We draw on our experience in institutional finance to build AI systems that take quantitative research from an initial idea through to autonomous trading.
These figures are average rolling 30-day counts for September 2026, with 165 tasks (less than 0.1%) requiring human intervention.
BanZhuan.AI
Evaluate price trends alongside SentiMove sentiment scores.
Build and test a strategy with explicit entry conditions.
Monitor funding, execution and portfolio risk.
Checking fees, funding and available liquidity before backtesting.
The 30/120-minute EMA strategy is connected. Waiting for execution checks.
Aligned BTC-USD prices with the bearish SentiMove sentiment signal.
Applied the $10,000 capital ceiling, 2× leverage cap and loss limits.
BTC-USD · Strategy development
| Study | Tool | Status |
|---|---|---|
| Sentiment alignmentCheck the timing and freshness of SentiMove scores. | evaluate_factors |
Completed |
| EMA parameter studyCompare crossover configurations with the sentiment gate. | run_backtest |
Running |
| Execution sensitivityAssess fees, funding and drawdown. | run_backtest |
Queued |
| Price data integrityCheck BTC-USD candles for gaps and duplicate timestamps. | evaluate_factors |
Completed |
| Sentiment coverageReview SentiMove availability across the research period. | evaluate_factors |
Completed |
| Signal persistenceStudy how sentiment strength changes after each update. | evaluate_factors |
Running |
| Market-regime robustnessCompare the selected setup in trending and ranging markets. | run_backtest |
Queued |
| Exposure reviewCheck holding periods and repeated entries against position limits. | run_backtest |
Queued |
| Out-of-sample validationEvaluate the selected setup on a separate, unseen period. | run_backtest |
Queued |
Compare OddSim outcome probabilities with market prices.
Test pricing edges after fees, slippage and exposure limits.
Monitor quotes, liquidity and event resolution conditions.
Reviewing liquidity, lineup changes and the match-resolution rules.
A 7 pp gross edge leaves 5 pp after costs, above the 3 pp minimum.
Compared OddSim’s 61% YES estimate with the market’s 54¢ ask.
Applied the $500 per-outcome limit and $1,000 daily allocation cap.
Dota 2 · Match-winner outcomes
| Study | Tool | Status |
|---|---|---|
| Probability calibrationCompare OddSim estimates with historical outcomes. | Calibration study |
Completed |
| Entry-cost sensitivityTest the effect of spread, fees and available liquidity. | Pricing study |
Running |
| Resolution reviewValidate settlement rules and lineup assumptions. | Market review |
Queued |
| Outcome mappingMatch teams and series identifiers to the correct YES outcome. | Market review |
Completed |
| Timestamp alignmentPair OddSim estimates with market quotes available at the same time. | Pricing study |
Completed |
| Roster freshnessCheck team lineups and stand-ins before reviewing model assumptions. | Market review |
Running |
| Series-format sensitivityReview probability calibration across different match lengths. | Calibration study |
Queued |
| Liquidity stressRecheck entry costs with thinner depth and wider spreads. | Pricing study |
Queued |
| Walk-forward validationAssess calibration on later matches excluded from model fitting. | Calibration study |
Queued |
Sentiment analytics for event-driven strategies. SentiMove combines data from SentiChain, decoded through Aparecium, with AI analysis to interpret the information behind market moves.
Launched in March 2025, it is evolving into a dedicated intelligence service for BanZhuan.AI, supporting research on Hyperliquid and, where relevant, Polymarket.
Agent-based simulation for pricing outcomes in MOBA games. Launched in September 2026, OddSim.gg studies how the behaviour of competing agents translates into outcome probabilities.
Integrated with BanZhuan.AI, OddSim.gg provides odds calculations for esports trading strategies on Polymarket.
An application-specific blockchain for sentiment data, launched in January 2025. It stores the embeddings that underpin SentiMove.
Our embedding reversal model, released in March 2025. It recovers text from embeddings for downstream analysis in SentiMove.
We are also incubating Beaker.gg, exploring trading through game experiences built on shared BanZhuan.AI infrastructure. Our design partnership with Bittenbin explores social games and online interaction.
Yototec began in Hong Kong in June 2018, developing quantitative models, event-driven strategies and trading infrastructure for hedge funds in the US and China. A focus on interpreting market language has guided our work ever since.
That research carried us into crypto markets and blockchain infrastructure, and through successive generations of language models, from recurrent networks and LSTMs to BERT, transformers and GPT. Agent systems bring these strands together.
In January 2026, we chose to conclude consulting and focus on our own platforms. BanZhuan.AI builds on the research and engineering experience accumulated through that institutional work.
Yototec has always been funded by its own operations, without external investment or government funding. Consulting supported our early development; today, proprietary trading through BanZhuan.AI supports the business.
Co-founder & Chief Executive Officer
Jenny’s background spans derivatives pricing, quantitative portfolios and AI-driven trading systems. She built machine-learning pricing models for securitised products at RBC Capital Markets and managed forex and rates quantitative portfolios at Complus Asset Management. She also worked as a Summer Fellow at KIMC.
She is an alumna of UCL and Cornell.
Co-founder & Chief Technology Officer
Edward’s background spans AI agent systems, quantitative engineering and decentralised market infrastructure. His fund work spans crypto with Time Research, US equities with Sheffield Asset Management and Chinese markets with Eastmoney. He has built trading agents, designed protocol infrastructure for the Autonity L1 blockchain and developed sentiment-driven strategies across these markets.
He is an alumnus of the University of Cambridge.
The business begins in Hong Kong, serving hedge funds focused on US equities and China A-shares.
Our focus shifts to crypto funds and blockchain foundations in Singapore and the UK, bringing our sentiment research to digital assets.
SentiChain development begins in July 2024 and launches in January 2025. Aparecium and SentiMove follow in March.
We conclude consulting and launch our agent-driven research and trading platform.
OddSim.gg launches, using agent-based models to calculate MOBA odds for trading research.
Headquartered in London with an office in Hong Kong, Yototec is led remotely by its co-founders in Irvine, California.
Company no. 15288135
86–90 Paul StreetBRN 71920954 · CR 2947411
Unit 1507B, 15/F, Eastcore