How the Gaming Token Index Works
The Gaming Token Index is a family of three market-cap-weighted benchmarks (base 100 on 1 January 2024) that track pure-play gaming tokens: a focused GTI-20 (top 20), a broader GTI-100 (top 100), and a full Composite. All three share the same eligibility and maintenance rules. They differ in how many names they hold and whether the weights carry a cap.
- Three indexes
- Top 20 · Top 100 · Composite
- Market-cap weighted
- 10% cap (GTI-20/100)
- Real market cap only
- Monthly rebalance
- Base 100 · Jan 2024
Overview
Each index answers one question: are gaming tokens, as a group, going up or down? They pick a clean set of pure-play gaming tokens with a market cap built from real data, weight them by market cap, and track their combined value against a base of 100 (set on 1 January 2024). The three differ in breadth and capping. Everything else is identical: eligibility, the real-data rule, monthly rebalancing, and the divisor that keeps the line continuous.
The three indexes
- GTI-20
- The top 20 eligible gaming tokens by market cap, each capped at 10%. A focused, large-cap read on the sector.
- GTI-100
- The top 100 eligible gaming tokens by market cap, each capped at 10%. Broader coverage with the same cap.
- Composite
- Every eligible gaming token, uncapped. Names weigh by their raw market-cap share, like the Nasdaq Composite or CoinMarketCap’s indices.
GTI-20 and GTI-100 always hold their full count. If a member drops out mid-month (see the data-integrity rule below), the next-ranked eligible reserve takes its place, so the line-up stays at 20 / 100 between rebalances. The Composite holds whatever is eligible.
Which tokens qualify
Candidates come from CoinGecko’s gaming-related category tags. To count as a gaming token, a coin must match at least two gaming categories, or belong to one core gaming category. A broad “Metaverse” tag on its own won’t do it, which keeps general AI, DePIN, and infrastructure tokens (RENDER, for example) out.
Eligibility & data integrity
These three take the raw top of the gaming market. There is no market-cap floor, no volume screen, no wash-trading exclusion. Every constituent must clear one bar instead, a market cap we can trust:
- Real market cap only
- At each rebalance a token must carry a real reported market cap from the data source, not one we estimate from price × last-known supply. A token with no known circulating supply has no market cap we can trust, so it stays out.
- Data-loss rule
- The real-data check runs every close, not only at rebalance. When CoinGecko stops reporting a member’s market cap, that name leaves the basket after 3 straight closes without real data, and its estimated value stops counting toward the level and weights. It rejoins after 3 straight real closes. The short streak each way avoids flicker from one-off gaps.
- Backfill (GTI-20/100)
- When a fixed-size index loses a name to the data-loss rule, the highest-ranked reserve frozen at the last rebalance takes its place, holding the count at 20 / 100. The Composite has no fixed size, so it shrinks by one.
- Manual overrides
- A small allow-list and deny-list handle edge cases where the automated rules produce wrong results: a legitimate gaming token miscategorised by the data source, or an obvious non-game that slipped through.
Weighting & the cap
All three weight by market cap. They differ on capping:
- GTI-20 & GTI-100 apply a 10% single-name cap: no token tops 10% of the index, and the excess spreads across the rest. When a winner drifts past 15% between rebalances, we re-cap the basket to 10% that day.
- The Composite runs uncapped: each token weighs by its full market-cap share, so the largest names dominate, as a broad market-cap index should.
Rebalancing & the divisor
We review all three monthly, not every second, so a token spiking for one day won’t push it in and out. Between rebalances the line-up changes only when the data-integrity rule drops a name (and, for GTI-20/100, a reserve backfills it), or when a 15% drift re-cap fires. In each of those cases the divisor absorbs the change so the level holds steady, as the rest of this section shows.
There’s an obvious worry. If a brand-new token joins, won’t the index jump just because there’s now “more” in it? And drop when one leaves? Sum everyone’s value and yes, it would. That misleads, because the index tracks whether gaming tokens are going up or down in price, not how many happen to be on the list.
To prevent that, the index uses a behind-the-scenes scaling number, the divisor. The index value is the combined market cap of its tokens divided by that divisor. On a normal day the divisor stays fixed, so the index moves only when prices move. The moment the line-up changes, the divisor re-tunes so the index reads the same right before and right after. Joining or leaving, on its own, moves the index by zero. Only real price changes move it from there.
Picture a class’s average grade. A straight-A student transfers in and the average jumps, but not because the other students studied harder. So you re-baseline the moment they join, and from then on the average changes only when students’ grades change. The divisor does that re-baselining for the index.
This isn’t just theory; it’s how the index behaves. To test it we added Bitcoin, roughly 300× larger than the entire gaming index. A naive index would have exploded; this one stayed put, because the divisor grew to absorb Bitcoin’s size. Removing Bitcoin returned everything to normal: the divisor shrank back and the index read the same as before the test.
The same chain-linking handles every other membership event: a monthly rebalance, a data-loss drop, a reserve backfill, or a 15% drift re-cap.
Pricing
Each index tracks its constituents’ daily closes, anchored to the 00:00 Europe/Sofia tick, and that same daily-close market cap drives eligibility and weighting. Intraday swings don’t move an index until the next day’s close.
Limitations
- Selection leans on the data source’s category tags, so a miscategorised token can land in or out by mistake. The manual override list patches the worst cases.
- Weights use total market cap, not free-float, so locked or vested supply counts the same as circulating supply.
- The data-loss rule leans on CoinGecko’s coverage: when it stops reporting a real market cap for a name, that name leaves the index until coverage returns, even while the token still trades.
- The Composite’s long tail of micro-caps carries near-zero weight, so the larger names move nearly all of the level.