Insight Memo - jaeaehkim/trading_system_beta GitHub Wiki

Insight

• 트레이딩 스타일 개요 : Risk-Free & Data-Momentum Arbitrage
• 왜 비트코인만 (Data-momentum) 트레이딩 하는 것으로 충분한가? : [거래비용을 고려한 회전율] X [선별된 Featrue로 부터 나온 확률우위]
  • 공급은 constant, 수요만 변수
  • daily transaction fee / daily issuance
    • lightening network
• 알트코인 트레이딩의 활용 방법
  • 비트코인 데이터로 학습된 모델을 활용해 공통요소와 개별요소를 구분지어 활용하기
  • transaction fee / issance로 비교 : https://xangle.io/research/62bd080e534a07d0b2ffb6aa
• Rule-Based vs Machine Learning 시스템의 본질적 차이는 무엇인가?
• mlops architecture 설계 : https://cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning?hl=ko
  • domain architecture
    • HRP & Cash design
      • base 5:5
    • Model Selection by momentum
    • Model Factory
      • Feature Importance
  • tech architecture
    • async trading
      • unit weight = model num / 10;
• Feature Idea
  • frame : common things vs individual things
    • log preprocessing : 자연물은 log로 가공
  • ohlcv, option, spot-futures, microstructure data (trade data grouping)
  • https://edition.cnn.com/markets/fear-and-greed
  • crypto fundamental
    • onchain data
      • 활성화 주소
      • 지갑 갯수
    • cryptoistic financial statement
    • supply/demand bitcoin
      • stock to flow model
    • hashrate
      • hash ribbon
  • spreadistic : ex) yield spread
    • 미체결약정 : https://blog.naver.com/PostView.naver?blogId=low01033&logNo=222383895185&parentCategoryNo=&categoryNo=18&viewDate=&isShowPopularPosts=true&from=search
    • cme - 바이낸스, 코베-바이낸스
    • BTC-ETH
    • USDC+USDT weight
    • spot-futures
    • funding fee
    • market cap diff
    • option data