Alright.  So I did a SS speed run of this in February 2010 in like 4:35 or something, and have been revising my route since then.  I've been pretty lazy about actually typing up my route and posting it here, but I suppose I should do it and get some feedback before I do any serious attempts.  I'm not entirely sure the best way to go about this, but I'll do my best.  I got a lot of help for this route from lich.  He gave me a lot of great suggestions, including RXR!  I was previously using LX to level in the beginning, and was leveling up a shot 2, and all kinds of other crazy stuff.  He improved my route in the beginning of the game significantly!

First off, I'll be using two robots for this run.  I make the 2nd robot in the Old Mansion ideally, and never bother making a third.  Awhile ago I consolidated my notes and finally made a spreadsheet for the stats and equipment of my robots, and I've been adjusting it as my route develops.  You can see it here:

ROBOT STUFF

So the general idea behind this route is pretty straightforward:
1.  Use the RXR combo to kill everything in one hit.
2.  Build Axe 3 level 9 as soon as possible (I'll abbreviate this as Axe 3.9 just to make it a bit easier).
3.  Get to level 20 as early as possible to make vanish (vanish = no more battles ever again).
4.  Max power and rely on evasion to kill bosses.

With this in mind, the route wasn't too bad to outline.  I guess I'll try to break up this game into chunks based on boss fights.


Beginning through Meta Crab

Collect: 500 GP & Clean (Hero's house), Repair & Clean (Father's house), 300 GP (Forest), 1000 GP (Mayor), Scrap 2, Scrap 5, Cure, & Repair (Tool Shop), Cure, 250 GP & Scrap 4 (Cave), 1000 GP (Meta Crab)
Total Gold after Meta Crab:  3050 GP, minus robot maintenance costs
Inventor's Friends: None!

I think the best option here is to just list at the beginning of each section what I will collect instead of trying to list it throughout the section.  Easier for me anyways.

In the beginning of the game, money and experience are very critical.  I do go a bit out of my way to collect the 300 GP and 250 GP in the forest and cave, respectively, but I think it is worthwhile at this point.  After I make my robot, I immediately program RXR (sword-bomb-sword).  RXR teleports the robot in front of each enemy in the battle and swings once.  With a high enough power stat, this is enough to kill nearly every single encounter through the middle of the game.  The only time it won't work is if there is a capsule 1 tile in front of an enemy, because the robot will land on the capsule and the capsule will damage the enemy first.  This is catastrophic.  More on that later I guess.  So yea, every battle will be fought with that strategy for a long time.

I need to reach level 3 after I collect the rusty drill in the cave to save a trip back to town and a lot of time.  Level 3 is a long way away normally, so I rely on some capsule bonuses to help me along.  I only fight mushrooms, mines, and gels at this point because they are very safe enemies and the experience doesn't vary enough with the other enemies to make it worth the effort.  I just want to hit capsules, hope for good bonuses, and then RXR to victory.  With luck, I won't need much extra farming to reach level 3 at the right time.

Now, after I reach level 3 I'll use the portable R&D to make the drill (rusty drill + clean) and a Shield 2 (scrap 5 + clean) and adjust my stats.  Shields are, hands down, the most broken things in this game.  It's not even fair how much they help.  Shields give a boost to defense and evasion, and have a chance to block stuff on their own.  Shield 2 has a monstrous effect on the speed run early on.  Now, the enemies in the lower cave (after you save the children and return to fight Meta Crab) are much harder than mines and gels, etc.  Spiders do a ton of damage, have good speed, and are resistant to the RUN command (RXR).  They also give 0.3 MEGs of experience and commonly drop repairs, both of which are very desirable.  My solution for quickly dealing with them is to pour my points into power and kill them with RXR anyways.    I chose a 1 energy, 54 power, 5 speed build here.  I only need enough speed to actually hit the spiders.  10+ speed seems to give a good hit chance.  54 power is the minimum value I have been able to use to guarantee a kill with even the lowest possible damage RXR hit.  With 5 speed and Shield 2, I normally hit them but there's still a good chance of a miss.  This is a really dangerous section.  I fight most battles down here until I reach level 5.

At level 5, I adjust my stats for the Meta Crab fight.  In my current route, I fight Meta Crab with Shot 1, but in the past I have upgraded it to Shot 2 to boost my damage a bit.  The damage goes up by ~6 or so per shot, but it's enough to save one round.  The only advantage, IMO, would be the 500 GP I would get from recycling scrap 2 instead of combining it.  I'm not sure what I want to do about that yet.  Anyways, Meta Crab is simple: walk to the second farthest tile I can, and fire.  Don't move at all after that.  Just keep firing.  I'll survive as long as he doesn't use the blue orb attack.  I pick up the cure outside the cave just in case he does use the blue orb attack.  With shot 1, I do ~38-41 damage, and with Shot 2, I do ~45 damage.  Meta Crab has 200 HP, so it's either a 6 or 5 round fight.  Not bad.


Old Mansion stuff through Mamurana

Collect: Shield 2, Axe 1, & 1000 GP (Grave), 500 GP, Repair, & Scrap 3 (Old Mansion), Scrap 1 (Clock tower), 1000 GP (Recycle scrap 2 & scrap 4),400 GP (drops in scrap 1 room)
Total Gold after Mamurana: 5950 GP - 4000 GP (Robot 2, Shield 3) = 1950 GP + drops from Ho-Ho
Inventor's Friends: Axe 1 (IF9), Shield Pack (IF10)

This section is a bit scary, too.  I want to adjust my stats after I level from Meta Crab.  I need that 5 HP because there's a fight with birds in the forest of illusion that is usually an unavoidable surprise attack.  With 1 HP, it would be game over.  The 5 HP is just for a bit of safety.  With 60+ power, sword 1 is more than strong enough to OHKO monks, ho-hos, and plasmokes with an RXR.  25 speed is enough to never miss with RXR.  Well, I don't know that for sure but I certainly never missed with that much speed here.  I loot the grave for useful stuff and money, and on the very next level that I get, I want to make an axe 2 (scrap 1 + axe 1).  While those 3 enemies aren't a threat to me, elesmokes are very threatening.  I need axe 2 and usually 80+ power to kill them with RXR because they are resistant to it.  They are rare enemies though so it's not usually a concern.

I run straight to Flavon's room, loot the dresser, learn how to make relay, and then spend the next few minutes killing ghosts in the library.  The library is a nice convenient spot to level because it has 2 ho-ho encounters and it has the IF books I want to read.  Ho-hos drop 500 GP pretty commonly, which is exceptionally useful.  Fortunately, with my gold route, I only NEED like 2 drops from them.  I usually get ~3 drops from them while I level, but it's nice to not have to rely on them.  Unfortunately, I run into a lot of troll capsules with these encounters it seems.  If there's a capsule directly next to an encounter with a ho-ho, I need to wait until the bonus timer expires.  If I RXR, the enemy will survive and kill me; if I try to run, I usually fail and die.  The only time it isn't an issue is in a battle with all elesmokes/plasmokes because I can just guard, let them break the capsule, and then kill them.  Ho-hos' attacks either pierce or are long range, so it doesn't work.  That is the only real threat.

I want to reach level 10 here because I absolutely need shield 3 for the rest of the game.  It is simply one of the best items in the game.  It is affordable, available early, and extremely effective.  I need IF 9 too because that lets me level the axe.  While I could do that with hammer 1 (IF6) and sword 1 (IF...2?), that takes far too long and I need to reach level 10 anyways.  After I reach 10 and read the IF books, I head to the R&D room to create relay, shield 3, and my second robot!  The second robot will be the one doing the fighting for the rest of the game....almost.  Robot 1 will be used in the next section, but after that he is just support.  Robot 2 will be programmed with RRR in the first slot and RXR in the second slot.  He is only RXRing for a few forced battles, not for leveling. For now, I give robot 2 a pair of shield 2s just so he can survive a bit and revive robot 1.  The roles are reversed in the next section: all robot 1 will do is revive robot 2 if he dies.  That is his role in this game.  The reason I use robot 2 for attacking is because he has a stronger melee attack by design.  I don't know exactly how much it is, but it is significant for the leveling against shells.  For this battle it isn't worth switching gear and whatnot, but it definitely is later on.

Anyways.  The rest of this section is straightforward.  Just go through the clock tower normally.  I make a detour to grab the scrap 1 even though it forces two battles because it lets me make axe 3.9 in the very next section, and it gives me an additional 400 GP from those battles.  The Mamurana fight is really easy.  She has 400 HP and with max power and axe 2, she gets hurt badly.  I think it does over 100 damage per hit.  It's nuts.  I can't reach her on the first turn so I go to the bottom row, move as far forward as I can, and defend on the first turn.  She can't switch rows so I'm safe.  I hit her next turn and she makes clones.  I just keep mashing away.  She seems to like staying in the middle, but I guess it's random so who knows.  She isn't really a threat because normally one of the wrong Mamuranas attacks instead and I'm safe, and if the Mamurana in my row attacks, shield 3 usually blocks.

Edit:  I made a slight miscalculation before.  I thought Robot 2 cost 2000 GP, but it's actually 3000 GP.  This makes funds pretty tight.  However, if I get two drops from Ho-hos, I am still ok.  If I only get 1, I can still make it work if I sell everything I have (shot 1s and sword 1s, maybe a big bomb if I got one).  Mamurana takes 6 hits (I do ~78 damage per hit) but it's a really safe fight still.


South Isle through Big Eye

Collect: 500 GP (volcano cave), 400 GP, 600 GP, 800 GP, Sword 4, & Scrap 9 (Volcano Base), Hammer 2, 1000 GP (hidden tiles in the shrine), 4000 GP (Recycle sword 4 & hammer 2)
Total gold after Big Eye: 7300 GP - 3200 GP (8 axe 1's) + drops from berets + whatever I had from before
Inventor's Friends:  None!

First thing I do here is make an Axe 3.9 by combining my axe 2 with the scrap 1 from the clock tower, and then with 8 more axe 1's to level it up.  It is definitely worthwhile to level this up.  No question about it.  With that finished, I head to South Isle.

My route here is a bit variable based on the drops I get.  There is a 500 GP capsule in the first room of the volcano cave, and a 400 GP capsule in one of the later rooms in the volcano base (the one before the R&D room I think).  I don't know if it's worth getting them or not.  I already grab an 800 GP capsule and a 600 GP capsule along the way.  I think I may pick up the other two if they are accessible.  (Edit:  I want to get these now.  They are along the way).

I go through the volcano base normally, exit out the garbage chute, and head through the volcano cave.  I don't bother picking up anything out of the way, and I avoid as many battles as I can.  They are simply not worth it.  In the very last room of the volcano cave, I do some more leveling on the berets.  They give 2.5 MEGs of experience each and drop 800 GP as a common drop.  That's absurd!  Robot 1 can kill them in one hit with axe 3.9 and 20 power.  The leveling here is a bit slow because I don't have boots 4 or turbo pack, so I have to guard for a turn while I walk over to them.  However, this is still the fastest option here I think.  The fastest possible time would be to just not level at all here and then level exclusively against shells later, but that is incredibly random.  I don't want to try that for SS because my times for leveling there have ranged between like 7 minutes and 25 minutes.  Not ideal.

The plan is to level to within ~5 MEGs of level 16, but I don't want to reach level 16 here.  I want to reach level 16 after I collect scrap 9 in the volcano base on the return trip so I can make boots 5 with the portable R&D before Big Eye.  The boots 5 adds a significant amount of evasion and guard that lets me use a power build against him without as much concern for my safety.  I need to experiment a bit more with stat distributions for Big Eye.  I have in my notes a 41-99-30-20-0 build for him, but I don't know why I chose those numbers.  I think I want to try a 1-99-0-91-0 build with boots 5 and shield 3 and just hope I don't get hit.  Big Eye only takes a few hits to kill (I think 4 hits), so there's a chance it'd work.  Since boots 5 are so slow in battle, the fastest option for Big Eye is to move down one row and defend.  He usually lasers and then flies over to the right side of the screen.  Then I just move to the center row and slug it out with him.  It isn't a bad fight.

Thanks to this really awesome Robotrek site, I found out that there are two hidden items in the shrine: 1000 GP and hammer 2!  Hammer 2 recycles for 3000 GP, so that was a massive find for this route.  So yea, I pick those up in the shrine because it's barely a detour and is a HUGE help in the next section.

Edit:  I got a HUGE amount of money from the berets in my practice run this morning.  I ended this section with 12000 GP, but I picked up the 500 and 400 GP capsules in the cave and base, respectively.  They don't cost much time and are totally safe so it's worth it.  Even if I get two or so fewer drops from berets, I can still make it work.

For Big Eye, if I stay in the very last 2 tiles against the wall in the bottom row, when he flies over to me he will be facing backwards.  He dies in 3 back attacks, which is great.  I decided to try the 1-99-0-90-0 build I was wondering about (with boots 5 and shield 3) and Big Eye didn't hit me a single time.  This is a fantastic strat.


Rococo through Papamecha

Collect: 1500 GP (tunnel)
Inventor's Friends: Shield Pack (IF15), Turbo Pack (IF18), Vanish (IF20)

I'm not going to bother trying to estimate how much gold I'll have from now on because the beret farming gives me a very variable amount of gold.  I sometimes have well over 10k GP, and other times like 8k or so.  It really changes significantly.

Killing Big Eye brings me to level 18.  Before I go to Rococo, I stop by Father's House to read IF 15 for Shield Pack.  After I return home from being captured, I read IF 18 (Turbo Pack) and immediately create one in the R&D in my house.  This is an item that I never even considered for a speed run until very recently.  You see, Turbo Pack boosts your mobility to maximum (highest movement speed, greatest range).  Since boots 5 gives the biggest evasion and guard bonus in the game except for shield 5, it is better to use boots 5 + turbo pack than to use shield pack + good boots.  I use turbo pack for leveling against shells in the tunnel here, and for fighting a few bosses later on.  It's one of the most underrated items out there I think.

So.  The ultimate goal is to reach level 20.  Instead of fighting against berets until then, it is faster to fight shells in the tunnel.  Shells have massive evasion and guard, and they flee whenever they get the chance.  With 99 power, 99 evasion, boots 5, shield 3, and axe 3.9, robot 2 will always kill shells in two successful hits, and will have the greatest possible chance of connecting with a hit (until he gets shield 5 at least).  Turbo pack will let him reach a shell in the middle row on the first turn, so they don't have as great of a chance of running away.  If the axe 3 is one level lower, or if robot 1 is used (remember, robot 1 doesn't have the power bonus), then you aren't guaranteed a 2HKO.  It is very tricky to kill them.  Why bother then?  Well, shells give 30 MEGs of experience!  I only have to kill 2 to reach level 20 from 18.  Conveniently, they are in the first room, too.

I kill two of them, reach level 20, backtrack to my house, read IF20, create vanish, and then finish this section without any fear of battles.  I have to keep unequipping vanish every once in awhile or else it'll go away forever.  That's basically the strat for the rest of the game.

Papamecha is actually a frustrating fight.  I keep the Turbo Pack equipped for him to get back attacks; with back attacks I think he dies in 3 hits.  Not much else to say.  Shield Pack might be worthwhile, but evasion doesn't help against crits and that is the only way Papamecha hits me.  If he crits me, I die.  It's better to use Turbo Pack and go for the extra damage from back attacks.

Edit:  I ended my practice run after I reached level 20 because that's where the strats stop changing.  I killed 3 shells in 4 battles.  It took under 2 minutes (like 1:40 or so?) to gain those two levels.  I may want to fight one extra battle after Big Eye (against Boss Toad maybe) to grab a bit more experience so I don't need to kill 3 shells.  The time would be about the same either way, but I might favor just killing two shells.

With my vast wealth in this practice run, I bought a Turbo Pack, a Shield Pack, and two shield 3s (10k GP total).  And of course, two vanishes once I reached level 20.


The uhh....the rest of the game

Yea.  Seems strange to just lump the rest of the game into one headline, but it really is that simple.  The rest of the game basically comes down to using vanish to get through dungeons, making sure to unequip it once in awhile so you don't lose it, and then killing bosses.  For forced battles, I switch from Turbo/Shield Pack to bomb 1 so I can use RXR again.  Some bosses are resistant to RUN commands, so it is best to kill them with normal swings.  I'll give a brief overview of how all the boss fights go.

Bugbug 1:  Use Turbo Pack and RRR.  He only has 1000 HP.  This is a quick fight.
Blackmore:  Use Shield Pack and RRR.  Since he has a knockback attack, it is best to defend in your spawn point for two turns while he slowly rolls toward you.  Then just roll up to him and RRR until he's dead.
De Rose:  Hardest boss in the game.  She crits very frequently and is resistant to the RUN command.  And she has a lot of health.  I just use normal swings and Shield Pack and hope for the best.  Backup robot is crucial. 
Bugbug 2:  Turbo Pack and normal swings.  Easy.
Gateau 1:  Turbo Pack and RRR.  He dies ridiculously fast with this setup.  This is where Turbo Pack shines.
Final Gateau:  Shield Pack and RRR.  I actually don't know if Shield Pack matters here or not because he hits with crits really frequently, but I use it anyways.

When I have the money, I buy the rest of the equipment for robot 1: a pair of shield 3's and a shield pack.  On Choco, I go to the tool shop to purchase another scrap 9 and make a shield 5 with it.  I make a pretty significant detour to make this, but if it prevents even one death it makes up for it...and it certainly does prevent one death.  It's an amazing shield.  I grab the scrap 9 in the sky base and all of the money capsules that are along the way.  I also buy a handful of repairs in Rococo before I head to Sky Base usually, but that depends on how lucky I got with repair drops from the spiders early on.


So that's it!  So far my current best with this (well, a slightly outdated) route is 4:01:XX hrs on the Japanese version.  I'll do some more test runs in the near future.

I just purchased Robotrek recently. I understand a small bit about it from the technical side, but next to nothing about the story or route. I'll play through it sometime this Summer and update you if I find any sort of improvements.

For drops from common enemies, it may be possible to manipulate some rare drops in a SS w/Resets run since battles proceed completely deterministically once you start them. That would take a lot of testing to find a quick method that works though. For a regular SS it wouldn't be feasible. The increased critical chance would be especially nice for the combination attacks, but I still don't know the differences in damage ranges that you'd be expecting between Punch 3 w/ 1/4 crits vs Axe 3 w/ 1/16 crits.

Still so much to investigate. Any possibility of reappropriating some of the robo points to defense/HP/etc for some of those otherwise suicidal fights?

I'll check again when I get some free time, but I'm fairly certain the frame counter advances even during dialogue.

And yes, getting a crit supercedes the check for whether an attack was dodged. I still don't know the formula for accuracy/dodging, but that will be one of the next things I look into when I get some free time.

Hey PJ, I'm glad you made this thread. Sorry for going MIA after saying I would test something out, I was in grad school at the time and it just kind of got away from me. But now that I'm done and I have tons of free time again, let me know if there's anything you want me to test out for your speedrun and I'd be happy to do it! I really want to see this run get made

I started compiling enemy information and investigating enemy tables in the ROM, but there's a ton of stuff to look into yet. The list below is mainly a reminder for me, but also to keep everybody up to date on what to expect:

-Formulas for dodging; finding the relationship between your speed and enemy speed.
-See how/where enemy attack crit values are stored.
-Enemy defense tables, and also checking to see if enemies have different defenses for different weapons.
-Any special weapon/shield modifiers that aren't obvious. Also, cataloging all of the crit values.
-Actual values relating to the robot "traits". I didn't see anything before, but I can check more in-depth now that I'm familiar with a lot of the battle memory structure.
-Actual drop rates, item track rates, and finding out when/how the drops are determined.

So yeah, lots of info to come... eventually.

EDIT:

Alright, I've dug into the drop probabilities a bit. I'm still trying to figure out where/when the "track" is determined, but it looks like the rest of it is pretty straightforward.

So the game uses what I call a crypto chain, which is sort of a fibonacci sequence of registers starting at 7E0539 and feeding down to 7E052A. It seems like only the lower numbers are used for RNG purposes. For drops in particular, the game uses 7E052A on a couple different occasions. First, immediately after a battle is over (during the "poof" animation) the crypto chain will be incremented once to determine whether or not a drop occurs at all. If the last 2 bits of 2A are 0 after the call, then a drop will be left behind. Then when you open it, the crypto chain is called again, and 2A is again used for determining the rarity category of the drop. If the last 4 bits are 0, then you get a category 2 drop. If all 8 bits are 0, you get a category 3 drop. Otherwise, category 1 (most common).

So to put this in perspective:
-25% chance to get a drop from any given enemy (1 in 4).
-6.25% chance to get a rare drop (1 in 16) so long as the drop occurs. Compounded with the above, only 1.56% chance to get a rare drop from any given encounter.
-0.39% chance to get a super-rare drop (1 in 256) so long as the drop occurs.

Note that in terms of per-encounter rates, this means that the probability of getting a rare item is around 1.56% and a super-rare item is only .098%. These numbers aren't entirely accurate either, since the 2 separate calls to the crypto chain are not independent. The probability of obtaining a 0 byte value after a prior xxxx xx00 value is quite a bit less than the direct product of the probabilities, but it's too late and not important enough for me to analyze completely. The point is: it is incredibly unlikely that you will obtain a super-rare item, and it's most likely not feasible to try to manipulate it outside of a TAS setting. EVen then, there are very few ways to force the crypto chain to be called arbitrarily, so it will be difficult to force an advantageous chain configuration.

On that same note, the subroutine for determining the item starts at 83/FCAB, and the drop tables are located at 88/D554. It uses an offset that I imagine describes the exact enemy and configuration of the battle, but I haven't traced it completely yet. It also looks like 7E052A and 7E052B are set arbitrarily at some points, just like 7E052C is set equal to the frame counter when a battle starts, but I don't know which values yet. Finally, 7E05C6 is a part of the description for the enemy configuration... cross-reference its value with those listed at http://www.hook-shot.com/games/snes/robotrek/codes.php for encounter types. If this value plays a role in determining the drop table offset, it does so outside of the scope of my current traces.

EDIT: New discovery! Enemies on the map are predetermined as to their configuration. There is absolutely no randomness in what set the enemy will be, only in how many will appear in battle and how many containers appear on the battlefield. I'm fairly certain that it's the configuration of enemy that determines the drop, so you can target specific enemies on the map for getting certain drop tracks.

Example: In the forest, the very first mushroom to appear will always use configuration 2A. The second mushroom will always use 29. The third mushroom (above the ball) will always be 02. Each of these are different enemy configurations, and also different drop tracks... even if the items received don't differ in this case, they likely will for other enemy sets. This can also be useful for grinding in some sections, since some enemies on the map may have easier enemies or give better experience than others. Still working on seeing how this plays into the item drop table, though. After that, I can start compiling all the information

OK, here's a bit of a data dump of what I've come up with this past week. These are taken verbatim from my notes, so I will try to paraphrase afterwards.



Basically just means that the enemy data is loaded based off of the configuration, and that the configuration has a pointer to each of the enemies in that configuration in turn.



In this case y is just incremented by 2 and the next value is loaded. The slots refer specifically to only some of the enemy attributes, but those are the ones that are loaded initially. I was not able to observe 32A0, 22C2, or 22C4 change at all during normal battle, but I was only checking specific actions. For defense, it it simply takes the power after other modifiers (they seem to vary by attack) and subtracts the target's defense to determine the final value. This goes for your robo too.



Yeah... the rabbit hole goes pretty deep and isn't all that important. Main point being, your probability of a successful dodge depends on the difference between your evasion and your target's evasion.

Below are my full notes which may have some other small tidbits. The main thing to note are some of the RNG pieces. Yes, some of the data is redundant. View only if ye are of strong heart and mind.

Good job PJ! Looking forward to maximum silliness

Yesh!  Yesh!  This is delicious!

Thanks!  I'm glad you enjoyed it. 

I watched a bit of it and decided to wait for some of that lovely audio commentary.  It's great to see a run finally get finished though!