Big thanks to everybody at SDA for keeping things running smoothly!

I need to give a HUGE shoutout to kareshi for being such an inspiration to me and for basically teaching me everything I needed to know to run this game. I learned a good deal of what I know now from his old ustream tutorials. He was constantly supporting me during my attempts, even if he didn't approve of my weapon choice. He definitely made the process much easier on me. Thanks so much, kareshi!

I also have to give a big thanks to Pasky for making a hitbox viewer for this game. I use this thing SO MUCH. It gave me a much better understanding of damage/health values and chest triggers, which were incredibly valuable. Being able to see the hitboxes is really useful for this game.

I have a long history with this game. I played this a lot as a kid, but never made it very far (I think stage 5 stopped me for a long time). Regardless, I loved playing it. When I started getting into "competitive gaming" back in my TG days (I know, I know), I knew I wanted to someday speed run SGnG. Kareshi's run looked completely insurmountable though, so I never really pursued this. When it came time to suggest games for AGDQ 2012, I offered SGnG without having any experience. Having it committed to the schedule would finally force me to learn the game. Kareshi offered to race me, so I really had to grind out this game or else I'd risk getting destroyed.

So we did the race at AGDQ2012 and I had a blast, but I really wanted to improve his run. I kept grinding strats for a few more months and made some solid progress. In the back of my mind, though, I wanted to use the torch because I had a sneaking suspicion that it'd be faster. Exo found a TA on nicovideo that used the torch, and I was amazed that it was so fast. Even though it was clearly segmented, I was motivated to relearn the game with torch. I spent a few more months rerouting the entire game with the torch and resumed attempts. After continuously modifying my route so it'd be safer and have a chance of finishing the game, I decided to just get a strong dagger run and then improve THAT with the torch. That would force me to stick with the dangerous torch strats (or else I'd risk losing time to a dagger run, which is unacceptable), and ultimately produce a better final run. The intermediate dagger run is the run you are seeing now.

This run was done on the Japanese version of the game (thanks for the cart, Jinny!), and on normal mode. There are no differences between the Japanese and US versions except for the end cutscene (which is ~25 seconds faster on Japanese). The RNG and enemies behave identically between the two versions. I only ran this because I didn't like how my other cart was behaving.

The choice to run normal mode is a bit more complicated. Kareshi's run was done on professional mode. On professional mode, all the random enemies tend to move faster or take action sooner (zombies, yellow devils, the falling skulls in stage 1, the swinging axes in stage 2), and the fuzzy bears have longer fire/ice breath. Aside from that, everything is more or less the same. Stage 2 is actually faster on higher difficulties because you don't have to wait for them to swing. You can proceed one axe cycle sooner. Anyways, all of the boss quick kills are the exact same on professional mode. Boss health is higher on higher difficulties, but not by very much; in general, attacks do so much damage that the same exact strats will work regardless of difficulty. For proof, I use the same strats in loop 1 and loop 2 in this run, and the bosses die in the same amount of time.

So if everything is so similar, why run normal mode? Well, it lets me take more risks in the stages. My goal for this game is to have the final run use the torch. Torch can do some silly things that are incredibly fast, but require very good luck from the enemies. These "good patterns" also exist on professional mode, but are more rare. Therefore, I had a few options: 1) run this game on professional difficulty, but take it a bit safer so I have a chance of finishing a run; 2) run it on professional and use the fastest strats, resetting until I get good luck; 3) same as option 2, but on normal. Since the runs produced by options 2 & 3 would be significantly more entertaining and faster than option 1, I wanted to do that. Choosing to run on normal mode just cuts down the number of required attempts by a few thousand. In the end, the run will look the same as if I did it on professional, but this just saves my sanity. Even though I'm not using the torch in this particular run, I still wanted to do it on normal so I can more accurately compare to it when I use the torch.

This run had the absolute strangest luck I've ever seen in my life. It is seriously a miracle that this run finished. I'm not even joking. There is no way this run should have finished, but I somehow saved it. I lost my shield in stage 5, I lost the rest of my armor in loop 2 stage 1, I had spawns that are normally undodgeable but somehow dodged them, etc etc. This run is really something special. I'm really glad it was this particular run that finished. Haha. This was my 3345th attempt at a run (~1600 were using the torch), and only the 6th run that made it to loop 2. Of the 6 that made it to loop 2, one of those finished (this one!), one made it to stage 3 and got hit by lag, and the other 4 ended in stage 1. Making it this far was truly something special.

I could talk for days about the luck in this run and specific details and room for improvement, but I did that in the audio commentary. I highly recommend listening to that if you want to know more detailed information about the run. Also, I forgot if this game had a "the end" screen at the end, so I accidentally cut the recording early. Oops! I promise I won't do that again. ^_^ I also left in a few frames of the previous aborted attempt, because that's the only way that the bow will ever be seen in a submitted run. Thanks for watching!

My 42:05 speedrun from 2008 was one I was very proud of. I didn't think anyone would bother to compete with it, and it took me 70 hours of attempts to get it done. But it had issues, mostly with the final 3 bosses, so after finishing my NES Ghosts 'N Goblins self-improvement, I got back to this one - strangely, not expecting it to take very long. A full month and 92 hours of attempts later, here it is - 39:41.

Immediately prior to this speedrun, I had done the following for Twingalaxies's online scoreboard - two Highest Score runs for SNES and Wii Virtual Console, and one Normal mode speedrun for Wii VC. The points runs touched-up my mental notes on enemy behavior, and the Normal speedrun got me back into the speedrunning mindset. Starting up the Professional attempts again, I managed in a short time to get a 40:52 - which beat my SDA time by over a minute. It was a nice time on the stopwatch, but the levels weren't nearly as well executed - in fact, every stage but the last one and the final boss were shorter on the SDA run. I played an aggressive Round 2 Level 7 and Final Boss, and those two alone had 90 seconds of improvement. But I decided (with help from loyal Kareshi TV viewers on ustream) to keep going until my levels looked as good as the SDA run. So I got another finished game somewhere in the 41:XX's, and one that was 40:48 that I almost stopped with. That run was good-looking overall, but had a few missed jumps and bad luck on some boss fights. Finally I got a run where the levels went great, the boss fights went great, and the final bosses went great too.

Most of this speedrun is not improvised, but instead tightly scripted based on knowledge of the enemy's behaviors. The enemies can't be counted on to do exactly the same thing each time, but most of them have only certain things they can do, with various timings. So I have to pay attention to when and how they're moving and be sure I can deal with them without running into other enemies or flying projectiles. I'm jumping as often as possible because it's slightly faster than walking.

During this run I was surprised to find so many new timesavers, since I thought I did a pretty good job of seeking those out back in 2008. The new ones this time around include:
- Better zombie spawning manipulation - all based on which ones you kill and which ones you keep alive.
- Better eyeball clam #1 strategy
- Shooting the spiky balloon while jumping the gap
- Better Level 1 boss strategies for both neck-version (see Round 1) and egg-version (see Round 2)
- A new shooting method using two fingers on B and Y means super-rapid fire if needed (comes up every now and then)
- New path through the ghost ship (thanks to a TAS path pointed out to me by any1else)
- Shooting more things off-screen while in the raft section reduces slowdown spots
- Some fire bear flame manipulation
- Shoot while jumping up to floating platforms in Level 2B (see Round 2 - it just may save your life!)
- Better location for the Dragon magic on Level 3 boss (hits him going up AND down... ideally. I only got a "down-hits" Dragon in Round 2.
- Can manipulate the worst-placed spiky balloon in Level 4B out of spawning. Doesn't save time, and he's not dangerous if I have Dragon magic, but it's a good thing to know for casual play.
- Can deal with the Bad Luck Flower (the second one in Level 5, after the spike pit) without getting hit or losing more than a second. Never got that right back in the 2008 attempts.
- Better manipulation of the behavior of the Level 5 boss
- Running under the bird head in Level 6 instead of killing it
- Better Levels 6 and 7 boss fight strategy - four knife shots replace one Dragon attack
- Much better Round 2, Level 7 Bracelet strategies vs. the bird heads and the bosses

The most time saved during this run is at the last 3 bosses. The Bracelet does more damage the closer you use it. Of course, this increases your odds of getting hit. If you're armorless but keep your shots close-range, you don't actually lose much time.

The final boss went as well as it could, with lots of point-blank hits. I did get hit, but was knocked ONTO a platform that I wouldn't have been able to reach had I fallen normally. So it actually saved me time. The rest is just up to the timing of the platforms and laser shots matching up.

Arthur dies in the ending credits. It's because he's trying to use B to jump, unaware that I switched that with A in the settings. He Game Overs because I had set the # of lives to 1.

Thank yous to all the ustreamers who supported me in this and watched so many hours of pain before this was finished.

This game was voted Second Hardest Game of All Time for a Twingalaxies article in September 2009, losing the #1 spot to Battletoads by a single vote.

The following list is copied and slightly modified from my 2008 comments.

"During the recording of this speedrun, I failed many times. The following is a list of things that went wrong and ended my attempt at least once - and in some cases many, many times.
1.Got the wrong weapon at the start
2. Hit by zombie
3.Hit by dog
4.Hit by skull fireball
5.Hit by falling skull
6.Hit by an eyeball shot by a clam
7.Hit by the clam
8.Jumped too soon and got swept away by a tidal wave
9.Missed the platform and jumped into the water
10. Jumped into a spiky balloon
11. Hit by Level 1 boss
12.Hit by ghost
13.Hit by axe
14.Hit by mimic
15.Missed a jump to the next raft platform
16.Hit by fish
17.Hit by star
18.Ran into an item drop that was a weapons change
19.Hit by shell shot by Level 2 boss
20.Hit by Level 2 boss
21.Hit by yellow devil
22.Hit by fire breath
23.Hit by falling lava
24.Jumped into lava pit
25.Hit the side of a bone platform
26.Hit by bat
27.Jumped into a gargoyle
28.Missed the jump onto a moving platform and fell
29.Hit by falling goblin
30.Hit by blue demon
31.Didn't make it to the rotating platform in time and hit ground spikes
32.Hit by green fireball
33.Fell off moving platform
34.Hit by gas jet
35. Hit by blue devil
36.Hit by fireball from Level 4 boss
37.Jumped into frozen thorns
38.Hit by projectile from flower
39.Hit by ice breath
40.Walked off the ledge
41.Jumped into roof spikes
42.Knocked into thorns by avalanche
43.Hit by fire breath of Level 6 boss
44.Hit by laser beam from Level 6 boss
45.Hit by bird head
46.Jumped onto fake chest
47.Hit by fire breath of Level 7 boss
48.Hit by laser of Level 7 boss
49.Hit by final boss's lasers
50.And sometimes I'd just quit because I spent too much time missing jumps or shots."
All other strategies used in this speedrun are self-apparent, if you'd like any more details on why I'm doing what I do, please ask me on the SDA forums or AIM me at Mozart2507.