Home | About | Tournament Winning Decklists | Forums

Skill Deepdive - Efficiency


#1

Originally published at: https://stimhack.com/skill-deepdive-efficiency/

Discuss the latest article here.


#2

Thanks @Elusive for the wonderful article!! Much <3


#3

I tend to disagree with any evaluation of a card that tries to account for the click to draw the card. You often have to draw anyway (such as, to find your breakers, other tech cards, etc…), and maybe the card is just in your opener anyway. My favourite way to evaluate the economy of a card is to subtract the credits you could have instead gained by clicking for credits during the time that you used it.

For example:
Sure gamble - 3 credits (+4, but -1 for the click to play).
Liberated accounts - 5 credits (+10, but -5 to install and use 4 times)
Guinea Pig - 5 credits
Daily Casts - 4 credits

The downside of this is that you don’t get the full info of the time taken to use the cards, or any other side effects, but I’ve found its a reasonable approach in general.


#4

It’s funny, that’s exactly how I evaluate economy cards too. I went so far as to chart it even including various levels of using Kati Jones (e.g. how efficient is she if you click only 3 times before unloading, 4 times, etc.).


#5

This seems to be a compromise between the ‘efficiency’ of the card which i discuss here and the ‘tactical value’ of the card which is the amount of power it can give you compressed into a single click. The tactical value is another property of the card that is useful to look at for short-term situations. It does not allow you to do proper math to maximize your resources however.
Drawing a card in an investment. In the model you use you cannot account for when you should stop investing clicks in cards and instead click for credits.

The reasoning that a card might be in your starting hand or is your mandatory draw and thus does not count is is a common one. However these cards are equally valuable as they represent an oppurtunity cost. If you had not gotten that card, you would have gotten another, no? Therefore it holds value. They are an investment. This value needs to be accounted for in some way.

The corp ‘mandatory’ can reasonably be seen as a forced ‘click to draw’.

EDIT: Thought about this for a sec. Your math works out just as well as mine to compare cards towards eachother. Thus is serves a part of the efficiency-evaluation well.
It is only in the comparison with basic action it falters, which, if you play a deck that never clicks for credits you would not notice. Some decks do however, those reliant on event-economy and early game pressure, and there that model breaks down.


#6

Great article, thanks! Like the series.

The refrain that you “shouldn’t count the draw” is a very common one. However, as you’ve pointed out, you really need to include this cost – in one way or the other – if you want to accurately compare against non-burst options.

This gets very complicated as the draw might not be as expensive as a click. If you have an earthrise out, the cost to draw the sure gamble is more complicated. But how expensive was it it to draw and pay for the earthrise? Depends on your aggregate credit geneneration and card generation efficiencies. As you point out, this is also complicated by the fact that you get 5 free cards at the start of the game! In short, counting one click to draw a sure gamble is wrong because you can often spend less than a click to draw a card. However, ignoring the cost to draw it entirely, also prevents an apples-to-apples comparison to other cards.

To try to understand this problem, I tried to derive a closed form analysis, but failed for all but the simplest cases. However, a simulation does accurately capture both the average efficiency, and the consistency (deviation). See my old articles (there are 3) on this. The first:

http://stimhack.com/a-comparison-of-event-based-runner-economies-in-anr/

Anyone who wants to hack the (old) code: https://github.com/gparmer/ANRsim. I’ll always accept PRs, and would love to see it updated to modern economies.

I couldn’t figure out a good way to assess the tactical aspect of the game. Your notion of deck access might be useful going in that direction. I focused on “quality draws” which was a draw that was not econ, and not a duplicate. This gave some indication of the efficiency of drawing “answers”.

Your article is great, and I really enjoyed reading it. I’ll encourage any push to make understanding the fundamentals of this great game more scientific. Thanks for taking the time to write these!


#7

Tactics evaluation is always going to be tricky to do in a context independent fashion. Deck access is a good measure of flexibility and consistency in a vacuum at least.

Probably the most useful skill i can think of to supplement all this is the ability to determine how long each game might go. As said in the article, your economic options are best represented as curves, and you’re always trying to figure out how quickly to burn your oil, both in deckbuilding and in play.


#8

This may be covered in the tactical article, but I think it useful to count the click to draw for in-game decisions. If you know you need a certain amount to get passed the rezzed ice and you want to have enough for unknown unrezzed ice, you do need to count the click to draw when you compare it basic actions or lesser econ. For, example if it takes you 5 credits to break ice on a remote and there’s also an unrezzed ice and you only have 4 credits, you need to calculate if it’s worth drawing for a gamble or Stimhack or just click for credits a few times before running.


#9

I remember that article of yours, excellent work. I might link to it in my index if you approve?

I agree that the compound effect of draw is neccessary to account for. I have a hard time keeping the right level on the article if going into that however… would be very interested in an updated modeling-supported analysis if you’re up to it!
Zer0 might be a good case-study :wink:

Tactical/tempo will be challenging, but that’s why i like this writing for myself as well. Really forces me to think logically about these things.


#10

If you don’t count that, Infiltration played for money becomes an economy card. Infiltration is a card that say “either expose something or pass with no or not much economy problems”. Its only economical effect is when you’re remote locking and expose when being ahead in terms of credits, and avoid a remote run you did not need.
(its a good card, every begineer should play it x1/x2 until they are confident).

A card analysis should be totally dual : you must count like you, and also you need to count after draw+install/play (that’s what I do since years anyway).

One click gives you one credit in the begining of a game but gives you more credits in the end (most matches ends in the 55-80c range, in say 15-20 turns, aka 60-80 clicks).

If you evaluate this only like that, then Temujin becomes a 20c-4-6 card = 10c - 3 free runs - 2 run costs (aka Stimhack range : 9c - run - one brain draw).

Sure gamble isn’t good for its economy (which is +2 / +0 in my “count the click draw/click play/clicks can give 2” book), it’s good for its alpha-shot : @5c most can’t run, +4c the click after, most can run.

If you have only one way of analysing cards, then you put Armitage in all your decks (+3 to -5 in my book, +4 in yours) ?

How I count for Armitage :

  • 8 clicks to give 12-1=11c
  • 8 click gives 8c or 16c --> +3 to -5 => mean is -1, bad econ.

How you count :

  • 7 clicks to give 12-1=11c
  • 7 click gives 7c -> +4, good econ (better than Gamble).

You have to count both and judge by the functionality. Armitage isn’t a bad card, but you need to integrate it within how you’re designing the flow of your deck. For exemple Armitage is a card I play in most of my MaxX decks : avoiding half of the draw clicks gives me half of the reasons I need (the other one is its a no larla MaxX, I need cheap staying econ to play zombi mode).


#11

Feel free to link to the articles if you think they have value. The other two articles are at:

Not sure which modern economies would be interesting. A list including the coding work that would be required:

  • As you say, a Zer0 deck with modern Anarch econ. Would require:
    1. a deeper understanding of hand-size (easy)
    2. net damage logic (easy)
    3. integrating heap-breaker install cost logic (medium)
  • Good-stuff anarch w/o Zer0
    1. Add liberated (easy)
    2. This would be a good starting space as it is pretty easy
    3. Add Maxx logic (easy)
  • An Aesops/Haley deck. Would require:
    1. understanding how/when to install programs (medium)
    2. understanding program/resource type of random programs (easy)
    3. adding Haley logic for click compression (medium)
  • A Kab Opus deck. Would require:
    1. Kab tutoring logic (easy)
    2. Kab integration with SMC (easy)
    3. Kab integration with Scavenge (hard)
  • Geist. Would require:
    1. Geist logic for draw (hard)
    2. Program install logic (as with Haley, above, easy)
    3. Understanding the trade-off between popping the trashables immediately vs. waiting (hard)
  • Sunny/419 Nexus tag econ
    1. Add tags to board state (easy), and logic to avoid them (medium)
    2. Add cards: Citadel (medium), Rogue Trading (medium), Credit Kiting (medium), Power Tap (medium)
    3. Add link into board state (easy)
  • Liza w/ and w/o tag clears
    1. Same cards as above, tags as above
    2. Paragon (easy, ignoring run costs)
    3. Liza (easy, ignoring run costs)

Am I missing any interesting econs? There’s little chance I’ll have time to do this in the near future, but would be happy to help out anyone else who wants a hack at it.

The largest failing of the current simulator (strictly: the gap between how useful this could be, and how useful it is) is that it doesn’t integrate breaker costs, and a measure of consistency for how long it takes to get each breaker out. Additionally, the efficiency of econs when needing to stay over a credit level (e.g. 8 credits).

Agreed. Thinking about this stuff deeply does tend to help our own understanding. Look forward to future articles.


#12

The Pirate Decks (in Geist and in Hayley) would also be interesting. Because they do so much to change the economy (replacing cost to break with cards/draw/installs) I think it would be hard to carry over other studies of credit economy.


#13

At least someone speaking about integration !

You were talking about special runner’s economies, you miss stealth stuff (that was big 6 monthes ago).


#14

Stealth is one of the ones that is not well approached by a simulation. Stealth’s benefit has to do with its specific interaction with ICE. As the simulator attempts to understand economies on the basis of clicks spent on econ, rather than interaction, this would be difficult.

In contrast, Liza decks also depend quite a bit on interaction. However, it would still be entertaining to understand the difference in the economic power going tagme vs. not.


#15

There is another article to write about flow and integration. The problem is most of us never theorized it and most people use instinct for this.

Also, the best deckbuilder for that purpose wen’t offline 3 or 4 years ago (that was sneakdoor zeta’s associate deckbuilder, I even forgot its name).

In this deckbuilder, you could filter by keywords and functionalities. That last information, that was not on cards but deducted, helped so much to build decks then. You could filter by program destruction, type of damage, bypass, click econ… Etc.

Filtering with those functionalities gave you a selection of cards when you wanted to put a theme in your deck, and then choose the one that works the best with the flow you were designing. This website was really wonderfull.


#16

I think there is a bit of information loss due to translation here.

Could you explain what you mean with flow and integration?