Leveling 3, 4 and 5 star heroes (Analysis, Advice)

Thank you for the advice

Sorry, I’m a bit confused… Is it better to do one 1* at a time or ten 1*s?

I would suggest going 10 at a time, for then you wouldn’t have to constantly be upgrading. If you’re creating troops via the training camp, say lvl 11 “Extra Low Cost” then you’re getting one troop every two hours per training camp it doesn’t take long to get 10 troops, and the % chance for improving their skill goes up the more of the same colored troops you use.

1 Like

Practical Answer:
It does not matter if you use one 1* hero at a time or ten 1* heroes at a time because of the way skill rolls currently work in the game.

3* Heroes can easily get 3/8 special skill at max tier, max level no matter what you do. Same for 4* heroes that lack Tier 4 ascension materials and 5* heroes that lack Tier 3 ascension materials.

For players who care about statistics and RNG:

1 Like

So basically the odds are the same whether I use one hero to train or ten? Though I technically increase the odds of not getting it maxed if I use ten and don’t get it. I still don’t understand the math -_-

Think of it this way

Part 1

1* Heroes give the best +% Skill roll per XP. Since your skill rolls are limited by how much XP a hero takes to level, 1* Heroes are the best ( there are special circumstances).

5* Heroes have the best special skills, you they are the most important to increase ( Note 1 )

Part 2
You have a week of vacation to finish a roof repair. You can leave work 1 hour early for 56 days ( roll one 1* hero ) or take the whole day for 1 week (batch of ten 1* heroes ).

You cannot work on the roof repair when the weather is bad ( fail your skill rating increase ). So leaving work early for 56 days gives you a better chance being able to work ( getting some increase ) Edit: because it is very unlikely to have a 56 day storm. But if you take the whole week off, and a 10 day storm shows up, you get zero work done ( no increase except ascension increases ) where it would have only ruined 10 hours if you had used the leave early option.

But the odds are better for 7 days in a row of good weather than 56 days in a row, so each individual day ( skill roll ) has a better chance of being good ( increasing the special skill ) with taking the whole day.

Basically the whole day decision ( batches of ten ) lumps a lot of your chances together for good or bad. Where leaving early ( batches of one ) spreads out your chances so you only miss the normal amount of days due to bad weather.

But even if the weather is really good all week or all 56 days, you may find unexpected damage and not have enough time to finish the repair before you run out of time ( you roll bad in batches of one or batches of ten and still get 3/8 3* 3.50 hero ).

Note 1)


I love your illustration!

Looking for advice. Is it worthwhile to sacrifice 3 star heroes to level up 4 or 5 star heroes, or even to level up other 3 star heroes?


Either ask you alliance fellows about your exact roster or start a new topic on the forums. A lot depends on the heroes involved, the heroes you have, the level of your base, what rare quests you can complete and what titans you fight.

I can complete all rare quests and have five Gunnar ( 3* blue tanky tank with Spirit link ), if I get a six I will definitely use him to level the other five. Seriously considering winnowing them down to three since 3* heroes are so squishy compared to 4* and his offense is low, so not good for clean up in war.

I’m levelling a Grimm currently, feeding only pack of 1 :star: and 2 :star: heroes (10x1 or 10x2) while feeding the other rainbow fodders to a fully maxed special hero.

What is the star ratings for common, uncommon, rare, epic and ledgendary. Ive been trying to get a 3*red card but keep failing in my attempts. Do i have to be at a certain level to be able to train one?

TC13 = 3* with a chance of 4*

TC20 = 3* with a chance of 4* and 5*

Elemtal/Epic Summons = 3* with the chance of 4* and 5*. No level requirement, just fork up the gems and pray you’re lucky.


My Wu Kong’s continue to be lucky. My second Wu Kong is 7/8 2.45 which guarantees he is 8/8 3.1 ( highest ascension with farmable ascension materials ) while Gryphonkit ( my wife ) is having trouble leveling her Wu Kong’s special skill.

Like Kiril, Wu Kong 3/8 or 4/8 is almost as useful as 8/8 because of their special skill design and how those skills actually work versus rare quests, farming, titans, raiding and war.

@Gryphonknight Thanks very much for the work on this(and a few other posts of yours I’ve been reading). I’d been hearing people argue for specific leveling up strategies, but this really puts everything relevant in perspective.

I’ve considered your argument of 4* and 5* being just as important considering their ascension limitations despite their larger amount of XP. I agree, but I was wondering if you crunched any numbers regarding special skill value probabilities. For example, using only 1s of the same colour what’s the probability that a 4 reaches 3/60 with special skill of 8/8. I guess that thought just puts the practicality of the argument into perspective.

[It’s not clear by my comment, but a lot of time passed at this point.]

Decided to learn math again and found a binomial distribution calculator. Here’s an example with results:

If I just pulled a 4* hero that I want to level up and put on my team, but I only have the ascension mats to get them to 3/60 this is the probabilities a certain special skill level was reached. 4* 3/60 takes 83871 XP. Special skills start at 1 and you would acquire 2 during the ascensions to 3/60. Assuming groups of 10 for highest probability.

538 1* random colour:

P(8) = 79%, P(>=7)= 90%, P(>=6) = 96%, P(>=5) = 99%

466 1* same colour:

P(8) = 97%, P(>=7) = 99%, P(>=6) = 99.77%, P(>=5) = 99.96%

207 2* random colour:

P(8) = 59%, P(>=7) = 78%, P(>=6) = 91%, P(>=5) = 98%

180 2* same colour:

P(8) = 91%, P(>=7) = 97%, P(>=6) = 99%, P(>=5) = 99.86%

299 (E)LC Training output 50% 1*/2* random colour:

P(8) = 65%, P(>=7) = 81%, P(>=6) = 93%, P(>=5) = 98%

255 (E)LC Training output 50% 1*/2* same colour:

P(8) = 93%, P(>=7) = 97%, P(>=6) = 99%, P(>=5) = 99.88%

Conclusion: Your theory hold weight even for 4*/5*s assuming ascension limitations. ELC output is decent. Colour matching is better.


I juste got a wu Kong is it a good or a bad hero do I start levling him or rely on my bane and keep levelling him better

1 Like

Keep both but level Wu Kong first

1 Like

Was bedeutet das P ?

P (8) = 65%, P (> = 7) = 81%, P (> = 6) = 93%, P (> = 5) = 98%

P = Probability

P(8) = Probability that Special Skill is at 8.
P(>=7) = Probability that Special Skill is at 7 or higher.

1 Like

Ooo. SCIENCE !! :grinning::grinning::grinning:

Yes, yes. its really. Ooo. MATH !!

I suspect whoever designed the skill level roll, and ascension requirement for each tier, used this particular data set. The P(>=5) = 98% is the most interesting data point.


That really makes 4* 3.60 the Most Valuable Heroes ( MVH ) of alliance war.

Now if we only had a 4* Heroes of the Month.


Argh. Why is my extra fast training ( recruit training level 19, 2 recruits per 1* hero ) taking so long.

Must also start farming rugged clothes in province 7.

As 3* heroes’ ascension materials are easier to acquire, but if the special abilities 8/8 is the main focus, would you say it would be best to still use 1* and 2* heroes of the same color at any stage of development?

3* heroes have 2 guaranteed special up opportunities (e.g. ascension time), so an ideal situation will be for them to get to 6/8 before their first ascension. But we have seen plenty of maxed 3^50 3* heroes but stuck at 4/8 to 6/8 skill and in some more extreme cases 3/8.
Let’s say for a maxed 50, just to complete his/her skillset (stuck at 5/8), would you recommend:

  1. feed a 1* or 2* of the same color one at a time (as each time is a separate occurrence)
  2. feed a mixture of 10x 1* or 2* of the same color at once
  3. go for the guarantee 100% and try to feed 4x of the same hero?