Nintendo’s answer to Skylanders will come with the caveat of giving Smash players, not new characters they control, but characters the artificial intelligence will control for them. While it’s possible the AI will be more intelligent than the standard computer opponents, namely because they can learn, it’s likely that their base knowledge and capabilities will all have a baseline from the game’s own AI. Therefore, it’s reasonable to conclude that a character the CPU is more adept with will be the better choice when facing other amiibos.
Thus, I have conceived the “Amiibo test”. Using the 3DS demo, and later the 3DS version of the game, I will pit the amiibo figures’ characters against each other in order to learn which of the characters the game itself thinks are the best, therefore determining who, in fact, is the best choice when purchasing an amiibo.
Since the amiibos presumably have the capability to learn from repeated play, this is the only way to maintain a baseline test, as the amiibos themselves will likely adapt as the test progresses, changing their basic behaviors and skewing results. The base game AI is not programmed with this functionality, and so services this test better (at least until we confirm whether or not amiibo data can be deleted).
Once the amiibos are released, a test can be performed to verify my findings by pitting them against each other (provided I have the funds to acquire them all). However, it’s important to note the results may not line up perfectly, due the the aforementioned learning capabilities.
Since only 12 characters will be available in series 1, only series 1 character will be used to start with. Series 2 characters will be tested in this fashion when they are announced. Also, I will redo the test once the Wii U version comes out in order to test for consistency between the two versions of the game.
Since the demo released on Friday, I was able to get some preliminary data on 4 of the amiibo characters. Since Megaman does not have an amiibo in series 1, I did not record any data with him.
For Fun preliminary test:
Test with highest level CPU players
Test for 25 rounds, changing character position in each fight to avoid any player preferences
Extend testing period if test period does not yield consistent or conclusive results.
Note: Test was extended due to perceived trends that characters might equal out number of Wins after additional rounds. This was not the case, but test was extended to 50 rounds to verify.
Record character with most wins and point totals (KO’s – Falls) per character.
53 KO’s – 48 Falls = 5 Points
46 KO’s – 56 Falls = -10 Points
52 KO’s – 56 Falls = – 4 Points
54 KO’s – 45 Falls = 9 points
Determine character with most wins and character with highest number of KO’s. Most wins will be best under the standard For Fun rules. Highest point total would likely be better in stock matches.
According to our results, Mario is the most likely to win under the “For Fun” conditions, winning over Link and Pikachu by a wide margin, and just barely beating out Villager. Since Mario is a fairly balanced character, it makes sense that the computer controlling him could handle many of the random elements the matches would throw at him.
Mario narrowly beat Villager in wins, but was in turn beaten in points. Villager too is rather flexible, but tends more toward the defensive. His recovery move is easily the best out of the four test subjects. Because of this, Villager ended up not sustaining very many Falls each round, even if he didn’t score as many points per round as others did.
Link and Pikachu have a weak and a difficult recovery move respectively, so it’s not a surprise that the computer struggled to keep them on the stage with basic techniques.
In conclusion, it appears that Mario and Villager are top contenders after this preliminary test. Mario is clearly the better choice in most “For Fun” scenarios, while Villager is a better option if looking at stock matches, or going for a strong overall score. Pikachu did not fair very well, and Link should be avoided under these conditions.
Once we get our hands on the full version, we can test how these characters perform in a vacuum with the “For Glory Test”.