Notes & Methodology

What this is

A slotting statistics viewer derived from 14,982 complete builds posted on the Homecoming forums. For each power, it shows how commonly it is taken, how many slots players typically put in it, which enhancement sets dominate, and which layouts are most popular. The goal is to give a quick, data-grounded answer to "what do people actually put in this power?"

Data sources

Builds were extracted from a snapshot of the Homecoming forum archive covering posts up to early 2026. Three inline build formats are recognised - MBD (Mids' binary JSON), MxDz (Mids' compressed binary), and Mids' plain-text export in both its compact and long-form variants, as well as .mbd and .mxd files posted as forum attachments. Counts by format for builds included in stats:

• text: 8,101
• MxDz: 5,014
• MBD: 1,867

Quality filters

Two filters are applied before a build contributes to any statistic:

• Minimum 12 slotted powers. The build-count distribution is strongly bimodal: complete builds cluster at 22–31 slotted powers; stubs and help-skeleton posts form a thin tail below about 15. The 12-power floor sits cleanly in the gap and removes 268 partial builds from stats without touching real ones.
• Maximum 50% unresolved enhancements. Builds exported from an outdated version of Mids sometimes reference enhancement IDs that no longer exist in the current database. If more than half of a build's slots are unresolvable it is excluded, as the slotting signal is too degraded to be useful.

Excluded builds are retained in the database and visible in the raw counts but do not affect any displayed statistic.

Deduplication

Builds are deduplicated by a signature derived from the archetype, powerset selection, and the sorted enhancement tuple for each power. The same build posted in multiple threads, or quoted and re-posted by another user, counts once. Iterative revisions of a build by the same author produce distinct signatures and count separately, which is the intended behaviour.

On data quality and sample bias

A common objection to forum-scraped build data is that it will be dominated by bad or incomplete builds. In practice the opposite pressure applies: people post builds publicly to show off a finished project, ask for critique on something they put real effort into, or reply to a request for examples. Placeholder skeletons exist but are removed by the minimum-power filter. The median build in this dataset has 26 slotted powers and resolves 99%+ of its enhancement tokens cleanly.

Sample bias does exist, but it runs in a more specific direction: the forum population skews toward players engaged enough with the game to seek out build discussion. Casual players who never visit the forums are not represented. This means the data likely over-represents IO-slotted builds relative to the player base as a whole, and under-represents early-levelling or TO/SO slotting. For the purpose of understanding how experienced players approach a power or powerset, this is the right population to sample from.

The data is also necessarily backward-looking. A build posted in 2021 reflects the meta and available sets of 2021. Enhancement sets added or rebalanced since then will be under-represented relative to their current prevalence.

Enhancement counting

Only named IO set pieces and generic Invention-origin IOs are counted. Training Origin, Dual Origin, and Single Origin enhancements are ignored. Serious builds at level 50 do not use them, and their occasional appearance in build exports is treated as noise rather than signal. The "Superior" tier of archetype-specific purple sets (e.g. Superior Defiant Barrage) is folded into its base set for counting purposes, since players use one or the other and splitting them would halve the sample size for each.

Credits

Archetype and powerset icons by Rylas (Homecoming forums).