Skip to content

Latest commit

 

History

History
73 lines (63 loc) · 2.28 KB

TODO.md

File metadata and controls

73 lines (63 loc) · 2.28 KB

compfipy todo / roadmap

Basic Code Modules

Calculator

  • General finance calculations
  • [ ]

Models

  • create stochastic models
  • [ ]

Asset

  • OCLHV time series data storage
  • OCLHV time series calculations
  • OCLHV time series technical equations
  • Performance measures
  • Risk measures
  • Market Comparisons
  • Summary Asset
  • [ ]

Portfolio

  • Aggregate specific Assets in Table
  • Define 'holding' of each Asset
  • Enter and Exit holding for one Asset
  • Rebalance whole Portfolio
  • Return weights
  • Total Unrealized Performance measures (based on specific time)
  • Total Unrealized Risk measures (based on specific time)
  • Total Unrealized Market Comparisons (based on specific time)
  • Total Realized Performance measures (based on holdings)
  • Total Realized Risk measures (based on holdings)
  • Total Realized Market Comparisons (based on holdings)
  • Summarize Portfolio
  • [ ]

Market

  • Aggregate of all Assets
  • Aggregate by sector?
  • Aggregate by cap?
  • [ ]

Events

  • Define event as some parameter (price, SAM, MACD, etc.) passing a threshold
  • Search an Asset for the event
  • Return booleans of when the event happens
  • [ ]

Strategy

  • Defines how a Events will affect a Portfolio
  • Define Enter and Exit Positions based on one or more events
  • Update Holdings of Portfolio based on the Strategy
  • [ ]

Overall Function and Flow

  • create trend trading algorithm
  • test trend algos on stochastic models, just for process completeness, not correctness at the moment
  • [ ]

Parameter Optimization

  • get asset history
  • design genetic algorithm to solve for stochastic model parameters to model asset history
  • use multi-temporal windows of asset history to design parameters
  • see how multi-temporal parameters change
  • find general best fit parameters for stochastic models
  • use genetic algorithm to train trending algos parameters, using some weighted combination of profit, volatility, risk, etc. as fitness function
  • once trained on blind stochastic model run on historical asset data
  • do this once for a simple case
  • repeat for maximizing return
  • repeat for minimizing risk
  • define a way to combine the two
  • [ ]