• Follow our Instagram!
  • Curious about what we're currently doing about bots? Check out the latest updates from staff in this thread.

Research Champions Battle Mechanics Research

Damage overflow to negative values is possible in Pokemon Champions via 32-bit signed integer overflow.

The setup for this was possible by assuming 32-bit signed integers were being used when storing numbers relevant to damage calculation. I show overflow setups with Metronome (the item) and a combination of Metronome + Fluffy. Among other things, this is pretty solid confirmation to me that 4096-scaled modifiers are still being used in Pokemon Champions.

Overflowing with Electric-type Stored Power​

Average LC Enjoyer messaged me about a setup they tried in Champions to deal maximum damage. I recommend reviewing the raw footage (one) (two) they sent me (in the video they talk about using Power Split but didn't actually pull it off in the 20-minute time limit).

Details and math
In this setup:
  • The attacker, Rotom-Heat:
    • Has a starting Special Attack stat of 172
    • Is fully boosted in all stats, resulting in a 860 BP Stored Power
    • Is affected by Electrify
    • Has active Charge
    • Is boosted by Helping Hand
    • Is boosted by Electric Terrain
    • Is boosted by Solar Power
    • Has a full-powered Metronome
  • The defender, Pelipper:
    • Has a starting Attack stat of 70
    • Is affected by Power Trick and Wonder Room (which effectively turns its Attack stat into its Sp. Def stat)
    • Is at -6 Special Defense
Here are the damage values I expect at various stages of the damage formula, assuming the SV damage formula but with 32-bit signed integers.
Final BP: 3354
Final Atk: 1032
Final Def: 17
Base Damage: 89589
After random damage rolls: [76150, 77046, 77942, 78838, 79734, 80630, 81525, 82421, 83317, 84213, 85109, 86005, 86901, 87797, 88693, 89589]
After STAB: [114225, 115569, 116913, 118257, 119601, 120945, 122287, 123631, 124975, 126319, 127663, 129007, 130351, 131695, 133039, 134383]
After type effectiveness: [456900, 462276, 467652, 473028, 478404, 483780, 489148, 494524, 499900, 505276, 510652, 516028, 521404, 526780, 532156, 537532]
After final modifiers: [-134776, -124024, -113272, -102520, -91768, -81016, -70280, -59528, -48776, -38024, -27272, -16520, -5768, 4984, 15736, 26488]

So assuming the math is correct, there was a 13/16 chance to deal negative damage, and a 3/16 chance to deal positive damage. It's very lucky that Average LC Enjoyer did not have an attack-lowering stat alignment on Pelipper, because if they did, it would have looped back around to positive numbers again and it would have never been noticed.

Attempting 16-bit signed overflow​

Since I wasn't sure initially what was going on, I tried some extremely basic setups to rule out 16-bit signed integer overflow. The first thing I tried was dealing over 32767, but less than 65536 damage. The second thing I tried was dealing over 65535 damage. Neither resulted in negative damage.

In the first setup, I used a Mega Floette with 227 SpA, and full 860 BP Stored Power into -6 101 Sp. Def Sneasler. Without overflow, I expected these possible damage rolls:
[46732, 47280, 47832, 48380, 48932, 49480, 50028, 50580, 51128, 51680, 52228, 52780, 53328, 53880, 54428, 54980]

I did the exact same thing in the second setup, but used Magic Powder on Floette to give Stored Power STAB. Without overflow, I expected these possible damage rolls:
[70096, 70920, 71748, 72568, 73396, 74220, 75040, 75868, 76692, 77520, 78340, 79168, 79992, 80820, 81640, 82468]

Attempting 32-bit signed overflow​

I figured at this point it was probably 32-bit signed integer overflow. I used two tests to confirm this, both exploiting 4096-scaled modifiers to do so. A full-powered Metronome is 8192/4096; Metronome and Fluffy combined are 16834/4096. So I just needed to hit 2^31 -1 / 8192 and 2^31 -1 / 16834 pre-Metronome, which is not unreasonable to set up.

Setup #1 - Power Trip
Raw footage of Power Trip setup
  • The attacker, Pangoro:
    • Has a starting Attack stat of 192
    • Has 32 accumulated boosts, resulting in a 660 BP Power Trip
    • Is at +6 Attack
    • Is boosted by Huge Power
    • Is boosted by Helping Hand
    • Has a full-powered Metronome
  • The defender, Pikachu:
    • Has a starting Defense stat of 54
    • Is at -6 Defense
    • Has been turned into a Psychic-type from Magic Powder, then given an additional Ghost-type from Trick-or-Treat
Here are the damage values I expect at various stages of the damage formula, assuming the SV damage formula but with 32-bit signed integers.
Final BP: 990
Final Atk: 1536
Final Def: 13
Base Damage: 51469
After random damage rolls: [43748, 44263, 44778, 45292, 45807, 46322, 46836, 47351, 47866, 48380, 48895, 49410, 49924, 50439, 50954, 51469]
After STAB: [65622, 66394, 67167, 67938, 68710, 69483, 70254, 71026, 71799, 72570, 73342, 74115, 74886, 75658, 76431, 77203]
After type effectiveness: [262488, 265576, 268668, 271752, 274840, 277932, 281016, 284104, 287196, 290280, 293368, 296460, 299544, 302632, 305724, 308812]
After final modifiers: [-523600, -517424, -511240, -505072, -498896, -492712, -486544, -480368, -474184, -468016, -461840, -455656, -449488, -443312, -437128, -430952]

My setup was designed to just barely reach over the 2^31 -1 boundary, and I consistently got negative values from damage.

Just to make sure, I then modified my setup to change Pangoro's Attack stat to 162. As predicted, there was no damage overflow this time.
Final BP: 990
Final Atk: 1296
Final Def: 13
Base Damage: 43427
After random damage rolls: [36912, 37347, 37781, 38215, 38650, 39084, 39518, 39952, 40387, 40821, 41255, 41689, 42124, 42558, 42992, 43427]
After STAB: [55368, 56020, 56671, 57322, 57975, 58626, 59277, 59928, 60580, 61231, 61882, 62533, 63186, 63837, 64488, 65140]
After type effectiveness: [221472, 224080, 226684, 229288, 231900, 234504, 237108, 239712, 242320, 244924, 247528, 250132, 252744, 255348, 257952, 260560]
After final modifiers: [442944, 448160, 453368, 458576, 463800, 469008, 474216, 479424, 484640, 489848, 495056, 500264, 505488, 510696, 515904, 521120]

Setup #2 - Pyro Ball + Fluffy
Raw footage of Pyro Ball setup (Gengar side) (Cinderace side)

In this setup:
  • The attacker, Cinderace
    • Has a starting Attack stat of 163
    • Is at +6 Attack
    • Is boosted by Huge Power
    • Is boosted by two Helping Hand boosts
    • Has a full-powered Metronome
  • The defender, Gengar
    • Has a starting Defense stat of 72
    • Is at -6 Defense
    • Has copied Bug/Steel type vs Reflect Type, and also been affected by Forest's Curse
    • Has Fluffy
  • Sun is active
Here are the damage values I expect at various stages of the damage formula, assuming the SV damage formula but with 32-bit signed integers.
Final BP: 270
Final Atk: 1304
Final Def: 18
Base Damage: 8608
Sun Damage: 12912
After random damage rolls: [10975, 11104, 11233, 11362, 11491, 11620, 11749, 11879, 12008, 12137, 12266, 12395, 12524, 12653, 12782, 12912]
After STAB: [16462, 16656, 16849, 17043, 17236, 17430, 17623, 17818, 18012, 18205, 18399, 18592, 18786, 18979, 19173, 19368]
After type effectiveness: [131696, 133248, 134792, 136344, 137888, 139440, 140984, 142544, 144096, 145640, 147192, 148736, 150288, 151832, 153384, 154944]
After final modifiers: [-521792, -515584, -509408, -503200, -497024, -490816, -484640, -478400, -472192, -466016, -459808, -453632, -447424, -441248, -435040, -428800]

The battle log's interpretation of damage overflow​

The battle log in Champions logs the actual damage you took from the perspective of the player that lost HP, and also logs the percentage of HP lost from the opponent's perspective. Here's examples of what that looks like:
QxYCN62OcGqDQsKa1jd6sg.png


ZqkGYNwMRRU-yNmcawlG8g.png


It is not obvious to me where these battle log numbers are coming from. They aren't consistent, so I assume it's got to be related to the specific damage roll in some way. I did hit one of the numbers in the Pikachu setup twice. For now, I'll just document them:
  • In the Pyro Ball Fluffy setup:
    • From Gengar's perspective: 64647 -> 167, 21255 -> 167, 30119 -> 167
    • From Cinderace's perspective: 299% -> 100%, 530% -> 100%, 755% -> 100%
  • In the Power Trip setup:
    • From Pikachu's perspective: 50246 -> 142, 62590 -> 142, 47494 -> 142 (NOTE: this was was a crit), 62590 -> 142, 15582 -> 142, 37886 -> 142

A note about theoretical max damage setups​

If implicit 4096/4096 modifier application is still present in Pokemon Champions as it is in the mainline games, then every attack will have to go through final modifier application. I did not create a setup to test if it is still there or not. Even if your setup precisely lands on 2147483647 during the initial multiplication of the final damage modifiers (which good luck doing exactly that), you always have to divide by 4096 afterwards. As is, I don't expect it to be possible to ever deal more than 2^31 -1 / 4096 damage, which would be 524,288.
 
i've finished research on the ranch movepools for m-b and m-c thanks to new methodology that allows me to confirm results a lot faster, and also added a link to a mega folder containing archived screenshots of all the m-b and m-c pulls i've observed just in case anyone else wants to do a more in-depth analysis than mine. some things to note:
  • nothing seems to have lost anything from their pools over the course of the last 2 regulations, except for archaludon, which lost mirror coat and metal burst because those moves were directly removed from its movepool
  • nothing appears to have gained anything either, including slash, which was newly introduced in m-c. i'm not sure whether this indicates that nothing in the ranch will ever gain anything new, but it seems to suggest that post-hoc modification of ranch pools will be rare if not entirely nonexistent. thank god
  • a handful of moves that weren't originally on anything in the ranch are present in the pools of new arrivals. these moves include healing wish, acupressure, noble roar, magnetic flux, and sweet scent (on mawile for some reason)
  • male indeedee has no flagship move, but female indeedee does (??????????????????)
 
DaddyBuzzwole said:
i've finished research on the ranch movepools for m-b and m-c thanks to new methodology that allows me to confirm results a lot faster, and also added a link to a mega folder containing archived screenshots of all the m-b and m-c pulls i've observed just in case anyone else wants to do a more in-depth analysis than mine. some things to note:
  • nothing seems to have lost anything from their pools over the course of the last 2 regulations, except for archaludon, which lost mirror coat and metal burst because those moves were directly removed from its movepool
  • nothing appears to have gained anything either, including slash, which was newly introduced in m-c. i'm not sure whether this indicates that nothing in the ranch will ever gain anything new, but it seems to suggest that post-hoc modification of ranch pools will be rare if not entirely nonexistent. thank god
  • a handful of moves that weren't originally on anything in the ranch are present in the pools of new arrivals. these moves include healing wish, acupressure, noble roar, magnetic flux, and sweet scent (on mawile for some reason)
  • male indeedee has no flagship move, but female indeedee does (??????????????????)

I heard Politoed lost Pound in reg M-C? Did it just never have it?
 
Last edited:
Ehmcee said:
I heard Politoed lost Pound in reg M-C? Did it just never have it?
my research is just on the moves the mons can have when you pull them in the ranch, not on their movepools in general. politoed could never spawn with pound. but yes, it did lose pound from its overall movepool and honestly i'm not sure why it had it in the first place (probably an off-by-one error somewhere, since pound's move index number is 1)
 
Last edited:
Someone suggested that: "Fire Mane may be a base power modifier, not Attack x1.5".

I'll copy the post here for better visibility so anyone interested can take a look:

Summary​

In the Champions mod, Fire Mane (Mega Pyroar) is inherited from the base data and implemented as a 1.5x Attack / Special Attack modifier. A community in-game measurement suggests that in Pokemon Champions it behaves as a base power modifier (or is applied at a different point), because the observed damage tops out at 100 while the attack-modifier implementation produces a maximum of 102.

I am reporting this so that someone with access to better data can verify it. The evidence is a single secondary source, so this may not be a bug.

Reproduction​

  • Format: [Gen 9 Champions] BSS Reg M-C
  • Attacker: Mega Pyroar (Pyroarite), Fire Mane, Hardy, 0 stat points (Attack 108)
  • Defender: Raichu-Alola, Surge Surfer, Hardy, 0 stat points (Defense 70)
  • Move: Fire Fang, Singles, no field effects, no critical hit
Damage per roll in the engine (forcing each of the 16 rolls): 85, 87, 88, 88, 90, 91, 91, 93, 94, 94, 96, 97, 97, 99, 100, 102

In-game observation​

Observed values: 84, 85, 87, 88, 90, 91, 93, 94, 96, 97, 99, 100 (Switch, Ver 1.1.0, measured 2026-06-21).

Source: https://note.com/yua_freewind/n/n21207ecbfe3d (Japanese). The author concludes that Fire Mane is a base power modifier. The source gives final stats only, so the natures and stat points above are back-calculated from Attack 108 and Defense 70.
 
Back
Top