Let's get this out of the way real quick: We don't really know what accounts for a player's power level. Most likely this is a combination of rank power levels (as described in https://splatoonwiki.org/wiki/Splatfest_Power) with an Elo-based score evolution (as seen in splatfest and leagues) plus another factor that I'll detail later which is more linked to the team-based nature of Splatoon.
The problem here has several faces to it. We need to explain the following elements we can observe daily:
- How one skips ranks even with cracks when one doesn't with no losses?
- How the gauge increases differently for apparently random reasons?
- How one gets cracks instantly and sometimes only after 3 games?
The starting point of my reasoning comes from what we could all see when splatoon was released or, more recently, when clam blitz came out. The basic principle is as follow: the game will make you skip ranks if it finds out your power level is too high compared to the rank you are in.
That simple principle explain a lot of what we can see:
- Many players will probably skip C ranks, much less B ranks and then a few A ranks.
- Once in S+, skipping rank is even rarer, as the gap between power levels tends to shorten but it's still possible as some other factor comes into play.
Now, this raises an interesting question: How the game is able to assert your power level with absolutely no references at all? Like when you start the game or a new game mode. As with any Elo rating system, your score rise based on your wins and losses and the score of your opponents. But at first everybody starts at the same point, and as your wins raise your score you should gain less and less point for each win.
This is where this other factor comes in: a K/D ratio bonus relative to the lobby.
In any ranks your power level gets an additional bonus/malus based on how you perform in game compared to the other players. If you're destroying your opponents while not dying much you get a power level increase that can outweight the Elo bonus/malus, and in the case you do poorly, the malus can also negate the Elo bonus for a win. This is always relative to the lobby as, for instance if all the players have 10-20 kills for 5-6 deaths, your 20/4 is less impressive than the same 20/4 when every other players have 2/10.
In the lower ranks, a lot of people are able to achieve this, which is why many people skip from C to B. Then, as you climb up the ranks it's much harder for most people to maintain that high K/D ratio, and they get less and less power level bonus that way.
And this explains why in S+ some players will skip 10 ranks even with cracks and some won't despite a full win streak in their previous rank:
- A continuous win streak while not being the top player will lower the power level bonuses and once you reach a new rank your power level is still in the expected range -> no skip.
- A somewhat chaotic session with many losses, even cracks but where you were constantly tearing up your opponents will results in a higher power level that the expected range -> skip
That also explains why slayer type gameplay is much more likely to get you to S+50 by skipping ranks.
Some believes the game expect you to win or lose games and fill the gauge accordingly (a lot if you were supposed to lose, not so much otherwise). I don't think it works that way as predicting stuff is way too complex and nintendo have showed us they aren't even able to balance weapons comps properly so foresight is clearly out of their league. I think it mostly works on the same principle as I describe above, if your power level bonus is much bigger relatively to the other players you've faced, you are more likely to see a big increase of your gauge versus the other players as the game needs you to rank up faster.
As for the cracks many have elaborate a theory with a hidden gauge that fills with each loss and that would trigger a crack when filled, and I also think it's probably something along these lines.
The amount of that gauge filled with a loss can vary depending on the rank and the lobby.
This is the tricky part imo. Because we know that the purpose of the ranking system is to place people were they belong I believe the gauge will fill faster if you have a power level that is somehow lower than the expected range for your rank and it will fill slower if you have a high power level despite losing versus lower level players.
Also, with that logic, we can explain various patterns, such as the fact that the first loss after a rank up rarely gives a crack (the gauge have been cleared, you really need to have a low power and have been carried all the way to your new rank to get that effect), and that several cracks can occurs consecutively (example: let's say that the gauge goes from 0 to 10, and I'm at 5 currently, one loss gives me a +7, I hit the 10 mark and get a crack, but there's a rest, and this rest is added to the now empty gauge, meaning I have now 2, another loss of 8 will give me another crack).
Skipping rank isn't only defined by wins/losses in a rank up, your K/D ratio probably play a bigger role in raising your power level and making you skip ranks.