Tool-assisted game movies
When human skills are just not enough

Submission #5478: McHazard's Arcade Arm Wrestling in 08:06.47

Console: Arcade
Game name: Arm Wrestling
Game version: unknown
ROM filename: armwrest.zip
Emulator: MAME-rr 0.139 v0.1-beta
Movie length: 08:06.47
FrameCount: 29188
Re-record count: (unknown)
Author's real name:
Author's nickname: McHazard
Submitter: McHazard
Submitted at: 2017-04-21 04:01:05
Text last edited at: 2017-05-08 13:56:42
Text last edited by: feos
Download: Download (131012 bytes)
Status: decision: rejected
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:
Quoting £e Nécroyeur:
Two independent parties were simultaneously creating a TAS...for this game? I love it! :)


  • Primary goal: Fastest in-game time
  • Secondary goal: Highest score
  • Tertiary goal: Fastest real-time
  • Sync setting: Movie starts unpaused
  • Manipulates luck
  • Has less rules than the WVBA

Why is there another TAS?

At the time that £e Nécroyeur submitted his TAS, I had completed this one, but was having trouble writing the submission text. Upon viewing the run, I found that we'd had different goal priorities; £e Nécroyeur had real time then score, where I had game time, score, real time.

It's my belief that for this game, real time doesn't make sense as a goal, primarily because of the post-fight score tally. The bonus points at the end are based on the time remaining, creating a Sonic-type situation where finishing a fight slower is faster overall.

£e Nécroyeur's submission explains things well (apart from a few off-by-one errors), so I'll instead be focusing on the differences between the runs here.

DIP Switch Settings

As in the other TAS, the difficulty is set to hardest. However, the coinage has been left at its default setting of 1 credit/coin. As the other submission notes, manipulation potential for Texas Mac 1's "Ready" timer is very limited without losing frames. But there is one type of input that is read between entering the name and starting the ready timer: the coin slot, which is ignored if free play is enabled.

...Unfortunately it turns out it was actually slower. <_<

Rematches are also lowered to 3, not that it matters.

Loop 1

Texas Mac (2.12)

  • Identical to £e Nécroyeur's submission.

Kabuki (5.28)

  • Identical to £e Nécroyeur's submission. Both this fight and the later rounds gave me a lot of trouble trying to find the fastest strat, and everything else I tried failed.
  • Annoyingly, Kabuki ALSO can give power on unblocked attacks on some arm positions, just like Texas Mac, but they're too far out of the way. One of them requires taking damage, which ultimately takes too long. The other is right before Kabuki is pinned, but is completely useless because he's programmed to always counterattack at that position.
  • High score is the secondary goal, and so the moneybag is caught in this and all other rounds.
  • Any resemblance to other fighters is entirely coincidental.

Mask X (8.03)

  • Identical to £e Nécroyeur's submission.

Alice & Ape III (2.60)

  • Identical to £e Nécroyeur's submission.

Frank Jr. (7.25)

  • This is the first place where the strats differ. Since in-game time is the primary goal for this TAS, the best thing to do in this and all other rounds is to manipulate a WAIT attack. The fire breath takes less in-game time to recover from, but is only available when Frank's arm reaches its lowest position.

Loops 2-4

Texas Mac (1.33,1.60,1.60)

  • Next.

Mask X (12.20,12.24,12.24)

  • The extra frame in loops 3 and 4 comes from Mask X's headbutt, which has a tighter dodge window.
  • Also, identical to £e Nécroyeur's submission.

Alice & Ape III (11.51,11.51,4.31)

  • The only fight after loop 1 that actually has a faster time in lower rounds. The culprit here was, to my surprise, this game actually has AI changes at certain times in the fight. In this case, the damage done by a double power attack in the starting position decreases from 64 to 48 once the clock hits 3 seconds. In loops 2 and 3, even tool-assisted button mashing isn't enough to beat this time limit, but in loop 4 the clock speed actually decreases, allowing an immediate win.
  • Again, in-game time changes the strategy here. On the first set of exclamation marks I stop short of getting all 10 and instead do a single power attack. At the resulting arm level, Ape III is guaranteed to counterattack instead of being guaranteed to just sit there and do nothing, giving the fight an extra clock stop phase.

Frank Jr. (6.40, 8.00, 8.85)

  • In addition to the reliance on WAIT attack clock stops the start of the fight is different. An immediate first press is of minimal help because it gives so few exclamation marks. The resulting choice between waiting for Frank to recover or pressing immediately and taking damage costs too much time. Instead I wait until the first press gives no exclamation marks, allowing an immediate second press.

Kabuki (15.10,15.10,15.10)

  • The end of input is delayed to gain an additional 500 points before Kabuki's arm is pinned. Otherwise these fights are identical to £e Nécroyeur's submission.
  • In the final rematch the window to counter Kabuki's final press is only 1 frame.

Total winnings: $923500

feos: This run needs my decision too...

feos: This submission is quite tricky. Here's a list of problems to resolve, as usual.

Tier. Looking at the percentage alone (77% support), one might think this movie was well received. Votes are somewhat minimal though, just 9. Now let's count the votes in actual posts, where people use to provide their take in a reliable way. 2 "yes" votes and 1 "meh". Doesn't look like people really got excited by this run. Lack of feedback is also a proof that it's not that entertaining to the audience.

And the gameplay kinda proves that: even though it has insane frame windows that can only be handles on a TAS level, and crazy point calculation system that can also only be routed precisely in a TAS, this boils down to optimization problems. And the impression from this gameplay is that it looks simplistic: you just have to pin the opponent's hand, that's all. Some comic take is there, but it doesn't change how narrow the range of actions is. And this game also loops the opponents for a few times, so once you see the first loop, you've basically seen all the gameplay has to offer entertainment-wise. Vault.

Maximum score. Some suggest that this might be considered a full completion type of goal in this game, since you can not collect an infinite amount of points. Indeed you can't, the levels have a time limit. But even though the count can not be infinite, or capped in a way similar to 999,999, where it either stops in some games or overflows in others, there's no designed limit where you could say "I collected all the points this game has to offer".

There's no fullness here, no count that would be equal to 100%. The score that gets calculated in such absurdly esoteric ways that it makes it impossible to prove that the score achieved in a given run is indeed maximal. The possibilities to test are endless, and even though it would require an overkilling amount of resources to do an exhaustive search, it would still be only "maxing out the score", not full completion. And therefore, it is an unvaultable goal.

In-game time. The argument given pro this goal is that in this game we deal with "a Sonic-type situation where finishing a fight slower is faster overall", and that it's "stomping opponents flat faster" than the other run. Both are false.

Sonic-type situation means that the player has to reach the end of the level as fast as possible, regardless of how many points the game will give, and ignoring the time the score tally will take. That allows the players to have a fair optimization competition, because the in-game time reflects the completion speed perfectly, and the real time fails at that due to the score tally. The level is still being completed faster and faster.

In this game, as the author admitted, ignoring the score tally doesn't give as much benefit, just a few frames, that can be easily lost if you wait too much. But what's more important, all the opponents that are different in this run are in fact defeated slower than in the run this one competes with.

IGT Frames
First Frank Jr. fight:
£e Nécroyeur 10:80 1142
McHazard 7:25 1282
Second Ape 3 fight
£e Nécroyeur 19:96 773
McHazard 11:51 778
Second Frank Jr. fight
£e Nécroyeur 15:16 933
McHazard 6:40 1198
Third Ape 3 fight
£ Nécroyeur 19:96 773
McHazard 11:51 778
Third Frank Jr. fight
£e Nécroyeur 18:95 933
McHazard 8:00 1198
Fourth Frank Jr. fight
£e Nécroyeur 20:96 933
McHazard 8:85 1206

The in-game time is only shorter because the author explicitly manipulates it to be shorter: he makes the opponent do the attacks that take more real time but freeze in in-game timer. It's not the score tally that makes this run longer, but the completion of the rounds that is slower.

Rejecting, with no foreseeable way to make a full completion run for this game.

Similar submissions (by title and categories where applicable):