The Same Game Test, or SGT, is a balance guideline used to gauge the level of power a character class or option brings to the table. It is derived from the definitions and explanations of encounter challenges in the Dungeon Masters Guide and Monster Manual. It states that a character of any given level should have, on average, a 50% chance to win an encounter against a creature with a CR equal to the character's level or a group of creatures in a single encounter whose EL equals the character's level. On this wiki we have assigned this level of aptitude to the Rogue balance level. Characters who perform noticeably better than 50% on an SGT, especially at later levels, are generally considered to fall into our Wizard balance level. Characters who perform noticeably worse than 50%, especially at higher levels, are generally considered to fall into our Fighter or Monk balance levels.
The test was originally written to test 10th level characters, but has been expanded to also include levels 5 and 15 on this wiki.
Running a Same Game Test[edit source]
Running a SGT is essentially a thought exercise where you attempt to gauge the results of a character attempting to overcome each the listed challenges. A character should be assumed to be at full health and ability before beginning the encounter, but should not be assumed to have any knowledge of it that would allow them to plan specifically for that encounter. The encounter must be beaten, in whatever sense that most applies to the encounter; the ability to escape or bypass a combat encounter is not treated as a win. The results of these individual challenges fall into one of the "Sure Win", "Likely Win", "Toss Up", "Likely Loss", or "Sure Loss" categories, and are often accompanied by a brief explanation to support that result. These are then totaled up, with each "Sure" result generally counted as twice as valuable as each "Likely" result to get a feel for the win / loss ratio of the character class or option.
Unfortunately, there is a gray area in this test. You have the option of testing a fully written up character with specific feats and equipment or testing a shell of a character with only a class, class features, and the basic bonus providing equipment we would expect them to have at their level. A fully written character, especially one who is reasonably optimized against monsters who are almost universally not optimized and generally do not get the benefit of equipment, will always give you better results than testing the character who is just a class. The reverse is also true, in that a character without specific feats and equipment will perform less well than a character with these things against monsters who do have them, even if they aren't optimized.
The results from a less complete character are less accurate, but can be generalized to more builds and actual games than a more specified character. If you are trying to gauge the power of a class rather than a specific build, the less specified build is often more useful because of that reason. It will under perform on the test because it lacks options it would normally have, but if you expect that you can balance appropriately.
A Level 5 Same Game Test[edit source]
- A locked door behind an arbitrarily high number of assorted CR 4 traps.
- A huge Animated iron statue in a throne room.
- A Basilisk in its desert burrow.
- A Large Fire Elemental in a mystic forge.
- A Manticore on the wing above a plain.
- A Phase Spider anywhere. They're tricky creatures like that.
- A couple of Centaur Archers in a light to medium wood.
- A Howler/Allip tag team in an abandoned temple to a dark god.
- A Grimlock assault team (4 members) hidden in a cavern.
- A Cleric of Hextor (with his dozen zombies) in a crypt.
A Level 10 Same Game Test[edit source]
- A hallway filled with magical runes.
- A Fire Giant.
- A Young Blue Dragon.
- A Bebilith.
- A Vrock.
- A tag team of Mind Flayers.
- An Evil Necromancer.
- 6 Trolls.
- 12 Shadows.
A Level 15 Same Game Test[edit source]
- A Marut.
- A Hullathoin (with its army of skeletons and bloodfiend locusts).
- A Nightmare Beast deep in a hedge maze.
- A Windghost in the sky.
- A Yakfolk cleric with a party of Dao.
- A Drow Priestess with an army of ghouls.
- A warparty of Cloud Giants.
- A Mature Adult White Dragon.
- A Death Slaad riding a Titanic Toad.
- A Cornugon.
- A Gelugon and his Iron Golem bodyguard.
- A Rube Goldberg series of contingent weirds triggered to a set of symbols of pain surrounding the artifact.
- A pair of Glabrezus
- A harem of Succubi.
- Twenty Dire Bears.
- A dozen Medusa archers mounted on Hellcats.
- A forest made out of lava and infested with hostile fire-element dire badgers.
- A pair of Beholders.
Using Your Results[edit source]
For classes, it is fairly easy to use the SGT to assign a balance level.
- Monk Balance Level - If a class wins less than 50% of the level 5 SGT, it is very likely a monk level class.
- Fighter Balance Level - If a class wins around 50% of the level 5 SGT but less than 50% of the level 10 SGT, it is very likely a fighter level class.
- Rogue Balance Level - If wins around 50% of the level 5 and 10 SGTs, and does not fall behind significantly on the level 15 SGT, it is very likely a rogue level class.
- Wizard Balance Level - If a class wins significantly more than 50% of the level 10 and 15 SGTs, it is very likely a wizard level class.
Also remember that we expect characters with specific, and thus likely optimized, feats and gear from character levels to perform better than a simple shell character with only class abilities and unspecified equipment. Such a character could perform a bit better than the percentages listed here without being above the level. Similarly, a character who is just a shell of class features and vague level appropriate bonus items might perform a bit below the percentages listed here without being below the level. The percentages listed here are not absolutes in anyway, and the expected deviations from them depend on what method of test you use.
For other character options, it is much harder to use the SGT to gauge balance. A test is often run with the option and with a different option of the same balance level, and the results then compared to make sure they still line up appropriately. If they do, the option is likely at the same level as the option it was compared against, but this is much more shaky ground than class testing.