DP Research Thread #4 ("Newer still")

Status
Not open for further replies.
The Metronome item doesn't increase Rollout's power each time it executes. It only increases it each time it's selected in succession.

Here's the data. Golem holding a Metronome using Rollout against a Marowak:

13
24
47
99
?? (Marowak faints)

?? (Pidgey faints while I revive Marowak)
57 (critical hit)
58
105
?? (I swap in a Staraptor to take the hit)

16
32
66
127
?? (Marowak can't survive the hit)
 
Lorak said:
Wow. That's depressing.
Click to expand...
I agree. I never thought to test that, and had been using Metronome on my gimmick rollout Lickilicki IRL all this time...

Completely off-topic, first post on Smogon in quite some time. Perhaps even since the new year.
 
The damage formula doesn't make it clear, but in the Defense modifiers, if Clamperl was Skill Swapped Color Change and its type changed to Rock, which would apply first to its Special Defense: Sandstorm (1.5x) or Deepseascale (2x)? Do we even know?

EDIT: Shoot, I didn't see on the OP that it hadn't been tested. ><
 
How does the item Metronome work with moves that call other moves?

If I use Return, then am hit with Spore, then Sleep Talk Return, then Sleep Talk Earthquake, at which point do I get a boost, if anywhere?
 
CardsOfTheHeart said:
The damage formula doesn't make it clear, but in the Defense modifiers, if Clamperl was Skill Swapped Color Change and its type changed to Rock, which would apply first to its Special Defense: Sandstorm (1.5x) or Deepseascale (2x)? Do we even know?

EDIT: Shoot, I didn't see on the OP that it hadn't been tested. ><
Click to expand...
We need that to be tested.
 
(probably not finding an answer to the following question in the Simple Question thread)

Anyone know where I can find a list of all of the modifiers to the Speed stat and the order in which they are applied? (My Gyro Ball formula depends on this.) If there isn't such a list available, could someone please make one--if it's possible?

I'm not looking for move priority; don't be the second to make that mistake.
 
Pokemon A (with Baton Pass) vs. Pokemon B (slower than Pokemon A and with Encore)

If Pokemon A uses Baton Pass to Pokemon C (with Baton Pass) and then Pokemon B uses Encore on the same turn, is Pokemon C forced to use Baton Pass?

And is it the same situation with U-turn?
 
what happens to a hyper cutter pokemon when exposed to memento does it memento fail or work but only reduces special attack or work as normal dispite hyper cutter
 
[.darkie] said:
Pokemon A (with Baton Pass) vs. Pokemon B (slower than Pokemon A and with Encore)

If Pokemon A uses Baton Pass to Pokemon C (with Baton Pass) and then Pokemon B uses Encore on the same turn, is Pokemon C forced to use Baton Pass?
Click to expand...

Encore fails. I don't know about U-turn.
 
Obi said:
Encore fails. I don't know about U-turn.
Click to expand...

No, Encore works and fixes Baton Pass or U-Turn. The encore itself, however, is not Baton Pass-able. So once Pokémon C uses Baton Pass, the next incoming Pokémon will not automatically be encored, even if it too knows Baton Pass. You could, of course, use Encore on this Pokémon too, continuing the chain.
 
I have a question involving Hyper Cutter as well.
Does a move that lowers the user's attack (say, Superpower) lower a Hyper Cutter Pokémon's Attack? (Use Pinsir/Corphish/Crawdaunt)
 
CPU V2.0 said:
I have a question involving Hyper Cutter as well.
Does a move that lowers the user's attack (say, Superpower) lower a Hyper Cutter Pokémon's Attack? (Use Pinsir/Corphish/Crawdaunt)
Click to expand...
It blocks enemy attempts to lower your atk, self inflicted stat drops still apply.
 
Common wisdom suggests that the success rate of Protect halves with each successive use, meaning that the first use always works, the second use has a 50% success rate, the third use has a 25% success rate, and so on.

I recently did 31 runs of Protect and here were my results (in number of consecutive successes):

1, 1, 2, 1, 3, 1, 1, 1, 3, 2, 3, 2, 4, 1, 1, 1, 3, 1, 2, 1, 2, 2, 2, 1, 4, 1, 3, 1, 2, 1, 2

Or, organized better:
1 use: 15
2 uses: 9
3 uses: 5
4 uses: 2

Now according to the above formula, each run has a 1 * 0.5 * 0.25 * 0.125 = 0.015625 = 1.56% chance of reaching at least four successful consecutive uses, and yet as you see I obtained 2 runs of four in my run of 31 trials. The odds of getting 2 or more runs of at least four uses like I did in 31 trials is equal to 1 – ((1 – 0.0152625)^31 + 31(0.0152625)(1 –*0.0152625)^30) = 0.0842748465 = 8.43%.

Doing a similar analysis on runs of at least 3 successful turns gives us a 0.0840156531 = 8.40% chance to get at least 7 runs of at least three uses in a run of 31 trials. These two statistics are not completely independent, but I think you see where I'm going here.

While not certain by any means, it's looking very likely from the distribution I got that the probability for consecutive uses of Protect is 50% for all but the first use.

Is somebody else willing to test and confirm this?

EDIT: Detect gives similar results, although I performed fewer trials with it.

EDIT: There's been some dissenting opinion in previous research threads about whether or not Sky Attack has a high crit rate. I did a test where I used Sky Attack 32 times on a Luxray with a Noctowl that was holding a Razor Claw and had Focus Energy Baton Passed to it. If Sky Attack has a normal crit chance, it should have only become a critical hit about once every three times under these conditions. If it has a high crit chance, it should have become a critical hit about once every two times. Here are my results:

Crits: 16
Not crits: 13
Misses: 3

I hope we can all put this to rest now, especially since the in-game description of Sky Attack reads, "A second-turn attack move with a high citical-hit ratio. It may also make the target flinch."
 
TheMaskedNitpicker said:
EDIT: There's been some dissenting opinion in previous research threads about whether or not Sky Attack has a high crit rate. I did a test where I used Sky Attack 32 times on a Luxray with a Noctowl that was holding a Razor Claw and had Focus Energy Baton Passed to it. If Sky Attack has a normal crit chance, it should have only become a critical hit about once every three times under these conditions. If it has a high crit chance, it should have become a critical hit about once every two times. Here are my results:

Crits: 16
Not crits: 13
Misses: 3
Click to expand...

First some ADV testing:

Peterko said:
SKY ATTACK

-tested this with lv.100 moltres against those route 1 (101?) lv. 2-3 pkmn
Results: the exact 8 PP tries (hits/CHs): 7/1, 6/2, 8/0, 8/0, 5/1, 7/0, 6/0, 7/1, 6/0, 6/0, 7/0, 7/1, 6/0, 8/1, 7/0, 7/1 ->

128 sky attacks
108 hits = 84,375% accuracy
8 CHs = 7,4% (8/108 ), but exactly 6,25% (8/128 ) out of all sky attacks, which is 1/16, the basic CH chance in the game

-> sky attack does not have a high CH chance (but it should have a 30% flinch chance)
Click to expand...


Then the follow-up testing in DP

Peterko said:
Lv. 35 noctowl @ wide lens using sky attack against partner Lv.35 wobbuffet

Sky Attack ingame description: "A second-turn attack move with a high critical-hit ratio. It may also make the target flinch."
Sky Attack 1st turn message: "Noctowl became cloaked in a harsh light!"

2/8 flinches, 0/8 CHs
2/8 flinches, 0/8 CHs
1/8 flinches, 2/8 CHs
2/8 flinches, 1/8 CHs
1/8 flinches, 0/8 CHs
1/6 flinches, 0/6 CHs, 2 missed
3/8 flinches, 0/8 CHs
2/6 flinches, 0/6 CHs

pachirisu struggle recoil fainted

TOTAL:
14/60 flinches = 23.33%
3/60 CHs = 5%

-> flinch rate seems to be 25% (we can tell that for sure after king´s rock has been tested)
-> although the description says so, sky attack doesn´t seem to have a high critical-hit ratio
Click to expand...
 
I have a question about Mist. Does it still make self-induced stat drops happen, or are they ignored with Mist? Eg. Draco Meteor's Sp Atk drop. Sorry if this is a simple question, but I'm not sure if it was looked into before.
 
I'm not too concerned with the Advance testing. I'm willing to believe that Sky Attack had a normal critical hit ratio in Advance. I just did another series of trials in D/P. This time I used a Noctowl with a Razor Claw, but without Focus Energy. I walked around Route 201, defeating Bidoof and Starly.

Assuming we can all accept as a given that Razor Claw raises the crit ratio by one level, the chance for a crit if Sky Attack has a normal crit ratio is 1/8 and the chance if it has a high crit ratio is 1/4.

Here are my results from 112 trials:

27 crits.
73 normal hits.
12 misses.

If Sky Attack has a normal crit ratio, the chance of getting at least 27 crits in 100 hits is well under 0.001 or 0.1%. The chance of Peterko getting three or fewer crits in 60 trials without Razor Claw is about 4.82%. I'm sure that his data is good, but he was doing research and tests on many, many things and it's not unusual for some flukes to show up. One out of every 20 tests is going to give a result outside a 95% confidence interval. I believe that this was one of those flukes.

I am now convinced that Sky Attack has a high crit ratio. If this data is not enough to convince you (Great Sage and Obi specifically), I'm going to need some specific info on what will convince you so that I, you, or some neutral thrid party (preferably all of the above) can run the appropriate tests.

EDIT:
Bologo said:
I have a question about Mist. Does it still make self-induced stat drops happen, or are they ignored with Mist? Eg. Draco Meteor's Sp Atk drop. Sorry if this is a simple question, but I'm not sure if it was looked into before.
Click to expand...

If I remember my previous testing, Mist doesn't prevent self-induced stat drops. I'll run a test now.

UPDATE: Confirmed. Mist doesn't prevent self-inflicted stat drops. (Tested with Overheat, if you're curious)
 
TheMaskedNitpicker said:
Well, normally you'd ask a question like this in the Ask a Simple Question Thread before you post here. Before that, though, you should check the description of the move in question.

To answer your question, yes Destiny Bond can KO a Pokémon behind a Substitute.
Click to expand...

Normally I do visit the Simple Question thread but I thought this was a bit more specific. Anyway, I didn't realize the Smogon move database was up-to-date and that specific. Thanks.
 
TheMaskedNitpicker said:
Common wisdom suggests that the success rate of Protect halves with each successive use, meaning that the first use always works, the second use has a 50% success rate, the third use has a 25% success rate, and so on.

I recently did 31 runs of Protect and here were my results (in number of consecutive successes):

1, 1, 2, 1, 3, 1, 1, 1, 3, 2, 3, 2, 4, 1, 1, 1, 3, 1, 2, 1, 2, 2, 2, 1, 4, 1, 3, 1, 2, 1, 2

Or, organized better:
1 use: 15
2 uses: 9
3 uses: 5
4 uses: 2

Now according to the above formula, each run has a 1 * 0.5 * 0.25 * 0.125 = 0.015625 = 1.56% chance of reaching at least four successful consecutive uses, and yet as you see I obtained 2 runs of four in my run of 31 trials. The odds of getting 2 or more runs of at least four uses like I did in 31 trials is equal to 1 – ((1 – 0.0152625)^31 + 31(0.0152625)(1 –*0.0152625)^30) = 0.0842748465 = 8.43%.

Doing a similar analysis on runs of at least 3 successful turns gives us a 0.0840156531 = 8.40% chance to get at least 7 runs of at least three uses in a run of 31 trials. These two statistics are not completely independent, but I think you see where I'm going here.

While not certain by any means, it's looking very likely from the distribution I got that the probability for consecutive uses of Protect is 50% for all but the first use.

Is somebody else willing to test and confirm this?
Click to expand...
If you try 31 runs of Protect, you should have:

31/2 = 15.5 of them that have a run of 1
31/4 = 7.75 of them that have a run of 2
31/8 = 3.875 of them that have a run of 3
31/16 = 1.9375 of them that have a run of 4

Since your values are very close to the above values (even though there were only 31 attempts), I think that Protect does work as we think it does.
 
X-Act, your method gives the likelihood of success assuming Protect has a 50% chance of working after the first. The previous thought was that it was 100%, 50%, 25%, 12.5%, etc. per turn, which would give (frequency distribution here)


1 = 100%
2 = 50%
3 = 12.5%
4 = 1.5625%

This means that given these values, we would expect, in a run of 31 Protects

31 of them would last at least one turn
15.5 of them would last at least two
3.875 of them would last at least three
.484 of them would last at least four

His data is

31
16
7
2



For one and two turns, 50% chance and half the previous chance give identical results and are thus irrelevant. After that, his data supports the flat 50% chance for success. We'd obviously need more data to be sure, but the odds of it not being a flat 50% shot seem pretty long.
 
Thanks to Obi for explaining that so well. The question now is if there's anyone else willing to do these tests in order to confirm my data. It won't take long and it would help us understand how Protect and Detect work.

I'm also requesting some input on Sky Attack. I'm convinced that it has a high crit rate, but I'm not going to change the move entry back until we reach a consensus. Does anybody dispute the data I've gathered? Obi, X-Act, and Great Sage, I'm looking at you here.
 
Status
Not open for further replies.
Back
Top