Ranking Theory

I’m writing this post in the midst of Xiu Min the Uber whale ranking. This post isn’t designed to give you an answer to all your damage capping woes but to help you understand the point trade offs in rankings to understand how to focus your team. Xiu Min is a good example because some of the faster builds miss out on damage capping which in the end just sets them back.

Points

The basic point structure is:
Combo = Avg combos x 5000
Time = remaining seconds x 500
Damage = Max damage x 0.00025 up to a max of 10,000
Button penalty = number of floors buttoned x -5000
Awakening multi = sum of the special awakening points you can earn in the new system (more on that later)
The total of these categories gives you your final score before checksum.

Awakening multiplier

This is a new mechanic Gungho released in JP a month or so ago to mix-up the ranking game. In the end it just makes it wildly more whale (from the rankings we’ve seen so far) as well as more RNG based. The basic premise is that you will be rewarded for matching certain awakenings like TPA, L-shape and others. I’m really not sure what they were trying to target with this because fixed teams means everyone has access to the same awakenings but what changes is the board distribution which can just inject more RNG into a ranking system that is already RNG based. Or, on non-fixed we get whale strategies that spam bicolors to stack L-matches and TPAs. This system ties in very closely to the combo score system because of the orb value as well as the time system for the time spent matching the extra awakening style and whether it’s worth the investment. Gungho sets the value of these awakenings on a per ranking basis. Therefore, if they set them too low they just won’t be worth matching, and if too high, they will just be abused. Let’s look at the other point systems and then see how everything fits together.

Time Left

Lucky for me, explaining time is pretty simple. It’s the number of seconds you have left when the clear message comes up. Unfortunately, this can include RNG and other factors as well. I’ve noted with toggling power saver on my Samsung for rankings that I can lose 5s or more in a 10 floor ranking to the phone speed change. This was tested on Indigo ranking. Others have noted that iPhone vs Android has different delays as well. Usually rankings have enough point diversity that iPhone or Android time differences shouldn’t make or break a crown (however, anecdotally Android feels a bit laggier). Just make sure to remove any power saver options you may have before starting your run! The second factor that can be purely RNG depending on the dungeon is the drop/no-drop RNG. This was very clear in DBDC ranking with Myne. Depending on how the ranking is setup, enemies might drop nothing, coins, eggs, or a possible chance of some of those. Myne was 1-2% chance of dropping an egg which is a shorter animation than dropping coins. Dropping nothing is shorter than both of those. It’s worth noting that Gungho has disabled all drops so this should not be an issue in the future.
Some general time cost timelines:
Every additional ping is 0.2s
Cascadeing takes an extra about 0.2-0.3 secs to finish each step
When brute forcing, start top to bottom. This is faster because any leftover orbs will already be at the bottom instead of having to fall down which takes another 0.1s
Each additional combo matched on the board is about 0.1s which is important to note as it directly takes out of your combo value

Combos

Combo score is your average combos to one decimal rounded down. This is multiplied by 5000 to give the combo score as mentioned earlier. Average combos is total combos in the dungeon divided by number of turns taken. This is very important to note because not everything takes a turn.
1) Refreshing the board and skyfalling combos does not take a turn. That means that yes, your refreshed combos add on to the average without taking a turn.
2) Combo addition does not add to the combo score so you can forget using your Planar for ranking.
With these things in mind, let’s talk about how the skill-only defeat is actually calculated.

Skill-only defeat

Any time you clear a floor without moving orbs (i.e: taking a turn). That means refreshing counts as a skill only defeat and will give you -5000 while poison means you have to pick up the orb and match which does indeed take a turn, therefore it is not a skill only defeat.

Max damage

As mentioned, this is your damage dealt multiplied by 0.00025, or a friendlier way to look at it, 2500 points for every 10,000,000 damage up to 10,000 points max. This means the closer you can get to 40,000,000 the higher your score here. Ideally you would hit that if the ranking allows for it. Now, what is max damage? This is taken as the total damage dealt to a single enemy. This is an important distinction because you might have floors with 5 enemies and you hit 8,000,000 each but your damage for that floor would only be 8,000,000. This often leads to people bringing a card that is stacked with killers against something in the dungeon so they can reach the damage cap with ease. For example, in Xiu Min ranking people were able to hit the damage cap for ranking with no leader skills. That’s quite the task considering your entire multiplier is based off spikes, combos and awakenings. When building a team the damage cap should be a consideration as you can find yourself several thousand points away from a crown despite having a better time and combo average.

How does all of this work together?

The first relationship I’ll discuss is the value of combos and dungeon length. I should be more technical and instead of calling it length we’ll call it the number of turns. Since it’s average combos and not total your score per combo is dependent on how many turns you take. This is easily calculated as (5000/turns). If you only have to take two turns each combo is worth 2500 points. If you have to take 14 then each combo is worth 357 points. This relationship is the most important to understand as it provides the basis for a lot of the tradeoffs.
Next let’s look at value of combos and skill only defeat. A skill only defeat does not take a turn and does not add combos (except refreshing) so you can consider it like a floor skip which reduces your turns taken. Tying in the points per combo we can see that buttoning a floor provides a change in combo value as
5000 * y/[x(x-y)]
Or written differently as
5000 * [(1/(x-y))-(1/x)]
Where x is the total number of floors and y is the number of floors you skill-only defeat.
This is important because it shows us that as y approaches x, the value of buttoning a floor in terms of points per combo increases. From that point of view we get the conclusion that when choosing to button floors, button more versus less if possible.
Now, we still haven’t factored in time. The second most important relationship is time to combos. First, given our 5000/x relationship for combo points we can directly relate combos to time. In a 5 floor dungeon, each combo is worth 1000 points (5000/5) and we know that each second is worth 500 points regardless of the dungeon. Therefore, in this example, one combo is worth two seconds. We know now that it is always more valuable to make a combo than save less than 2 seconds. When we get to a 10 floor ranking, a combo is only worth one second. Any combo you make needs to take you less than 1s each. I will include a table for quick reference at the bottom of the page up to 14 floors.
N.B.: I did not include the time that it takes to resolve the combo into consideration here. The game still has to count the combos and complete hte attack animation which takes fractions of a second per combo. Cascades also take time to resolve. Therefore, give yourself a buffer on the results. Check the section on time to see the rough time estimates to take into consideration.
Now we’ve compared combos to time, combos to skill-only defeats. We can build a basic decisioning model on combo or no-combo based on these relationships. For all intents and purposes, the assumption is that damage score and awakenings score stay constant, otherwise for specific cases the player needs to take it into consideration.
If 5000*y < 5000 * [(1/(x-y))-(1/x)] * z + (t*500)
Where 0<y<x, z is the expected number of combos, t is the expected total time saved from skill only defeats.
t is where you will need to either estimate or test run to get a better unstanding as this is a very personal number. I know people that can regularly full combo a board in under 4s and others that take 9s. Along with the added time to resolve the board your time may very greatly from mine. My general rule of thumb is that it will take roughly 6 to 7s or most players to full combo a natural board and another 2-3s to count. Along with more pings than a skill-only, it will take more than 10s total. Meanwhile, comboing a bicolor board all factors included should be under 10s.
Considerations to be taken for the value of z is that when comboing less floors on a Skyfall ranking, you could adjust the combo score higher to reflect that a good run is more likely to happen for the shorter dungeon than a good run for a longer dungeon. In other words, you need less good RNG per floor and therefore in sets of runs you can expect a higher combo when matching less turns. Getting 4 skyfall per floor on 3 floors is more realistic than 4 skyfall per floor on 10 floors.
From here you have a decent system of looking at whether, in a combo scenario, you will be buttoning floors and how many. Usually if you’re going to button it’s always more beneficial to button more floors when carrying high combo averages but when going for speed, saving the time to 1c is not worth the 5000 point cost.
The final consideration you need to look at after all of this is whether you’re going to low combo through or high combo through. We can look at the comparison as:
(t*(x-y)*500) + (y*5000) – ((c-1)*5000) < 0
where t is the time per floor when comboing fully (as I mentioned this is very personal, should do your own testing in endless)
x is the number of floors in the dungeon
y is the number of floors you button
c is the average combos you’re aiming for on a combo run
If this statement is less than 0 then you’re better off going the combo route, otherwise try going the time route. Remember, when looking at your time you need to consider the time to resolve the combo (count / pings as well).

Conclusion

This article was mostly designed to give some insight into the ranking point system and why you might see some rankings where people go for strictly buttons, or strictly speed, etc. It also doesn’t include the relationships between the point topics covered and the awakening point system since that is heavily ranking dependent and what point scores they assign. Good luck in future rankings and I hope this provides some insight.