Add pity to card upgrades#7
Conversation
| `Upgrading the \`${card.mapper.username}\` card from from **${condition.id}** to **${condition.nextUpgrade.id}** has a ${Math.floor(condition.upgradeChance * 100)}% chance of success.`, | ||
| condition.previousUpgrade ? `If the upgrade fails, there is a 50% chance the card will be downgraded to **${condition.previousUpgrade.id}**` : `If the upgrade fails, nothing will happen.`, | ||
| `Upgrading the \`${card.mapper.username}\` card from from **${condition.id}** to **${condition.nextUpgrade.id}** has a ${Math.floor(upgradeChance * 100)}% chance of success.`, | ||
| condition.previousUpgrade ? `If the upgrade fails, there is a ${Math.floor(downgradeChance * 100)}% chance the card will be downgraded to **${condition.previousUpgrade.id}**` : `If the upgrade fails, nothing will happen.`, |
There was a problem hiding this comment.
I don't think the pity factor should be advertised anywhere
There was a problem hiding this comment.
it displays the odds and the odds change, felt like basic transparency
There was a problem hiding this comment.
I'd still prefer not to give it away, ppl should not really be aware of the pity system
There was a problem hiding this comment.
that's a fair standpoint, can remove when we've figured it all out
| const success = random < upgradeChance; | ||
| const downgrade = !success && Math.random() < downgradeChance; | ||
| if (success) { | ||
| currentCard.condition = condition.nextUpgrade; |
There was a problem hiding this comment.
pretty sure pity should be reset on success
There was a problem hiding this comment.
typical pity systems would do that, but I don't think it would fit here
the main killer in the upgrade system is how much it can fluctuate, in 10% of cases your card downgrades 3+ times by going from good to poor and back on repeat for example
so resetting on success would be a bit pointless since it'd continue to allow those sequences to escalate costs anyway
at that point just reducing overall costs or reducing chance to downgrade in general would be more effective changes, which people would probably also be happy with
| } | ||
|
|
||
| const upgradePrice = condition.upgradePrice * (card.foil ? 2 : 1); | ||
| const pityFactor = 0.07 * card.pity; |
There was a problem hiding this comment.
This should probably not be linear, rather something that slowly approaches but never reaches 1
There was a problem hiding this comment.
Can agree with making it non-linear, though making it "slowly" approach anything seems a bit pointless since most upgrade sequences currently don't even reach 10 total rolls
; this isn't like spending 100 gacha pulls to get your guaranteed SSR character
There was a problem hiding this comment.
How about this? It's pity / (pity + 8)
There was a problem hiding this comment.
Comparison with previous formula
There was a problem hiding this comment.
simulated this and got this, it going up steeper early has a stronger effect it seems
but having the average coincidentally hover around the price of a superdrop seems rather neat
| Costs | |
|---|---|
| Mean | 1501 |
| Median | 1300 |
| Std Dev | 713 |
2 participants
Implemented this as a way to propose its workings, I think the results are solid
basically applying an "what doesn't kill you makes you stronger" philosophy to limit extreme sequences of bad luck, without tampering with regular fail chains
tested through simulating upgrades from badly damaged to mint, the system has the following effects:
Aka it mainly aims to lower the standard deviation, as most of it is built up in the upper end due to Mean - Std Dev going far below the minimum cost to upgrade
Below are 2 graphs for comparison across the board
No Pity (current)
Pity (implemented in this PR)
