Forums-->Queries and help-->
| Author | Set Traps |
|---|
After playing with Barbs of the Past, I've been spoiled by 100% movement reducing probability.
How is the probability for Goblin Trapper's Set Traps ability calculated? Does it depend on the number of tiles the enemy unit has left in its path when it steps on the trap? | | Base 25% chance if goblin HP is equal to walker HP, + or - minus HP ratio. Caps at 5% and 75%. | The trap is at least dependent on the current trapper HP at the time of the trap getting set off.
In any hunts with disproportionately large trapper stacks, the target will almost always get completely stopped with a few failing cases here and there where the trap belonged to a worn-down or perished trapper stack.
E.g. https://www.lordswm.com/war.php?lt=-1&warid=785214226
So it at least doesn't share the same 75% caps compared to triggering abilities.
Also based on the rendering code (no formula unfortunately), there are only 6 cases:
1. Failing (trapper loses 1 creature-turn worth of initiative)
2-5. Reduce walking distance by 1-4 tiles
6. Complete stop
(search for 'snu' here: https://dcdn.lordswm.com/js/usemagicp.js)
So the formula should also distinguish or explain these cases as well. | | Well, I once asked on .ru why my goblin traps were so successful even with tiny stacks and I was tought about the 5% cap. They didn't mention 75% though, that was me jumping to conlusions. | I am not ruling it out entirely, and it's also a very sensible starting point.
Just that the standard triggering formula alone doesn't fully explain the trapping mechanics and there needs to be other components in play to account for the varying modes of success.
A bit like how dwarf bears also obeys the same rules with [0.05, 0.75] bounds, but the actual success rate can still reach ~90% after walking 7 tiles given 28% base trigger rate from equal HP.
There is one theory in .ru forum based on the original HoMM series:
https://www.heroeswm.ru/forum_messages.php?tid=2713371
https://www.heroeswm.ru/forum_messages.php?tid=1681485&page=2081#47501706
where the probabilities are split into 4 categories:
fail, -1 tile, -2 tiles, and complete stop
and each category cannot be lower than 5%.
But it is unclear how the mechanics translates to the lwm case.
The method also gives at least 15% chance of successful trigger in the worst case (5% for -1, -2, and complete stop each) which is probably too high (e.g. for 1 goblin). | The Paw knock of bears is slightly different.
It is capped by the same rule [0.05,0.75] but since its "This ability attempts to trigger as many times as many tiles were covered before the attack; thus, if this creature attacks from place on any turn, the chance of this ability to trigger is 0."
Even if I did see it trigger on retalition rarely.
That means that one hit with 7 movement is the ability applied 7 times.
So even with 0.5 chance, the componded chance of not being knoked is 0.78%.
For a 25% chance (equal hp), that is ~87% of knockback after 7 tiles.
Also hp is calculated after the damage of the attack is applied and before the retaliation, which combined with the high number used does help a lot.
Even with minimum percentage (like in hunts), after a 7 momvement, you still have ~30% of a know back with the base chance of 5%.
It would be interesting to know if the trap is calculated the same way, apply the trap to each of the remaining movement attempted and reduce movement for any success. It might explain why the AI always put the trap as near as possible to the starting point.
If you need to move 7 points after encountering the trap that would greatly increase the chance of the trap workingm because it would be applied 7 times. The chance of stopping you completely would not be great unless hp is much higher (lie 15~20 times higher), but the chance of reducing movement by at least one would be high even with just one goblin, making the trap work and stopping you just short of the enemy. It does depend on how much the unit can move, so it would be less effective against slow units. |
This topic is long since last update and considered obsolete for further discussions.
Back to topics list
|
