Barry Farmz Here -- Comprehensive data driven farming guide

You know … it actually is good game design that if you want to farm just to get recruits and 1* ascension mats to run your training camps, 1:5-8, 1:6-8, 1:7-7, and 1:8-7 cover almost as good as it gets*. There are lots of other good, valid reasons why you need or want to farm past that stage, mostly in S2. That players who have unlocked S2 and can farm there don’t grow their rosters twice as fast because they farm there makes for much more balanced gameplay.

*Farming backpacks in 2:1-7/8 Hard does considerably beat 1:5-8, but that alone isn’t game changing.


Haha, true! I’ve done it when in a hurry to get somewhere at the same time a monster chest timer ended. But generally speaking, it’s pretty rare that I can’t spare ten minutes to auto farm 1-8-7 eight times.


Players should be notified the data about best season 2 farming location for recruits being 15-8 Hard is not necessarily accurate. Depending on the amount of world energy a player has in their possession influences the amount of recruits they can obtain, it is possible to obtain more recruits from other unlisted stages because you possess more attempts and can make up the difference through a higher frequency of attempt.

I’m not quite sure I understand this statement.

To explain what I do and don’t provide in my sheet … I list the recruits per WE for each level and identify where that is best.

I don’t try to identify what combination of levels are best to farm with any specific amount of WE.

2:15-8 Hard at 12 WE per run would be an inefficient way to use 35WE, much better using 36WE, if that is the kind of thing you are referring to.

In my favorite farming levels, I’ve identified top recruit farming at 3WE, 7WE, and 12WE as well as acceptable recruits at 4WE.


I’ve updated my sheet to include the last two avatar missions, Gadeiruses and Atlantean Drones.

The list of levels should be accurate, but the Atlantean Drone counts aren’t going to be very accurate, so the WE to complete numbers are similarly sketchy. Edit: Changed which levels I call out)

Looking at the number of waves as a proxy for drone counts and preliminary recruit numbers, here are the levels I’d recommend, despite what my spreadsheet’s data says today!

  • 2:25-1N - 5 waves; .94 Recruits/WE is the worst of Gad/Drone stages; My SS currently identifies it as fastest to complete the mission.
  • 2:27-3N - 5 waves, 1.67 Recruits/WE is way better; I expect with better data, WE to complete will be similar to 25-1
  • 2:26-8N - 4 waves, 2.26 Recruits/WE appears to be the best of any stage that completes the Gad/Drone mission, but I’d expect it to take longer than either of the two above.
  • All others - 4 waves, recruits vary between the extremes listed. I wouldn’t bother with any of them unless my preliminary data is badly wrong.

Bazza my man! I love your work, have now gained my ice kitsune and kirin avatars. Was about to embark on the kappa and gashadokouros missions but it seems to have a typo! 14-3 is listed as one of your favs, but it doesn’t have a gasha??

1 Like

Let me check and I’ll fix it. Thanks for telling me.

FYI, the estimate of kappas per stage are probably pretty good, but the next mission’s Atlantean Drone counts are based on pretty sketchy data. Which levels have them SHOULD be correct, but I don’t believe the WE to complete numbers are all that accurate.


1:23-11 has the Highest XP/Energy of all, found by an older Project:
And yes, Barry’s table has the same conclusion.


We are not talking about the same experience.

You are talking profile xp.
I am talking about hero training xp.

1 Like

Oh, thank you. And now, i have more posts understood in the right way.

1 Like

Hello @BarryWuzHere ,

Looks like a fantastic job. I’m really happy to have found this.

How is it possible to help uploading runs? I’ll be glad to contribute with this farming guide.

1 Like

YES, I’d love your contributions!

Easiest if you can do the data entry directly, which more-or-less requires a computer or tablet; Google Sheets is pretty much unusable on a phone for this at least. My Line ID is barrywuzhere if you use the Line chat app, that’s the easiest way to contact me privately and I’ll set you up for it. (If not, there is a collective effort to contribute loot screenshots and enter the data as well, although the it is running behind on data entry last I checked)

1 Like

Barry, I can’t help but wonder if you have adjusted or made allowances for people at different levels of the game providing data that does equal a set, but it is kind of unqualified? I only noticed recently after I first started farming, installed a couple of application updates and while I was levelling up frequently, I noticed a dramatic change in both the type of materials being collected and at what province they first appear. Most levels are essentially the same as their basic description in guides but for some things the math does not seem to work out; we literally just can’t run through the same stage 10 times a day, create a predictive data set and then level up the following day and find out the data is not quite 100% anymore.
Meteor frags and those rare nuggets turning up at province 5 clued me up when prior I hadn’t seen them until much later. There also now appears there is a limit for some materials that equals about 2 - 3 passes at any level before that rare item just stops appearing until the next day when you can do it over ~2 - 3 times again. I’ve noticed a couple of other odd correlations but wont list them in public for fear it might come to an end too. I figure the code knows I will likely need these 3 and 4 star items, but for the life of me, I don’t know why now.

Please don’t think that I’m looking to be negative, those spreadsheets are way better than guessing. I’ve gotten up to 23 stars worth of loot, 10 recruits, 10 monsters and two 1 or 2 star heroes for dinner for 5 WE. I call that good value.

Sft, the Bounties are very randomly, with a light weighting. If you have over 15k Datasets, you become a statistic, that close enough to the best performance.

Barry, can you give me an access to the table ‘Barry Farmz Here: Data
Entry’? I haven’t Line, but you can found me in Google Mail under the Name ‘ralgertempires’.
Did you need the first Row? I have translate the Data names, in a copy of your table. It is right, the ‘Adventurer’s Kit’ is the backpack?

Actually, no, I haven’t made any attempt to figure out if player level changes farming results or not … and since player level isn’t recorded anywhere in the data I have, I don’t think it can be done very easily.

That could be the case … but with low probabilities (my best guess/measurement is that of all crafting items found, 8% are 3* and 2% are 4*, with lower levels generally getting a total of 5-8 per run), odd patterns will occur … and it would take hundreds of runs to prove if they are random coincidences or not.

Btw, you will find yourself using them to craft items to battle big titans later when your forge is higher level and you can craft things like timestops and tornados. Also they can help you complete difficult challenge stages, rare quests, or class trials.


Cheers, I used to balk at the cost of some of the more useful battle items but have noticed that if I’m not careful, I have to spend my food/iron or lose it. Only food really, everything pretty much costs iron.

The software update may well have cured some odd things I noticed, and goods do seemingly be shifting back to where they should be. Probably more in accordance with the trends you have observed over time.

Best I can tell, Atlantis Rises will show up this Thursday with the portal, and my farming guide now shows the best estimate of how that will be that I can give. (I haven’t even tried to estimate the number of unique monsters or extra Atlantis coins they drop)

There’s now a tab for Atlantis Rises with all my current data for S2 levels adjusted based on -3WE per run and 50% more loot per run, identifying which levels should be best to farm for everything.

I’ll probably put my favorites for Atlantis Rises in the “Barry’s Fave Farming” tab as well.


It would be nice if you could add 95% confidence intervals to your estimates of drop rates. A Bonferroni correction is probably needed as well?

Just some thoughts to consider.

Yes, it would, except for how awful the 95% confidence intervals will be for the most part. Food, iron, and recruits probably aren’t that bad. But even common items like 1* ascension items and 1* and 2* crafting items have low enough drop rates that only 50-100 runs of an individual level (which I might have for 1/3 of the map if I’m lucky) should give a statistically valid result.

Unfortunately, 2*/3* ascension mats, 3*/4* crafting mats, heroes, and troops have much lower drop rates, and 100 runs really is marginal for that; I probably have sufficient data to measure them accurately on less than 10 individual levels.

I’m working on a different approach toward that, whihc I call synthetic farming rates, my best estimates of the actual drop rates for things by level. It looks like a lot of those drop rates can be broken down into different components to measure, and that aside from recruits, they don’t vary across a province, and fortunately, I do have enough data for good results in most provinces.

I hope to release this soon, probably after Atlantis Rises so I can check my estimated farming rates against some measured farming rates.


It would be interesting to see your estimated farm rates vs measured farm rates in an independent sample. If they are at least moderately correlated, that would be reassuring.