GET
get-market-data
Pull the public tape. Close the candle. Reject anything stale, incomplete, or out of sequence.
- CCXT
- Binance US
- SQS out
QNTGY//RESEARCH UNIT · EST. 2024
Alpha,
wherever
it hides.
Immaterial algorithmic search across crypto, sports, and every other market willing to leave a trail of data.
LIVE SYSTEM TRACE PAPER / ARMED
21:00:00.004 EVENTBRIDGE tick/hour
21:00:00.119 GET_MARKET_DATA binance_usdt
21:00:02.884 SQS candles: validated
21:00:03.021 ANALYZE_MARKET_DATA signals: 03
21:00:03.506 RISK_GATE mode: paper
21:00:03.509 EXECUTE_TRADES waiting_
01INGESTMarket data
02ANALYZETestable signals
03CONTROLRisk first
01 / The machine
Three Lambdas.
One opinion.
The machinery is public. Every idea leaves a trace from raw candle to rejected—or accepted—trade.
ASK
analyze-market-data
Make the strategies disagree in public. Preserve the signal, the price, and the reason.
- MACD
- RSI
- Patterns
ACT
execute-quantegy-trades
Assume every message is dangerous. Check limits, duplicates, age, and mode before action.
- Risk gate
- Kill switch
- Audit trail
02 / Artifact no. 025
The receipts.
One sports-model experiment, unvarnished. Wins, losses, and recorded book value—research evidence, not a victory lap.
Latest model value—Loading dataset
Change from start—Historical experiment
Recorded decisions—2025 sample
Aggregate hit rate—Across recorded picks
MODEL / VALUE
Recorded valueExperimental book value
MODEL / ACCURACY
Daily hit rate
Source: Quantegy internal research datasets · Loading latest observation
03 / House rules
No magic.
No alibis.
Alpha is a hypothesis that survived costs, uncertainty, and repeated attempts to kill it.
Read the source ↗Evidence over confidence
Track assumptions, baselines, and failure—not just the favorable result.
Risk is part of the model
Position limits and stop conditions belong in the system, not in someone’s memory.
Automation stays accountable
Every signal and action should be attributable, reproducible, and reversible.
Have a difficult signal to find?