Derived Character Fate Costs in the L5R LCG

by Slaven

Fantasy Flight’s new L5R LCG has a fate economy that is simple in presentation, but strategically interesting.  The choice of which characters to purchase, and how much fate to put on them is an important strategic choice, as games are often won and lost on the margins of having just one more personality an additional turn.  In the past, I built models in Excel for the old CCG (here and here), and wanted to see if I could do anything to provide similar insight on the fate economy of the new LCG.  I have tried to keep this as broadly readable as possible, so apologies if I keep things a bit simpler than you stats-nerds are used to.

 

I took data from https://fiveringsdb.com/ and formatted it to specifically break out fate cost, military strength, political strength, glory, the presence of certain traits (shugenja, courtier, bushi), and the presence of a dash for military or political strength.  My analysis focuses on the cost of Characters, as they’re the centerpiece of any deck, and provide the most room for this kind of analysis.

 

The model found that several of these aspects of the character had no significant contribution to the cost of the character. Glory didn’t significantly contribute to the fate cost of a character in the model.  Understandable, as 0 glory is occasionally helpful. Having one or more stats that is a null (“-“) did not significantly contribute to the fate cost of a character in the model. The Shugenja, Bushi, and Courtier traits did not significantly contribute to the fate cost of a character in the model.

 

Per the model, the implied ‘base’ cost of a 0/0/0 is 0.53 fate (with a 95% confidence interval of 0.25 to 0.81).  For each point of increase in what is the greater of that character’s political or military stats, the cost is .54 additional fate (with a 95% confidence interval of 0.43 to 0.65).  The cost of the lesser of that character’s political or military stat is .42 fate (with a 95% confidence interval of 0.29 to 0.56).  This is intuitive as well, as giving a Lion bushi additional military is more likely to result in an unbalanced personality than increasing their politics by one (in general).

 

From there, you can build an equation to estimate the fate cost of each character in the game.  Estimated Fate Cost = .53 + (.54*X) + (.42*Z), where X equals the greater of that personality’s Political/Military stats, and Z equals the lesser of the two.

 

That equation lets us make a few interesting observations about the Core set (and the handful of Imperial Cycle Characters previewed when this analysis was conducted). The “average” personality has an ability (and/or keyword), so the 0.53 fate ‘base’ cost of a 0/0/0 personality is not for a blank 0/0/0 personality. Assuming all personalities are largely balanced, we can assume that if a character is overcosted using this equation (their estimated fate cost based on stats is less than their actual fate cost), that said character’s ability is particularly strong.  By this logic, Ikoma Eiji is estimated to have the strongest ability in the game using the model, and certainly being able to pull Lion’s Pride Brawler  into play from your discard is a strong ability.  Similarly, Doomed Shugenja is estimated to have the weakest ‘ability’ in the game, and dying after one turn is a relatively significant drawback (to say nothing of her status as “Limited”).  Our logic here works for Shiba Peacemaker as well, which has the 2nd best estimated stats per cost, but also a significant drawback.

 

You’ll note that many of the “best abilities” characters are expensive, and many of the “best stats” characters are cheap.  This isn’t that surprising, as the model only includes fate cost, and ignores the opportunity cost of the province that each unit flops up in.

 

Implied 10 best abilities  (or the 10 worst stats per fate cost)  

Ranking Name Clan Cost Estimated Cost Difference in Estimate vs Cost
1  Ikoma Eiji Lion 4 2.56 1.44
2  Honored General Lion 4 2.56 1.44
3  Yogo Hiroue Scorpion 4 2.68 1.32
4  Isawa Kaede Phoenix 5 3.94 1.06
5  Radiant Orator Phoenix 3 2.02 0.98
6  Venerable Historian Lion 2 1.06 0.94
7  Guest of Honor Crane 4 3.10 0.90
8  Political Rival Crane 3 2.14 0.86
9  Vengeful Berserker Crab 3 2.14 0.86
10  Shiba Tsukune Phoenix 5 4.36 0.64

 

 

Implied 10 worst abilities   (or the 10 best stats per fate cost)

Ranking Name Clan Cost Estimated Cost Difference in Estimate vs Cost
113  Doomed Shugenja Dragon 1 3.40 -2.40
112  Shiba Peacemaker Phoenix 1 3.10 -2.10
111  Hiruma Yōjimbō Crab 2 3.94 -1.94
110  Obstinate Recruit Lion 0 1.60 -1.60
109  Doji Whisperer Crane 1 2.14 -1.14
108  Bayushi Liar Scorpion 1 2.14 -1.14
107  Matsu Berserker Lion 1 2.14 -1.14
106  Tattooed Wanderer Dragon 1 2.02 -1.02
105  Naive Student Phoenix 1 1.60 -0.60
104  Ikoma Prodigy Lion 1 1.60 -0.60

 

Shoju by this method has the 16th worst ability, which is a sick, sad joke.  There’s a reason he’s one of the better champs. Similarly, Lion’s Pride Brawler ‘pays’ almost nothing for her (excellent) ability, and one of the best characters in the game.  Each character is judged by their printed stats, which does have a few issues, for example, Honored General, which appears to be one of the most overcosted personalities, is based on his 3/1 statline.  However, he honors himself when he comes into play, bringing him to be a 5/3 for 4, which is actually the strongest stat line of any personality in the game on a per-fate basis.

 

Full Excel File For Stat Nerds

 

Full Table

 

6 thoughts on “Derived Character Fate Costs in the L5R LCG

  1. I was wondering if you assumed linearity in the model or if that was a deduction from the data?

    It would be an interesting fact when determining how ffg see the benefits of a single big guy vs the benefits of a bunch of small guys. Linearity would mean they see them as about even right now.

  2. Author here: Mucking around with nonlinear methods is (a) not my area of expertise and (b) would be a PITA for the layperson to understand. The results of the linear regression were pretty neat/clean, so I figured I’d just leave it at that. Now that the data is here (and formatted) it shouldn’t be hard to do that analysis, if you’re keen on it. If you can make a better model, by all means! I’d love to see it

    1. Thank you for compiling and supplying the data so neatly as well as the nice piece of analysis. It definitely saves some work. The idea of mapping how an average personality is costed according to stats is interesting and one I doubt I would have tought enough about without you.

      I am hoping to have a dig around the data this weekend but I am less than certain I will get a better model for a few reasons but it should be fun to try. I agree with your points and a linear assumption covers will give most of the information you were trying to get anyway, I am always just curious about taking these things an extra step. Sometimes it is helpful and sometimes it is not. I will certainly report back anything I figure out. I’ll look forward to your next piece.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.