How to Create a Custom Index From Scratch

Creating a custom index from scratch means writing a rulebook that turns your own choice of securities into a single, trackable index level, then backtesting and following that level over time. A custom index is not a random watchlist; it is a defined set of rules for what goes in, how much of each you hold, and when you reset the weights. Once those rules exist, the index becomes measurable, repeatable, and comparable to a benchmark. This guide walks through the creation process end to end. It is educational and is not investment advice.

Start with a clear purpose

Every good index answers a question. Before picking names, decide what your index is meant to express. Is it "the twenty largest US banks," "a basket of AI infrastructure companies," "an equal-weighted slice of dividend payers," or "the top crypto assets outside BTC"? A sharp purpose makes every later decision easier, because you can ask whether each rule serves that purpose. A vague purpose produces a vague index that is hard to interpret later.

Define the universe

Turn the purpose into a concrete list of tickers with explicit inclusion rules. If the theme is an idea rather than a fixed set, write down what qualifies a security for membership, for example "publicly traded, primary listing in the US, revenue mostly from cloud infrastructure." Clear rules mean the index can be maintained consistently and re-derived later. For turning a theme into names, see thematic investing.

Choose how to weight the members

Weighting decides each member's influence on the level. The main options and their character:

• Market-cap weighted. Bigger companies get bigger weights. Low maintenance, but can become concentrated in a few giants.
• Equal weighted. Every member counts the same. More diversified, but needs regular rebalancing to stay equal.
• Capped. Market-cap logic with a ceiling, such as "no name above 10 percent," so no single company dominates.
• Custom. You assign the weights directly to express a specific view.

You can experiment here with different weighting schemes and compare how they behave. For the trade-offs between the two most common ones, see market cap versus equal weight.

Set rebalancing rules

Prices drift the weights away from target, so decide how and when to reset them. A common approach is a fixed schedule (say quarterly) optionally paired with a tolerance band ("only rebalance a name if it drifts more than 5 percent from target"). Tighter rules keep the index closer to its intent but imply more turnover. Our overview of index rebalancing covers the choices.

Set a base value

An index is quoted as a level, not a dollar sum. Pick a base date and a base value, commonly 100. From there the index moves with the weighted basket. If the basket rises 9 percent, the level goes from 100 to 109. The base value makes the index easy to read and directly comparable to a benchmark on the same scale.

Backtest the rulebook

Run the completed rules backward over history. A backtest applies your universe, weighting, and rebalancing to past prices and produces a hypothetical track record. This tells you how the index behaved: its volatility, its worst drawdown, and how it moved against a benchmark. Two things to keep front of mind. A backtest is hypothetical historical performance, and past performance does not guarantee future results. And you should resist tuning the rules until the past looks perfect, which is overfitting; see backtesting mistakes to avoid.

Track it against a benchmark

With the index live, follow the level over time and compare it to a fair reference such as the S&P 500 or, for crypto, BTC. Benchmark comparison tells you whether the index earned its complexity rather than just riding the general market. Watch the relationship, not only the absolute number.

A compact example

That is a complete custom index. You can now backtest it, track its level, and compare it to a benchmark, then refine as you learn.

If you want to create and follow a custom index without spreadsheets, Indexes lets you build a weighted basket of stocks or crypto, backtest it against real history, and track its level against a benchmark. It is built for learning and measurement, not trading, and nothing here is investment advice, so treat any index you create as an educational study of your own rules.