Bittensor Subnet 46 RESI Deep Dive – Every Bittensor Subnet Gets Exploited

In this video, TAO Templar (Travis) does a deep dive into Subnet 46: Resi, walking through what’s working, what’s still rough, and the typical messy path a young Bittensor subnet takes while refining its incentive mechanism.

A useful framing point he opens with: every subnet gets exploited at some point. The question is how the team responds, not whether it happens.

What Resi is building

• Resi is building real estate appraisal infrastructure on Bittensor. Miners train AI models, validators check those models against recently sold properties, and the goal is highly accurate property valuations.
• The long-term vision is bigger: accurate appraisals unlock RWA tokenization, letting users tokenize their property and take quick loans against it. Lending is targeted for end of year.
• The subnet is about 9 months old and backed by DSV (Siam Kidd, Mark Creaser) and Stillcore Capital (Mark Jeffrey, Rob Greer).

Their new dashboard & the appraisal portal

• The dashboard claims Resi is hitting ~98% accuracy vs Zillow’s estimates, but the top miner on the leaderboard is only at 94%. Travis flags this discrepancy as something worth clarifying.
• He warns about a common pattern: cherry-picked accuracy streaks. A single miner can hit 99% five times in a row without the overall model being that good.
• His suggestion: prove appraisals with historical timestamps and Wayback Machine links, showing that Resi’s predictions don’t change after the fact, while Zillow’s estimates quietly do.
• The appraisal portal is live with a small number of paying users. You can try it free with an account. Paste in any property and the current best model returns an appraisal you can edit and resubmit.

Reading the on-chain signals (TAO Stats)

• The subnet is currently at 100% miner burn — no miners are being paid right now. On the surface that looks like a red flag.
• But Travis reads it as healthy behavior: emissions have been toggled on, broken, fixed, turned off, turned on again. That’s exactly what a subnet owner should do when the mechanism is exploited — stop paying miners to exploit you while you patch it.
• No buybacks or manual burns yet, which makes sense for a pre-revenue subnet still refining its core mechanism. He’d like to see consistent buybacks once the mechanism is leak-proof.
• Backing this up: lots of code commits per day, active Discord and X presence, and Bitcast campaigns getting the name out — far better communication than most subnets.

Discord & community signals

• A pinned message from about a week ago confirms their incentive mechanism was recently exploited and a patch is in progress. Normal Bittensor subnet life.
• All-time revenue is $15,000, mostly generated when they were selling predictive data — small, but it exists.
• Subnet owner Sevi is very responsive and has a clear long-term vision. Travis calls this out as refreshing compared to “moai-silent” subnets.
• He recommends following Rado (Trend Setter Capital on YouTube) — a former real estate agent who got so absorbed by Bittensor that he learned to code and became a Resi miner. His domain expertise makes him a useful voice on whether appraisals are being done properly.

Bottom line

• Resi is actively developed, well-communicated, backed by serious investors, and run by someone with a clear vision.
• The open question is still whether they can build a leak-proof incentive mechanism that consistently produces accurate appraisals — and once they do, whether they can drive real revenue to start burning supply.
• No major red flags. Travis is rooting for them, because if the lending thesis plays out, there’s a lot of money in what they’re building.

If you want Travis’s full walkthrough with the dashboards and Discord scrolling, the video is here: