The research workflow
Five steps, in order. The first two are on this site, the next two are on the subnet’s own channels, and the last one is the calendar.
- Sort the directory by emission shareOpen the subnet directory and sort by emission. That column is the share of every block’s new TAO the subnet currently captures, and it is the single number that summarises how the rest of the network is voting with its capital. Note the direction as well as the level — a subnet moving from 2% to 4% is a different proposition from one that has sat at 8% for a year.
- Open the subnet detail pageThe detail page carries the numbers the directory row cannot: how many miners are actually registered and earning, how many independent validators are scoring them, how concentrated the staked alpha is across coldkeys, and how the alpha price has moved. Concentration is the one people skip and later regret.
- Read the scoring code on GitHubEvery subnet publishes a repository with its validator logic in it. That code is the contract: it states exactly what is measured, how quality and latency are weighted, and what a response has to look like to score. Ten minutes here tells you more than any dashboard — and if the repository is thin, stale, or undocumented, that is itself the finding.
- Read the room in DiscordSubnet repositories link a Discord. Operational problems, imminent scoring changes and owner responsiveness all surface there weeks before they show up on-chain. Lurking for an afternoon before you commit is free; discovering an unannounced mechanism change after you have bought hardware is not.
- Watch it for two weeks before committingOne good week is noise. Track emission share, miner count and alpha price over at least two weeks and look for the trend rather than the level. This is the step that separates a decision from a reaction, and it costs nothing but patience.
What you are looking for
The same page of numbers answers two different questions. Read whichever panel matches what you are about to commit.
Staking into it
- Emission share, and its slope. Rising share means capital is arriving. Falling share is the market voting against the subnet in real time.
- Pool depth. Your exit price comes from the same AMM curve your entry did. A thin pool is a good entry and a bad door.
- Stake concentration. If two or three coldkeys hold most of the alpha, their exit sets your exit price, and you will not be told in advance.
- Validator count. More independent validators means scoring that is harder for any one party to bend.
Mining on it
- Emission per miner. Emission share divided by active miners is the number that pays you, not emission share on its own.
- Hardware fit. The repository states the model and the latency budget. Under-spec hardware does not earn less — it scores near zero.
- Scoring clarity. A documented, readable scoring function is a target you can optimise against. An opaque one is a guess.
- Registration cost. Burned, not refunded. Price it as the cost of the experiment, and check it the day you register.
Red flags
Three findings should stop the process rather than adjust the price. Stake concentrated in a handful of coldkeys, because it makes your exit somebody else’s decision. Fewer than about five active miners, because a subnet earning emissions with almost no competition is usually earning them through validator relationships rather than through work. And an emission share that has fallen for several consecutive weeks, because that is the market completing its own research ahead of you.
Start with one
Miners should run one subnet until they are in the top half of its ranking. The learning curve is per-subnet — the scoring function, the configuration, the failure modes — and spreading across three before mastering one reliably produces mediocre scores everywhere. Expand once you have operational confidence, not once you have spare hardware.
Stakers face the opposite failure. A single subnet is a concentrated bet on one team’s execution, so anchor most of the position in two or three established subnets and keep the higher-conviction, earlier-stage ones small. Holding several positions is also how you learn what normal emission and price behaviour looks like, which is what makes an anomaly on any one of them visible.
Ready to compare? Open the subnet directory → or price a stake in the trade workspace →