Mike Tyson's Punch-Out!! (ntscus) (nes) [Any %] [Single Segment] [0:15:12.14]
Run Information
Completion Date June 30, 2016
State Published
Runner 'Summoningsalt'
Internal Comments:
Comment State Information Checkpoint None

This game is timed by adding up the in-game time you get on each fighter. I did this while I performed the run, so I know the exact time for this run is 15:12.14

Verification Thread
Run Comments

As of July 2015, the record for this game was 16:24 by sinister1.  Now, in July 2016, just a year later, the time has been pushed all the way down to 15:12.  This is partially thanks to a number of new strats being developed by several different MTPO runners.  I'll go over those when I recap the run.

This game is incredibly frustrating to speedrun.  On top of extremely precise timings and often difficult inputs, the luck required to get a run like this is absolutely insane.  And, of course, at the end of it all is Mike Tyson, where if you don't land at least 20 frame perfect punches on top of getting a good pattern, you're out of contention to get a world record.  The luck that I got in this run was the best luck anybody had ever gotten in an MTPO run, and although the execution wasn't perfect, it was good enough to acheive my goal of sub 15:20.  Now then, on to the fights:

Glass Joe (42.00): Thanks to a strategy first thought of by ouijawii and later simplified by Chambers, 42.00, which incidentally ties both the IL record and the TAS record, is a guaranteed time.
Von Kaiser (36.25): This strategy is a modified version of the IL strategy, which zallard1 figured out.  I got both stars in phase 2, which guarantees a time under 37 seconds.  Best possible luck on this guy.

Piston Honda 1 (49.97): Pretty much the only fight with bad luck in the entire run.  He blocked my 2nd face punch in the first phase of the fight, which ended up costing me about 3 seconds.  Although not necessarily a bad fight, most future runs will gain time against this split.

Don Flamenco 1 (15.00): Couldn't quite match the TAS time, but thanks to zallard1's buffer strat, a time of 15.00 or better is guaranteed.

King Hippo (40.99): About 70% of runs don't make it past this guy, because in order for me to get past him, King Hippo needs to open his mouth a lot.  He has a 37.5% chance of opening his mouth on any given punch.  He opened 3 out of the first 4 opportunities in this fight, which happens extremely rarely.  Only crazy RNG can gain you time in this split, but probably about 85% of the time, you would be losing time.

Great Tiger (48.25): Pretty typical fight here.  This is the last fight without any RNG, and ironically, it comes right after a fight with one of the most RNG. Getting a .25 means that my execution was reasonably tight.

Bald Bull 1 (1:02.25): I had figured out a few strats for Bull 1 if you didn't get the lucky star in phase 2, but I did get it, along with a bunch of other stupid pieces of luck.  I got the lucky star (50%), rolling jabs in phase 2 (25%), the good refill in phase 3 (25%), and rolling jabs again in phase 3 (25%). That works out to 1/128, or a 0.78% chance of getting this luck.  Absolutely unreal.
Piston Honda 2 (45.97): A few months back, zallard1 found a new phase 1 strategy, which allowed you to get a 27 second time going into the second phase.  He then messed with the idea of using max damage stars in phase 2, but I wasn't able to land them consistently enough.  Then, I realized that if you buffer a left quick dodge, hold start and up, and then let go of up once his guard raises (4 frame window), it lets max damage star punches land.  Doing this was much easier than tying to time the stars manually.  It ended up being a huge discovery, as it ended up letting me save up to 10 seconds.  I ended up getting the good refill in phase 2 as well, which was yet again the best possible luck.

Soda Popinski (51.25): This guy and King Hippo are typically regarded as the two biggest run killers, because they are the most random.  In each phase, Soda can either throw uppercuts (good), or throw hooks (bad), with a delay somewhere between 0 and 4 in-game seconds right before them.  Not only did I get 3 uppercuts in a row, which is quite rare itself, but I got extremely small delays before each one.  The odds in getting a fight this good is probably about 1/30-1/40, so, once more, this was another fight with incredible luck.

Bald Bull 2 (1:30.25): A new strat I figrured ot for this guy was eliminating the chance of him blocking without retaliating in phase 1.  Unfortunately, he still has 3 chances to do it later, and he probably has about a 1/4 chance of doing on each of those punches.  He did it to me once, which is technically considered bad luck, but we only lost about a second because of it.

Don Flamenco 2 (1:26.99): Previously, this fight would typically yield times in the 1:40-1:50 range, or if you were really lucky, you could maybe squeeze out somewhere in the upper 1:30's.  Although these times are still very possible now, and you still will get a time in the 1:40's aroud 80% of the time, there now exist strategies where you can get times in the lower 1:30's, or if you're really lucky, even into the 1:20's.

Zallard figured out an opening for phase 1 where you have 9 chances at getting a random star.  Given that you have a 1/16 chance of getting a random star on any punch where you're holding a star, that works out to roughly a 44% chance of getting a star in phase 1.  After he found that, I put together a bunch of strategies of what to do depending on when you got a random star, how many you got, and what refill he gave in phase 2.  The odds in getting a fight this good ended up being around 1/50.  Funnily enough, I actually messed up pretty badly in phase 1- he blocked the first punch I threw.  The run probably would've been over if he didn't give me a random star right after, and he proceeded to give 2 more after that, along with a good refill.  Yet again, this fight gave the god luck  
By this point, I knew that this run was something special.  I felt like I was dreaming.  It was almost as if I couldn't get bad luck; the game wouldn't give it to me.  I was just hoping that I would be able to get a few decent fights to close out the run, but as it turns out, the game wasn't done giving perfect RNG.

Mr. Sandman (2:20.00): I press select to start with half health, because I want to take a knockdown late in the fight to get Sandman to do another dreamland express.  In order to get my health low enough, I take a few hits intentionally in the first phase.  

The two main pieces of RNG in this run are the refill Sandman gets for phase 3, and the delay he does once you get up from your knockdown.  I got the good refill (75%) and the shortest delay (50%), and this caused my time to be 2:20- pretty much as low as this fight can go for single segment.

Super Macho Man (55.97): In late 2015, MTPO speedrunner Ouijawii found out a way to sneak in extra hits on Super Macho Man.  When he told us the following day, the top 4 runners at the time (sinister1, zallard1, ouijawii, and myself) all collaborated for hours on sinister's stream and came up with 2 different strategies that use this mechanic to our advantage.  The first one gets him down at 30 seconds with either 1 or 2 stars, and the second one gets him down at 32-33 seconds with 2 stars guaranteed.  Having 2 stars going into the second phase is extremely beneficial, so I decided that I would use the guaranteed 2 stars strat if I was on a godlike pace, and if not, I would use the risky strat.

As it turned out, this was the most godlike pace I could imagine, so I went and did the safe strat.  I ended up getting the best luck possible in this fight, and although I don't know what the odds were in getting it, I would say probably <25%.  Even with the risky strat, you can't get less than a 53, so getting a 55 was pretty much ideal for the strategy used.

Before the Tyson fight, I had to pause to sort of appreciate the moment.  This was the best pace anybody had been on by a long shot, and it wasn't really because of execution.  It was because the game literally was trying to do everything nice to me that it could.  This pace was practically like a dream.  Nearly every fight had given me PERFECT RNG, which I didn't even know was possible.  Had I been on a pace a few seconds worse than this, I probably would have felt more nervous before Tyson, but instead, I was too in shock of the pace I was on to react.  I pretty much just assumed that the run was dead, and that I would choke on Tyson, or that he would give the 8 second delay.  As I pressed start, I didn't know what to expect, but I knew that regardless of how the Tyson fight went, this run would go down in history- whether it would be known as a world record or the greatest choke of all time, I wasn't sure.

Mike Tyson (2:27): If you don't know how the Tyson fight works, here's a run-down. In round 1, for the first 1:30, he exclusively throws uppercuts, and for the last 1:30, he throws hooks.  You can hit him twice after each punch, regardless of whether it's an uppercut or a hook. For the uppercuts, if you just dodge and hit him with the correct hand (same side of the screen as he threw his uppercut), your two punches deal 5 and 1 damage, respectively.  However, if you delay your 2nd punch just enough, that punch deals 5 damage instead of 1.  The timing on that window?  1 frame.  If you're too early, it deals just 1 damage, and if you're too late, he blocks it, and it deals no damage.  Tyson starts with 96 health.  That means that to do the first phase of the fight optimally, you need to land all frame perfect punches except for 1 (9 punches from Tyson where you deal a total of 90 damage, plus 1 punch where you deal the 6 damage necessary to get 96).  Unfortunately, you need to miss this punch by being early, or else you fall 1 damage short at 95.  He then would need to throw 1 more punch that you would need to dodge and counter, which on average, wastes 5 seconds in phase 1.

Once Tyson switches to hooks, your punches deal 2 damage each.  Meaning, each time he throws a hook, you can deal 4 damage to him.  Similar to uppercuts, there is a way to deal more damage- if you wait to throw your first punch until the last frame of Tyson's vunerability, it will deal 5 damage, and you won't get a chance to throw your second punch.  This is the optimal way to deal with his hooks, but it is very risky- if you're late, he blocks your punch, and instead of dealing 5 damage (or 4 if you're early), you deal 0 damage.  Note that even if you are perfect with your hooks, you can't deal as much damage to him here as when he does uppercuts.  This means that any time you lose to missing uppercuts in phase 1 ends up costing you more time overall, because the more uppercuts he throws in the 2nd phase of the fight (before 1:30), the more damage that can be dealt.

Tyson is also incredibly random.  He can do a variable delay after nearly every uppercut and hook, but the big killer is the 8 second delay that he has a 50% chance of doing in phase 1.  It actually wastes more than 8 seconds (typically between 10-16) because of the property that I explained above, so Tyson doing the 8 second delay pretty much kills any run.

In this particular fight, I start off by landing every frame perfect punch.  Then, I'm 1 frame too late, and I get blocked.  Thankfully, Tyson didn't do the 8 second delay, but I get my first knockdown at 1:02, which would've been 57 if I wasn't blocked.

He threw 5 uppercuts in phase 2, and he just BARELY got that 5th one out in time.  I hit all 5, and send him down for the second time at 1:42.  Given the same execution and pattern in phase 2, if my phase 1 had been a 57, the 2nd knockdown would've been 1:32.  

In phase 3, I hit 6 frame perfect punches on his hooks, which was good enough for a 45 second phase, ending the fight at 2:27 (would've been 2:17 if I wasn't blocked in phase 1).

In total, I hit 20 frame perfect punches.  If I had missed that one punch in the first phase by being too early instead of too late, it would have been 10 seconds faster.  In other words, even with the same amount of mistakes, this fight could have changed drastically.  Although, to be honest, it's a miracle to me that I pulled out a 2:27 on a pace this good.  2:27 is a really good fight, and although it could've been better, I'm fine with it on a run like this.

I'd like to thank the two guys who influenced me the most in terms of running this game- sinister1 and zallard1.  The reason I learned to run Mike Tyson's Punch-Out!! was because of watching sinister's stream back in late 2013/early 2014, and his series of tutorials helped me learn some of the more difficult fights.  Zallard also was a big influence for me, as I watched his stream shortly after finding sinister's, and his encouragement helped me push this time down to as low as it is.  Without these two, this run would not have been possible.

In all, although this run is beatable, the luck required, as well as the execution on Tyson, would have to both be insane.  There are people who could do it.  However, I will say this- if this run does get beaten, without the use of any new strategies, I will not be the person to do it.