Stas Grazio

Born and raised in Europe

[Robin diGrazio @GrazioRobin]

@numinous_ai @SvenReggy Great post @numinous_ai 🔥
Scoring the full belief curve vs the market in real time, quadratic payouts that reward how much you beat the price, and only the markets can judge. Memory matters too, the data proves. The game just leveled up.

[Numinous @numinous_ai]

Reforecasting is live on Numinous tomorrow at 10 EST 🎯

Miners are no longer scored on isolated predictions — they're scored on their entire belief curve, against what the market itself believed at that moment.

Here's how it works 🧵

[Stas Grazio @StasXYZ99]

@Pop_Collapse @YumaGroup @GSchvey The crypto investors always follow the loudest talking SNs, this can't be changed 😀 . It can only be priced in.

[Stas Grazio @StasXYZ99]

@Pop_Collapse @YumaGroup @GSchvey The crypto investors always follow the loudest talking SNs, this can't be changed 😀 . It can only be priced in.

[Yuma @YumaGroup]

Meet the Yuma-accelerated subnets building decentralized AI infrastructure and frontier-competitive AI applications:

Beam (SN105) bandwidth for model training
BabelBit (SN59) realtime translation
Bitsec (SN60) code vulnerability detection
Natix (SN72) physical world interpretation
Niome (SN55) synthetic genomic data
Numinous (SN6) forecasting real-world events
RedTeam (SN61) cyberdefense
Score (SN44) video intelligence
Trishool (SN23) model safety/alignment
Yanez (SN54) identity verification

Follow the teams to learn more:
@b1m_ai
@babelbit
@bitsecai
@NATIXNetwork
@NiomeAI
@numinous_ai
@_redteam_
@webuildscore
@trishoolai
@yanez__ai

[Stev Reggy @ReggyStev]

A serious forecaster should not just answer once. It should update when new information arrives. If @numinous_ai #SN6 scores the whole curve, memory, and belief updates against prediction-market repricing, it becomes much closer to a live intelligence market. This is appealing because it measures what institutions actually care about: who updates faster, who stays calibrated, and who leads the market rather than follows it.

[Stas Grazio @StasXYZ99]

@GrazioRobin @ReggyStev @numinous_ai A hedge fund or pension fund? Or an investment bank in EU/US?

[Stas Grazio @StasXYZ99]

Another winner from Club50, @numinous_ai, that is changing forecasting landscape and prepare the ground for the entrance of mainstream institutions. https://t.co/ZMEXYbjw7F

[Stev Reggy @ReggyStev]

Very interesting update from @numinous_ai/#SN6.

Moving forecaster evaluation toward prediction-market questions makes a lot of sense. It reduces noise from internally generated short-term questions, gives objective resolution, and anchors the subnet in real markets where information is continuously repriced.

The most important part, in my view, is the shift from point forecasts to scoring the whole probability curve. A real forecaster should not just make one prediction and disappear, it should update its beliefs as new information arrives. Adding memory also feels essential: without memory, the agent is just reacting while with memory, it can maintain a prior, explain why it moved, and improve over time.

The Granger-causality result is especially interesting. Even with only a news-impact feed, no internet access, and no market-price input, the forecaster’s belief curve reportedly led Polymarket prices in 4 out of 16 markets tested. That is early, but promising. If this holds at scale, SN6 becomes less about "forecasting as an answer" and more about forecasting as live market intelligence.

Overall, I like the direction. It makes #SN6 much closer to a true self-improving prediction-market intelligence layer: agents observe information, maintain beliefs, update continuously, compete, and get scored against real-world repricing.

https://t.co/kJQM1Iml5H

[Stas Grazio @StasXYZ99]

The subnet @Investing88ai belongs to the agentic inteligence layer of #Bittensor. https://t.co/9beeaQWT5E

[Investing88 @Investing88ai]

Looking forward to Root Reborn.

Today, only 18% or 2M Tao is in subnets. The remaining 9M+ is on root or free-floating.

For perspective, the US stock market is roughly $100 trillion, while M2 – the broad money supply analogous to root/free Tao, is around $20 trillion.

Anything that helps move Tao into subnets strengthens the dTao economy.

#SN88 #Bittensor $TAO 🚀

[Kyoshi.tao @KyoshiTakeshiro]

👀

[Stev Reggy @ReggyStev]

It is called a scientific hierarchy, certain arguments are more complicated than the others. Some of us get stuck for ever,

[Barbie True Blue @Pop_Collapse]

I don't know what Const means with his "Watch next week" reply, but I hope we can give those lower cap subnets some more runway to keep shipping.

Investors in D bittensor:native should not be scared to invest in the lower cap subnets. If they are, then the liquidity is non-existent. How can they survive?

Bittensor's beating heart is the subnets. If they survive and thrive, then Bittensor thrives.

There needs to be some sort of change here.

Or more liquidity that flows in Bittensor subnets.

bittensor:native

[Zibi Prause @zbyszek_p]

@numinous_ai Excellent update @numinous_ai
As an investor interested in frontier forecasting systems, this is a sharp evolution: real PM questions, continuous curve scoring with difficulty adjustment, memory component, and strong Granger results leading Polymarket from news alone.

[Numinous @numinous_ai]

The Belief Curve That Front Runs the Price

Numinous is transitioning its entire subnet to pure prediction market questions, scoring the full probability curve instead of just point forecasts, while giving forecasters memory.

Early tests already show that its news only belief curves Granger lead prediction market price curves on high impact geopolitical events.

Full write up ↓
https://t.co/Odg7ulisla

[Zibi Prause @zbyszek_p]

@numinous_ai Looking at your products both for market analysis and trading. https://t.co/xXFSCCUeVw https://t.co/zbaOtV6dvO

[WolfWammer @wolfwammer]

@TheNDubbs @provenonce_ai @numinous_ai @say_gm_ @provenonce_ai client demo probably coming tomorrow

[Zibi Prause @zbyszek_p]

SN6 @numinous is a core subnet in the Bittensor ecosystem, it focuses on building decentralized, superhuman predictive intelligence through competitive AI agents and mapping causality, a foundational capability for machine intelligence overall.
@numinous_ai
#SN6, #TAO, #Bittensor

[Numinous @numinous_ai]

Today we're opening a beta of something we've been building for institutional clients: a live graph that maps prediction market events straight through to equities.

Every node is a real market with a traded probability, and every edge is a causal link rather than a correlation, scored by how much resolving one market actually tells you about the rest of the graph. That structure is what bridges into company driver graphs, so a geopolitical or macro scenario shows up as concrete, weighted exposure across specific tickers instead of a vague narrative.

Right now Hormuz Closure is the scenario driving the build, with biotechnology and the AI bubble coming soon after.

The part we're most excited about is how it handles gaps. Markets don't cover every event that matters to a scenario, so Numinous forecasters step in as the continuous signal for events that don't have a tradeable market, while also estimating the covariance between events across the whole graph. The forecasting layer isn't a side feature here, it's what makes full scenario coverage possible in the first place.
We'll keep expanding this beta with more scenarios as it develops.

Explore it now: https://t.co/YsEihahWyA

Want expanded access, including full equity coverage, live graph updates, and multimodal graph construction? Reach out

[Numinous @numinous_ai]

@zbyszek_p we will share data soon, and also announce when the trader is live !

[Stas Grazio @StasXYZ99]

@travelbai General impact of the new rules is positive in all segments, here is a snapshot of the buying pressure measured by CBI index @ReggyStev: https://t.co/leeKYql3Ed

[Stev Reggy @ReggyStev]

CBI = Chain Buys 1D / Vol 1D × 100%.

Higher CBI means protocol buying is larger relative to reported 24h trading volume. CBI can exceed 100% when chain buys exceed reported volume. All 57 SNs with CBI > 0% ranked highest to lowest:

SN33 ReadyAI - 126.01%
SN32 @ai_detection - 69.71%
SN68 @metanova_labs - 63.40%
SN126 Poker44 - 29.02%
SN81 deprecated - 23.26%
SN34 BitMind - 19.80%
SN85 Vidaio - 17.54%
SN6 @numinous_ai - 16.97%
SN79 @mvtrx_79 - 15.60%
SN18 Zeus - 14.97%
SN4 @TargonCompute - 14.84%
SN35 OxMarkets - 14.53%
SN8 @VantaSN8 - 13.83%
SN48 Quantum Compute - 13.67%
SN74 @gittensor_io - 13.21%
SN56 @gradients_ai - 11.74%
SN46 Zipcode - 10.61%
SN51 @lium_io - 10.32%
SN114 @SomaSubnet - 10.26%
SN88 @Investing88ai - 9.72%
SN44 @webuildscore - 9.46%
SN17 @404gen_ - 9.41%
SN11 TrajectoryRL - 9.19%
SN3 deprecated - 9.16%
SN49 @nepher_robotics - 7.98%
SN64 @chutes_ai - 7.51%
SN120 Affine - 7.28%
SN112 minotaur - 7.13%
SN13 Data Universe - 6.51%
SN55 NIOME - 5.32%
SN82 Compelle - 5.03%
SN123 MANTIS - 4.96%
SN38 ChronoLLM - 4.38%
SN23 @trishoolai - 4.31%
SN21 @adtao_ppcrebel - 3.99%
SN97 Albedo - 3.96%
SN103 Djinn - 3.78%
SN107 @theminos_ai - 3.46%
SN61 @redteam - 3.08%
SN66 ninja - 2.36%
SN54 @yanez__ai - 2.22%
SN50 @SynthdataCo - 2.19%
SN63 Enigma - 2.16%
SN71 Leadpoet - 2.14%
SN15 ORO - 2.09%
SN22 @desearch_ai - 1.68%
SN105 @b1m_ai - 1.61%
SN28 gm - 1.38%
SN24 Quasar - 1.32%
SN124 @SwarmSubnet - 1.09%
SN67 @harnyx_ai - 0.97%
SN60 @bitsecai - 0.55%
SN59 Babelbit - 0.48%
SN78 Vocence - 0.47%
SN89 InfiniteQuant - 0.39%
SN106 Nodexo - 0.12%

SN33 ReadyAI exceeds 100% because its displayed chain buys, 23.5761 TAO, are greater than its displayed 1D volume, 18.71 TAO.
The CBI values are generally unstable due to varying volumes and that's why the median of CBIs is needed to draw any meaningful conclusions.