We Built a Conviction Score Engine — And It Immediately Flagged 2 Positions

Root Reborn is coming. In roughly two weeks, the way Bittensor allocates emissions will fundamentally change. Validators will decide which subnets deserve emissions and which ones get cut. If you're staked in the wrong subnet when that happens, you could lose a significant portion of your position.

We needed a system to tell us — objectively, with data — which subnets are real and which ones are coasting. So we built one.

What is the Conviction Score?

The Conviction Score rates every subnet on Bittensor from 0 to 100 across 8 signals:

• GitHub Activity — Is the team actively shipping code, or has the repo gone quiet?
• Emission Status — Are emissions active, or has the OTF already cut them?
• Price Momentum — Is the alpha price trending up or bleeding out?
• Validator Confidence — How much TAO have validators staked behind this subnet?
• Buy/Sell Pressure — Are people accumulating or dumping?
• Development Shipping — Has the team released anything meaningful recently?
• Risk Signals — Drawdowns, zero volume, high concentration?
• Network Importance — Is this subnet structurally important to Bittensor?

Each subnet gets scored, rated (Strong Buy / Buy / Hold / Watch / Cautious / Avoid), and analyzed by our Conviction Agent in plain English.

What We Found

Out of 128 subnets on Bittensor:

• 6 scored Strong Buy — active development, emissions on, positive momentum
• 24 scored Buy — solid fundamentals with manageable risks
• 63 have emissions already cut — nearly half the network is at risk
• 19 scored Avoid — multiple red flags, high risk of further decline

How It Changed Our Portfolio

The moment we ran the Conviction Score on our own portfolio, it flagged two positions:

SN95 (Actual) — Emissions CUT by the OTF. Despite being up 35% on the month from hype, the on-chain data told a different story. We sold.

SN118 (Ditto) — Emissions CUT. Influencers were bullish on this one. The Conviction Score wasn't. We sold.

Both positions were converted to TAO and held as cash — ready to redeploy after Root Reborn into confirmed winners.

The Lesson

We almost made the opposite mistake. When we first looked at SN9 (iota) and SN44 (Score), the low earner count made them look dead. One earner? Must be a ghost subnet.

Wrong.

• SN9 (iota) is Macrocosmos — they published a research paper on distributed AI training and launched a consumer app. Low earner count means the bar is incredibly high.
• SN44 (Score) published results showing their 19MB vision model beat GPT-4o, Gemini, Claude, and Grok at object detection.

Both scored 76-77 out of 100. Strong Buy. The Conviction Score caught what vibes-based analysis would have missed.

The Conviction Agent

Every subnet includes an AI-generated analysis explaining in plain English what the data means and what to consider doing. You can also ask questions directly — "Which subnets should I avoid?" or "Compare SN9 vs SN44."

Why This Matters Now

Root Reborn will redirect emissions from underperforming subnets to performing ones. 63 subnets already have emissions cut. That number could grow.

The Conviction Score tracks all of it, daily at 6 AM ET.

Try It Free

The Subnet Intelligence page is live at subnetaiq.io/subnet-intel. Every subnet scored, every score explained.

Not financial advice. Always do your own research.