A Comparison of Event-Based Runner Economies in ANR

Originally published at: A Comparison of Event-Based Runner Economies in ANR - StimHack

Discuss the latest StimHack article by @YCombinator here.


Really loved the article. One major flaw I see in the DE calcuation:

Doesn’t seem fair to only include one draw package of (Shaper) draw events and then ProCon as the only Resource-driven draw engine. What about Mr Li? What about Wyldside? What about Earthrise? What about Inject? There are many other draw engines in the game.

Thanks for the analytics!

1 Like

I don’t have enough time to read to do this really justice with the comment yet, but this is something I’m going to really enjoy, thank you for the time put into it!

in before spags meme


I sincerely appreciate the suggestions, and couldn’t agree more. Look for future articles that venture into other draw and credit cards. There is far too much to analyse to fit into a single article.


I was hoping to see my pet (loses-to-NEH) shaper econ engine, 3 opus + 3 SMC (for consistent turn 1 opus) + 3 diesel + 3 QT. But it would probably show up a lot worse in the graph without accounting for astrolabe. Likewise, I’d like to see graphs that account for LARLA. Popular wisdom is that PPVP is average until the levy, and utterly bonkers afterward.

Not to mention sec testing + desperado + masanori… Nevertheless, really great, groundbreaking article. It’s good to have some formal analysis to confirm/reject the community’s hunches.

Reminds me of a few months ago at my FLGS, someone asked “How many credits do you suppose a click is worth nowadays?” “Probably like $1.5” Based on this, looks like it’s still 1.5 for the runner, but for the top corps (blue sun glacier, RP, and NEH), I feel like their clicks are worth closer to $2.


Based on the article it seemed like it was still around 1 for the runner, right? Where are you getting your 1.5?


“By the end, efficiency is topping out at around 1.4 credits per click. Note that the procons decks increase in efficiency as the clicks go on due to variance in if we drew Professional Contacts early or not.”

But only at the end of the graph, so the real average probably is closer to 1.
Surprisingly I think I understood most of this article. Thanks for all this rigorous work and analysis, can’t wait for more of your articles and DES!

Best ANR article to date. (Yes, better than mine.) The only thing that could improve it is a link to a GitHub or Bitbucket repository. (Python? Looks like you used matplotlib for the graphs, at least.)

Looking forward to more, but this gives me enough to go on when looking at econ engines for MaxX and Hayley.

1 Like

Supplement the credit/click efficiency with the draw/click efficiency and in terms of actual usefulness a click could possibly be more like 1.5c rather than 1.0c

1 Like

Excellent paper! I have just one question: in order to produce your plots I guess that you assumed a standard 45 cards deck and included 3x of the mentioned cards for each of the mentioned engines, right? If so, it might be interesting to investigate the changes in the efficiencies you pointed out with respect to different deck sizes and quantify the loss of efficiency for decks that do not include the complete play-set of the mentioned cards.

Completely agreed. The outlay of future articles:

  • What must change in an econ package to change the meta. Look at some new cards WRT this.
  • Look at the new event-based anarch draw/cred cards.
  • Look at non-event-based econs.
  • Look at identities. Andromeda vs. 40 card decks vs. Maxx
  • …and about 6 others.

Suggestions like this are great, thanks!


Man, I can’t believe I didn’t think to add LARLA into the sim. Great suggestion that I’ll take into account in the future.

Other econs (Noiseshop, crim, etc…) will be considered in the future, but the software isn’t quite to the point where that is possible yet.


When I calculate my click efficiency, I calculate a draw efficiency - Ie., how many clicks to draw out my deck.

I then figure that I will have drawn all my econ. I add the clicks to run the econ. This then gives me a rough $/click. I also calculate how much surplus cash the (ie., not needed for installing econ, breakers etc). This tells me how much money I have for runs.

Useful, as it tells me when I’m severely under econ’d.


Mr Li is actually really hard to use in a simulator like this - whether you chose the econ card or the action card (meaning non-econ card) varies based on a lot of factors, including how good either card is, both in general and in the current situation (for example, you’re likely to ignore a redundant breaker, unless it’s a Cerberus/CyberCypher/Faerie), This leads to a lot of complicated decision making for mathematical analysis. And, whatever system you build, it has to fire every single time you draw - which adds up.

Inject also provides pseudo card selection, allowing you to recur the trashed cards. When do you do so?

The shaper packages are easier to analyze, because they work off of raw draw power, and they are generally turn timing independent. You just spend clicks and cards and get money and cards.

1 Like

Yeah, I get that it’s easier, but there are always assumptions made in a simulations like this and they never tell the whole story.

I’d think with Mr. Li you could prefer a Quality Draw over an Econ Card (does the simulation track multiples of breakers and/or other one-ofs?). With Inject the number of quality draws would be equal to the number of non-unique-type cards you kept in your hand (maybe half of a Quality draw for shit in heap?) and also track the economic swing.

On the Security Testing/Desperado side of the economy, you could make general assumptions on how often you will hit Security Testing/Des (maybe 2 out of every 10 clicks or something?).

Love that you are doing this @YCombinator!

Got some time to sit down, read through the definitions you’ve put down, and this is interesting stuff! I had never looked at the game in that way, and esspecially the derivative graphs were very informative.

ty once more.

I don’t understand all the love for Mr Li. I did an analysis about 18 months ago (in a Stimhack thread) that demonstrated that you’re almost always better off simply drawing all your cards blindly with mass draw events than searching for “quality” cards using Mr Li. Admittedly I didn’t factor in the $ cost in playing cards like Quality Time or the deck-building restrictions regarding influence that might morivate one to choose Mr Li, it was just a click counting exercise.

The main case for Mr. Li is the fact that there is a hand-size limit that means you might not want to over-draw. Fast-forward to 2015 though and we find that the heap is now a more significant resource for the runner and there are a greater number of options for recursion (as well as more draw events now / imminent in O&C), which would seem to make the case for draw events even stronger.

1 Like

Cool article. Wouldn’t an infinite hand size skew the draw efficiency analysis? I realize it’s hard to factor in what cards to keep.

Isn’t Mr. Li mostly played in criminal decks where recursion is at a premium? Sure there’s SOT, but everything else comes at a cost.