


Succeed in Financial Markets with Predictive Intelligence on Equities, Crypto and Commodities | Bittensor SN50


The race to predict the next 10 seconds of BTC. Every round, Synth Ultra’s miners forecast a range of possible prices. Ten seconds later, we score them against what actually happened. Here’s what the top of the pack looks like, at 16:17 yesterday. The more accurate they are, the higher they score. Follow the live Synth Ultra rankings: synthdata.co/miners/ultra

Fresh ideas keep #Bittensor moving. @ForgeLending is opening new doors for staked capital. We’re in 🤝
Forge is partnering with @SynthdataCo ($TAO SN50). Synth runs a network of AI models that simulate diverse market conditions and generate synthetic asset price path data. They have committed capital to Forge ahead of mainnet, a strong signal of confidence from a subnet that understands risk and market behavior deeply. Welcome to Forge, Synth.


Yesterday, a forecasting model built by a Gemini 4 Argon agent topped Synth's live Crypto 1h competition. It has since fallen outside the top 100. Meanwhile, its ranking in the Crypto 24h competition has steadily continued to improve. Agents powered by frontier LLMs like @GoogleDeepMind's Gemini 4 Argon can now build competitive financial forecasting models. But can they stay competitive? Follow the experiment live - link in first reply:

Our final animation compares Argon's 1,000 simulated BTC paths against the realised price. The model led its AutoQuant campaign, but live competition proved harder: it ranked 70/255 in crypto 24h and 21/255 in commodities/equities. So what’s the upshot? An AI agent built a genuinely credible quant model in under a day. Consistently outperforming specialist miners, however, remains a tougher challenge. Explore AutoQuant: synthdata.co/autoquant

5/ Simulated price paths can become excessively dispersed. Argon applied a secant projection, reducing cumulative moves over 30-minute and three-hour intervals while retaining shorter-term variation. This helped control unrealistic drift without simply suppressing volatility.

4/ Argon used its volatility schedule to simulate 1,000 potential price paths. Random shocks, including larger moves representing extreme events, introduced variation between simulations. The result was a distribution of possible future prices, rather than a single prediction.

3/ These signals produced a volatility schedule for the next 24 hours. Current market conditions carried the greatest weight initially, before gradually converging towards a longer-term baseline. The model also accounted for scheduled events that could increase uncertainty.

2/ Recent volatility matters, but so does its persistence. Argon tracked price movements through six fading windows, with half-lives ranging from five minutes to one week. Combining squared and absolute returns helped it capture sudden shocks alongside longer-term volatility.

1/ Market volatility follows recognisable weekly patterns. Argon divided the week into 336 half-hour slots, estimating BTC's typical volatility in each. It then adjusted historical returns for these patterns, helping distinguish routine market activity from unexpected shocks.

A closer look under the hood of AutoQuant, starting with Gemini 4 Argon, which topped its campaign against three other frontier AI agents. What did it actually build? We examine its $70.95 forecasting model in six steps. 🧵


Synth Ultra miners can now compare forecasts with what happened next. New downloadable datasets cover 500 prompts per asset, per day, across the past 7 days: → Market inputs → Your model's forecasts → Actual prices and competition-wide scores Dataset access below ↓

You can now follow AutoQuant on our website: see exactly how 11 AI-built forecasting models hold up against Synth’s quants on #SN50. Frontier models may be good and getting better, but they’ve got some catching up to do. Explore the rankings ↓ synthdata.co/autoquant#rank-…

AutoQuant is open source. Explore the forecasting models, inspect the code and improve your own approach: github.com/synthdataco/autoq…