost tied to the copy of cards, is easier to explain to the playerbase and easier to implement.



As the provision cost is not updated “on the flight”, you would still have unbalanced cards. Cards that have a high value / provision ratio in that specific season.

In short, I fear you would turn Gwent into a large optimization problem. Something like the “Traveling Sales…”, pardon “Traveling Witcher Problem”. While your suggestion is well written and explained I don’t think it would be beneficial for the majority of the players.

You seems concerned with players understand his proposal.
I think this is not something for us to be concerned off. An auto-system to catch the % to defines what is a balanced card what would help a lot providing statistics for the developers decide what should be changed or not.

I liked a lot most of his ideas, indeed some of them seems convoluted, but I think we should let the developers figure it out what from those ideas can work and what may not, them put into test, and adjust accordingly the players feedback after a closed test in a test server for example.

 

[sfruzz]

You seems concerned with players understand his proposal.
I think this is not something for us to be concerned off. An auto-system to catch the % to defines what is a balanced card what would help a lot providing statistics for the developers decide what should be changed or not.

As the system would affect deckbuilding, the players should be comfortable with the idea and having a general grasp on it. It's not just affecting the raw value of cards, but also "provisions", which I think would make the system hard to understand for the majority of the players. Plus, probably the developers have already implemented some big-data analysis tool, which they rely on when tuning the +/- point strength of the cards.

Still I like the constructive attitude of the suggestion and I'm in favour of a simplified version of it.

 

[ClimbHigh]

The core idea is in (#1 #2 #3). And of course there are some catches for any systems thus I add a few sub-points to supplement (or convolute) the core idea.

Provisional cost is a new & major thing that CDPR has to balance in HC, they does not have good track record to balance (CDPR either change the card completely or do +/- 1 pts, while leaving most cards unusable). The core idea should helps a lot to avoid another balancing nightmare, while all cards should see play. Should be much better than any non-big-data methods.

The the provision cost is not updated “on the flight” because I want a common worldwide cost so you can build a deck same as any other player for that whole season. Make the cost stable for a reasonable amount of time, so no one need to fit in the ‘last card’ of your deck before each game you play, if the cost is updated ‘on the flight’

 

[Witcher3point0]

Why not try going with deck may contain up up to two sets of 3 bronze cards and up to 3 sets of 2 bronze cards, all other cards may not be duplicated during deck construction.

Or did I miss the the mark altogether?