Star Wars: Age of Rebellion – A Deep Dive on Dice Probabilities

A little while ago, I gave Star Wars: Edge of the Empire a reasonably positive review. After another year of playing the game, it’s time to check in again see how it’s faring, and to try to pass along some playing tips for dealing with some of the game’s quirks. If you haven’t played Star Wars: Edge of the Empire or Age of Rebellion (they’re exactly the same game system), you might want to catch up by reading that review – this will go pretty deep into the game’s probabilities and what they mean for actual play.

The dice in Star Wars: Edge of the Empire, image courtesy FFG. Click for product page.

The central, most intriguing, and most opaque idea in the game is the set of customized dice that are used to form dice pools for task resolution. At their most basic, they are simple and elegant: just add positive ability (green 8-sided) and proficiency (yellow 12-sided) dice for your level of skill, negative purple 8-sided difficulty dice for the difficulty level negative red 12-siders for reasons, and then even some more dice (positive blue boosts and negative black  setbacks, both 6-siders) for situational modifiers (cover, time pressure, assistance from an ally), grab them all, and roll them. Net out success and failure icons, and advantage and threat icons. What you’ve got left is the result. If you end up with at least one success, you succeed. Remaining threat or advantage may give an additional bonus or penalty. It’s a pretty cool idea and rolling lots of dice is fun. However, as I played more, I kept noticing unexpected quirks in the results.

The big thing was that critical hits (which generally require a successful combat check with three advantages) seemed exceptionally rare, and even weapons special powers were hard to activate (they generally require successful check with two advantages). A little bit of analysis of the dice and this made sense. The dice faces never have more than two symbols and rarely show both success and advantage. Good dice only have good faces, and bad dice only have bad faces. So canceling out all the failures and all the threats while having surplus successes and advantages is obviously rare. While there are in theory four different possible “quadrants” the result of a dice roll could fall into (success or failure with advantage or threat), in practice two of them (success with threat and failure with advantage) are predominant, with critical success and critical failure seeming quite rare unless you have a big imbalance between good and bad dice in the pool.

Another odd thing was that it seemed like having high characteristics (Agility, Perception, Brawn, etc.) was a much bigger deal than actually having ranks in skills. A skilled character with a lower characteristic (so, say, a character with Piloting 2 and Agility 2) did not seem to do as well as a character with a slightly higher characteristic (so, no Piloting skill but Agility 3).

Then finally, because nobody really had any idea what the magnitude of the effects of adding various dice were, it made it hard to know what to do with the Destiny Pool, or how to rate the importance of boost and setback dice, or how to wisely spend advantages in combat. The rules seemed to imply that a boost/setback die was a lesser effect while upgrading a die (for example, turning green ability die into a yellow proficiency die) was a bigger deal. But is this really true? It didn’t feel like it.

Now, a little bit (or even a lot) of opacity in the odds is not a bad thing. In fact, a major appeal of the Star Wars system is that dice pools are so intuitive to construct but the probability curves they produce are so complicated as to be essentially incalculable.  That said, we still need to understand some basic things to play the game well. And the only way to get there seemed to be to run some Monte Carlo simulations and see what the numbers looked like. So that’s what I did. Knowing these details definitely improved my game, but they also raised some real questions about whether the designer understood how the dice pool actually works in practice, and if there are fundamental aspects of the game system that need to be re-calibrated.

The dice pool is extremely flexible and it can generate a huge range of probability curves. So I focused on a just few questions: how common are critical successes? What are the quantitative impacts on success and critical success of adding the various different dice to the pool? How big a deal are the challenge dice? To do this, I looked primarily at a few common cases:

  • A difficulty 2 check, which is a typical ranged combat check at medium range or when shooting at a similarly-sized ship
  • A check made by a moderately-skilled entry-level character, with two yellow and one green dice
  • A highly skilled character rolling one green and 3 yellow dice

For a look at a sampling of the numbers generated, you can check out this Google sheet. I’ve focussed on positive results, because they get the most attention from the system. The SUC+2 or SUC+3 columns give the percent chance of a successful check with at least that many advantages. I didn’t go deep, running stats for 2 or 3 or more boost, setback, or challenge dice, but this should give you a feel for how they work. You could really go crazy with the data, but the visualization problems get out of hand very quickly. Also, you need to run a surprisingly large number of iterations to get the results to converge. The numbers for the very large dice pools are not exact.

First a few general observations:

  • If you’re trying to activate weapon properties or score critical hits which require a success with +2 or +3 advantages, basically forget about it unless you have a very large dice advantage. The odds of a less-experienced character with good skill (a dice pool of 1g+2y) activating her twin-linked cannons on a typical shot (difficulty 2p) are only 9%. The odds of a crit are only 2%. A ridiculously skilled gunner on that same shot (5y) still only gets +2 advantages half the time, and a crit 30% of the time (remember a crit is generally not “you’re dead”, or even double damage, but something like reducing the target’s speed by 1 or removing a defense die). Also: reducing your crit rating is a really big deal. A vibrosword with a mono-molecular, serrated edge is extremely nasty (vicious 2, crit 1). Very nearly as good as a lightsaber.
  • Quantity of dice is better than quality of dice. In virtually every case, your odds of success with N green dice are better than with N-1 yellow dice. For Piloting checks, someone with Agility 3 and skill has worse odds than someone with Agility 5 and no training until the former has trained up to 5 ranks in Piloting. There are minor exceptions when a skill level is high and characteristic ratings are only different by 1, but they are quite small. A boost die has a bigger impact than a die upgrade, and it can significantly increase your chances of success and critical success.
  • In the same vein, upgrading a purple difficulty die to a red challenge dice just isn’t that big a deal. Outside of the triumph and despair symbols (see next), the effects of the yellow and red dice on your chances of success vs failure or advantage vs threat are negligible (they help; just not a lot). In most situations a challenge die is about a 5 percentage point hit on your chances of success, with minimal impact on your chances of critical successes (again, outside of despair, which I’m getting to). Adding a setback die is significantly more impactful than upgrading a check, in that it has a somewhat greater negative impact on your chances of success, and a much nastier hit to your chances of critical success (roughly halving them in the 1g+2y vs. 2p case).
  • By far the most important impact of the proficiency (yellow) and challenge (red) dice are the 1:12 chance for a triumph or despair.  The 1g+2y shot vs. 2p has only a 3% chance of a critical, but it has a 12% chance of a success + triumph. For anything that requires three or more advantages to activate,  triumphs are the way to go. Even when you are enormously skilled and the task is easy, you are still more likely to get triumphs than 3 advantages. The overall impact of the challenge dice on the game (and so the despair symbol) is less pronounced simply because there aren’t very many of them flying around. You’ll occasionally get them through the GM spending Destiny, and a Nemesis’ Adversary talent; that’s about it.
  • The bias for failure + advantage and success + threat is very noticeable when the pool is balanced (equal green and purple dice). Due to the extra threat and missing failure symbols on the purple dice (compared to the green dice), until you have a big dice imbalance success with threat tends to dominate the nontrivial results, with failure plus advantage coming next, and then failure plus threat and success plus advantage only filling up as the dice become overwhelming in one direction or the other.

Some of this makes sense, some of it is decidedly odd.

I am definitely not a fan of how it makes levels of skill relatively unimportant. You might think 3 ranks of Piloting gives you some niche protection in that area; but it does not, someone with 4 Agility is basically as good as you. High characteristics are a big deal, and spending any of your initial XP on anything other than characteristics doesn’t make a ton of sense, from a pure min-maxing perspective. This is not great from a “making characters interesting” perspective.

The flipside of the relative weakness of skills is the imbalance in the Destiny Pool. For the GM, spending a dark side Destiny Point to upgrade the difficulty of a skill check just doesn’t do a lot – it’s usually about a 5% hit, the equivalent of a -1 in a d20-based system, with a small (8%) chance of a despair symbol. Barely worth the effort. The FATE-like “Luck and Deus Ex Machina” function of the Destiny Pool is great and I like it a lot. But spending points to upgrade checks and difficulty is fiddly and low-impact.

In fact, many of the parameters of the game seem to be built on a profound misapprehension about how powerful the challenge and proficiency dice are and the frequency of surplus advantages. There are weapons with critical ratings of 4 or 5, as if the chances of that level of surplus did not round to zero. In table 7-5, which talks about how to spend advantage and triumph in starship combat, two options for spending triumph are: “Do something vital to turning the tide of the battle, such as destroying a capital ship’s shield generator or losing a pursuing ship in an asteroid field”, and “Upgrade an allied character’s next Piloting Gunnery, Computers, or Mechanics check”. Which is deeply weird. That first use of a triumph is obviously extremely powerful – much more powerful than a simple critical hit. Yet for a character with skill, getting a triumph is much more likely than rolling a critical hit! The second effect, on the other hand, is comparatively trivial. Not only that, but it’s usually weaker than adding a boost die to the same check, which you can do with a single advantage! The proficiency dice do come with triumph symbols which boosts do not, which complicates the comparison slightly, I think the general rule holds. Boosts increase you chances of success more than proficiency dice do.

One of the interesting ways in which the dice pool works that did make sense is how armor affects defense. Armor is generally modeled through setback dice, and the interesting thing as that while it’s generally a modest reduction in your to-hit chances (on the order of 8%), it really nerfs your chances of getting surplus advantages (the impact on triumphs is less noticeable).

At the end of the day I’m not sure what to make of all this. I like the dice pool mechanic quite a bit, and if you look at it as a core system, it has lots of interesting features and it’s fun to roll lots of colorful dice. For purely narrative checks it works great, as long as you have a basic grasp of the frequency of some combinations. However, a lot of the crunchy superstructure built on top of it seems deeply suspect.

In the long term, to make a more smoothly functioning game, I think we need a serious recalibration of many of the parameters of the game system. Fortunately, it’s not too hard to figure out some tweaks to make your game run better. Here is what I’ve taken away from it for my GMing:

  1. Be very generous in handing out setback and boost dice; they are the most interesting dice to add to routine checks. Use the boosts like candy; give them to players as a reward for trying something cool cinematic, in addition to their modeling function. They are fairly strong, and in order to get surplus advantage to trigger interesting game effects (primarily in combat), characters are going to need the extra punch provided by boosts. The point of diminishing returns on the chance of success is at about 85-90%, so dice beyond that point (roughly +3 dice, although obviously the system has a lot of variability) are mainly going into generating advantage. A lot of talents also remove setback dice, so in order for that to be interesting you need to be handing them out fairly routinely. As a corollary for players, all those things you can do to gain boost dice in combat (aim, spending advantage) are worth doing.
  2. Don’t overdo the triumph symbols, either narratively or in combat. When characters are skilled, triumphs are not that uncommon (12-15% for a balanced dice pool with a couple Proficiency dice). Feel free to go nuts with multiple triumphs, but a single triumph should not be allowed to dismantle a scene or conflict.
  3. Be aware of the system’s significant bias for success with threat and failure with advantage in routine checks. This is actually mostly a feature, not a bug, and allows you to make success more complicated and (more usefully) mitigate failure. But it does get repetitive, so don’t get too worked up about it; it’s OK for the benefits of rolling an advantage or two on a failed check to be small and transitory, since it’s very common. It also argues for rolling dice only when it’s genuinely interesting to do so, but that’s good advice for any game under any circumstances.
  4. The Destiny pool doesn’t really work, because the impact of a single challenge or proficiency die just isn’t big enough. I don’t really have a solution for this. The house rule I’m considering using is to have the GM spend them for Numenera-like intrusions as the flip side of the players’ “Luck and Deus Ex Machina”, but what the system seems to really want is just a much more potent die.
  5. Find ways to get more challenge dice into the game, just so you can play with despair. As written, the system favors adding setback dice to modify difficulty, which I think generally makes sense, but we’re just not rolling enough red dice and they don’t have enough impact on the game. While excess threat is generally easier to find then excess advantage, we still want to see those despair symbols occasionally! Consider giving more opponents the Adversary talent. Minion groups especially currently really suck, and giving more imposing ones (Stormtroopers, TIE wings) some kind of levels of Adversary would help mix things up.
  6. Speaking of Minion groups, they do really suck because the extra proficiency dice they get for being in larger groups just aren’t hugely significant. I think this is fine and generally the intent of the game, but just bear it in mind. Big groups of minions are far more imposing on the page than they are in actual play. The number of groups is far more important than the numbers in each group.
  7. During character generation, I like to give players an extra 40XP to spend after spending their initial allotment (I think this number could actually be even larger). The system is so heavily biased in favor of characteristics that players are going to sensibly spend as many of of their initial points as possible on those. Making sure your character has one characteristic of 4 is a huge deal. Low characteristics, 1 or 2, can limit you because they cap the number of yellow dice you can roll, and so in the long term limit your ability to generate triumphs no matter how skilled you become. Giving players some points they must  spend on skills & talents to differentiate the characters seems wise. In an unrelated point, characters need a lot more starting money – 1500 credits instead of 500 seems closer to right. Yes, there are options for spending Obligation or Duty for more equipment, but messing with this is awkward and 500 credits is just ludicrously low.

This may all sound negative, and I do think it’s true that there is too much in the game that just doesn’t make much sense as designed. Unfortunately, there are a lot of things in the game – weapon ratings, talents, advantage spends – that clearly are at least somewhat misaligned. But, the cool thing about the system is that fundamentally it’s quite simple and elegant and has a lot of potential. Once you’ve gotten a handle on how things actually work, it’s not that hard to hammer things into some form of order. I still need to find a fix for the destiny pool, but I feel like as I play more and get more of an understanding of how the pieces fit together, and how to use them properly (even if that’s not exactly what the rules say), I’m happier with the game. While it’s always going to be a game in need of constant tinkering to keep working, that tinkering is not particularly onerous, Star Wars is fun, dice are fun, and the core system is good.


16 thoughts on “Star Wars: Age of Rebellion – A Deep Dive on Dice Probabilities

  1. This going to be my first tabletop game (I’ll be playing it a friends home soon) and I’m reading all this stuff online..and it’s kind of intimidating for someone whose too young to have played games like this when they were first popular.

    Still, your original review and this analysis are very helpful 🙂

    • If you’re new to the genre, there are two Beginner’s Games for this system (I think they’re just called “Edge of the Empire Beginner’s Game” and “Age of Rebellion Beginner’s Game”) which are really very nice products – they have all you need to play and are *much* easier to get into than trying to wade through the morass of the core rulebooks. They include nice maps and playing pieces and dice. In fact, the Beginner’s boxes are in many ways superior products to the core game (they have some really good GMing advice that isn’t in the core books, for example).

      Anyway, it depends a on whether you’re a player or planning to try to GM. The system is pretty easy to get into as a player, so I wouldn’t be too worried. If you’re planning to GM, I would really recommend starting with the boxed sets.

  2. This is interesting. My group had noticed that pumping the characteristics is the way to go, and I had noticed that getting success + threat etc was very common but the rough odds changes were a mystery to me.

    The original dice system was developed by Jay Little for the WFRP 3rd edition, it was then transplanted into Starwars by someone else. It would be interesting to see how the odds vary between starwars and WFRP3 and what was changed. Was it a straight port or did they try to scew the results in a certain direction, if so for what reason.

    I play a number of OSR (old school dnd clones or varients) and as the numbers system is open and rather similar between designs it is often a point of discussion as to how the numbers have been tiled one way or another. In general in DnD line games you can see the designers intentions at a mechanical level.

    • Jay Little is also the lead designer on the SWRPG. He talks a bit about “terminal outcomes” here:

      Where he talks up the vast number of possibilities the dice can generate. Which is true in theory, but in practice it appears many of them are vanishingly rare.

      I do really like this idea of the dice pool in theory, but in practice it seems simpler and more transparent systems like Numenera or Night’s Black Agents get better results. I really think the dice pool in SWRPG needs at least one more major revision before it works as the designer seems to have intended.

  3. Very nice run-down of the quirks in the system! I have only just started playing the game recently and I can already see some of the issues that you high-light. Particularly, I think I made a critical error in designing my character. In the interest of making my character more well-rounded, I spent most of my initial XP on Skills instead of boosting my Characteristics. This has resulted in me not being able to do much of anything in gameplay and has been kinda frustrating. My dice pool is never very large and since we are almost always operating as a team, I keep having to hang back and hoping other players fail just so I get a chance to play. Not very fun. You make some good points that I will bring up to our GM.

    • Yeah, the bias for characteristics over skills is a really critical thing about the system that if you’re generating your own character and you miss it you can be screwed. I’ve actually played SWRPG primarily with pre-generated characters and one-off adventures, not as part of a campaign, so this issue didn’t become clear to me until I had played a fair amount. In most traditional-style games I like to borrow the idea from Gumshoe that in investigative-type circumstances a player with a rank in a skill can succeed automatically. So I would ask the players, “who a rank in Mechanics” or “who has the best Mechanics”, and someone would have a rank or two, but then a character with a high Int would protest that she was actually better at it. We then looked and a simple analysis of the dice revealed that just in terms of adding success symbols, adding a green die is noticeably better than upgrading to a yellow die, and that’s where things got weird because we realized skill ranks weren’t working intuitively.

      There are a lot of things in the SWRPG that bug me a little, but don’t really affect my enjoyment of the game that much. But this dominance of characteristics over skills (and the fact that buying skills during character generation for the most part is a trap) bugs me a lot, and I think if you’re a GM you really have do something about it – either by giving characters another 40-50XP after initial character generation and making sure they realize they should spend most of their initial XP on characteristics, or something (I don’t think *limiting* what players can spend on characteristics works, I think you just want to start people at a higher power level).

      The One Ring RPG gets this whole idea much closer to right I think. As much as I like the idea of SWRPG’s dice pool, I think The One Ring executes the idea significantly better.

  4. We have been playing for a few months now, and the more mathematically inclined among us are starting to note the tendencies in results you describe so well here. I am glad to see an analysis which tries to get to the heart of how to approach upgrades versus die adds, and your findings here are quite close to what I have been adjusting the pools to be over time, with the exception that I do use Challenge Dice a fair bit, and the group still reacts to them with respect (however unwarranted that might actually be). I imagine based on what I have read here, that this respect will fade.

    One small thing that I would add to this mechanical analysis is that the game asserts quite often that the rules are there to support a narrative approach. When my group first hit the, “Can I do skill X unskilled?” question, even without that sort of assertion, the game’s style means that even with experienced players with established habits from other games, we can generate a culture of play which collaborates on and evaluates what is needed in a scene. We can rule on a case-by-case basis if they could use an unlearned skill or not, and assign whatever setback or difficulty level was appropriate if it made sense for them to try. I won’t be put in a situation where a character with a 4 Agility is going to out-fly a Pilot with 3 Agility and 2 Proficiency. If they can even get off the ground, they will be facing very different challenges than even a beginner pilot in identical situations. The example you use makes a good point, but it is not going to be a determining factor for many GM’s, I think. Even when it is something simple like telling a lie, it is not hard to swallow that a person with more Presence will be able to do so better than character with less – even with training to mitigate that lack. Where this example starts to bother us is when it occurs in a skill roll they shouldn’t have made in the first place. We are what keeps those odd situations from happening~

    Thanks again for the great post. I found it to be very useful!

  5. One thing to keep in mind is that the important aspect of the dice pool is not the cumulative probability or the expected payoff from thousands of rolls. There are going to be a modest number of rolls in one night. I would guess my group that plays 5 hours every other week makes 100 – 200 rolls a night among the 7 of us. The major factor is the standard deviation which is high. That is where the gambles pay off.

    Also, the dice are narrative so that is where their real promise is. 3 greens don’t have the same narrative strength than two school buses (aka Proficiency dice!)

    Great article!

  6. In WFRP 3rd dice gained from skills add to your stat dice rather than replacing them. The fact that yellow dice upgrade and replace green dice in EoE seems to be one of the biggest changes to the dice mechanics of the two games. I wonder if making red and yellow dice additional rather than upgrades would fix some of the odd odds you have found.

  7. This is very interesting. I bought into the idea that Y was better than G, even though I felt that not spending everything you had at creation on characteristics was a big mistake. I changed my mind later on when I felt that to be better at specific things I could get more out of a beginning character by spending on skills, which the data obviously does not support.

    This does seem like a very large oversight. I know you said in the comment section that you don’t think XP spent on characteristics should be limited more than it is, but that might not be such a bad idea if you want to play a long-term game. I haven’t played enough long-term games to say either way, but I would like to know how things really get based on enemy matchup suggestions due to character earned XP levels, vs true power of the characters that did not buy all the characteristics they could.

  8. A great breakdown of the only marginal effect of the yellow and red dice. I have a house rule that
    1) a double advantage/threat on a yellow or red also adds a single success/failure respectively
    2) a double success/failure on a yellow or red also adds a single advantage/threat respectively

    The base odds with **3 ability dice** are
    At Least 1 Success 87.5%
    At Least 2 Success 59.4%
    At Least 3 Success 28.9%

    At Least 2 Advantages 59.4%

    The changes results in the following with **2 ability and 1 proficiency** dice

    At Least 1 Success 91.7%
    At Least 2 Success 66.7%
    At Least 3 Success 34.9%

    At Least 2 Advantages 60.4%

    At Least 1 Success 93.8%
    At Least 2 Success 69.8%
    At Least 3 Success 37.1%

    At Least 2 Advantages 63.5%

    Now the odds are slightly better if you have a proficiency in a skill

    This also means that **2 proficiency** dice now have better odds overall than 3 ability that just feels right

    At Least 1 Success 88.9%
    At Least 2 Success 55.6%
    At Least 3 Success 19.4%

    At Least 2 Advantages 41.7%

    At Least 1 Success 93.8%
    At Least 2 Success 64.6%
    At Least 3 Success 22.2%

    At Least 2 Advantages 47.9%

    This is also calculated by the amazing dice statistics spreadsheet found here

  9. My gut reaction to the skill/attribute problem is to make a penalty for non-skilled attempts at something. Something like adding an immediate +2 difficulty die (or setback die) trying to do something unskilled. That way it pays to get at least one rank in a skill. Sure, Pilot 1 Agility 3 is marginally better than Pilot 3 Agility 1, but Pilot 1 Agility 1 is a lot better than Pilot 0 Agility 3 (I think).

    • Using Monte Carlo rolls, adding 2 Setback dice is what it takes to not only sufficiently lower Success chance but also beat back the glut of Advantage that another Ability die brings.

      Since players tend to get thrilled by the possibility of Despair, you could instead add just 1 Setback and upgrade a Difficulty to Challenge. But, owing to the weakness of straight upgrades, the high-Characteristic-no-skill attempt will still have a probability edge in both higher Advantage and lower Threat.

  10. I like Ant’s idea of adjusting the yellow and red dice to get results that feel right for my game. I haven’t done the numbers yet but another adjustment I might investigate is allowing each yellow die to add one Advantage automatically. The idea would be that skilled actions substantially reduce risk of disadvantage and increase the chance of crits and special moves.

    The same principle would work for the red die. This method means it’s very simple to remember and apply, and the dice can still be read as they roll – with skill giving a little extra boost.

    Possibly, allow spending a strain before the roll to automatically add a success as well, but my sense is that might have too much of an impact. I’d want to look carefully at the numbers.

  11. Great article! I appreciate your in depth analysis of system. It seems like a lot of work to fix the game with house rules. I think you have convinced me to use a system other than FFG’s game the next time I want to play Star Wars. I like Savage Worlds and there are two great fan made Star Wars settings for it ( and I think I will go with that.

  12. Thanks for the article. Still relevant 2 and a half years later.

    A character’s characteristic rating does have a cost, as you know, of being twice as expensive as skills and really only being able to be modified at character creation or deep into a tree. So I assume they figured that it warranted being just as powerful to your pool as your skill rank.

    Still, I agree with you. I think that if you are skilled, you should always be better than someone with “natural” ability. So to treat the skill as being more important, I think this may work:

    Base dice are set by your skill rank and your characteristic rating is added on as upgrades.

    So let’s take a character with Agility 4 and no ranks in Piloting (Space).
    The pool would be: 2Y (0 base dice and 4 upgrades)

    Now let’s take a character with Agility 2 and 2 ranks in Piloting (Space).
    The pool would be: 2Y (2 base dice and 2 upgrades)

    The only other thing you might want to implement is that you can only have as many triumphs as you have ranks in the skill, otherwise a triumph counts as 1 success and 1 advantage.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s