running_todo.todo

Running Todo:

    ☐ Write steps to add new strategy to backtest framework
    ☐ Write steps to add new strategy to live trading framework
    ☐ Check if thetadata v3 is backward compatible with v2
    ☐ Write proper docstring for binomial model in optlib
    ☐ Complelely decommission old method of getting rates.
    ☐ How do we handle reverse splits?
    ☐ Write a ChainDataManager
    ☐ Add full dataframe tp Spot Result
        - Include price property to pick from columns. Create enum for standard columns
    ☐ Remove all import for riskmanager.market_data
    ☐ Save scaling in position sizing info to database
    ✔ Make order picker faster & more efficient @done(26-02-17 07:17)
    ☐ Singleton Metaclass with attribute picking
    ✔ Find a way to save historically closed in signals tabled. @done(26-02-17 07:17)
    ☐ Add slippage pnl to live attribution
    ☐ Aggregator should take trades & equity data only
    ✔ EVB should use StrategyBase class (Optionally, dataframe, but probably should discourage it) @done(26-02-17 07:17)
    ☐ Switch to TimeseriesDataManager for pnl attribution
    ☐ Discontinue all use of module_test.raw_code.DataManagers
        1. Option.py
        2. xmultiply_attr.py
    ✔ Add refresh information to MarketTimeseries to avoid constantly refreshing data when not needed. @done(26-02-17 07:19)
    ☐ Move all vol and pricing logic to use the new OptLib
        ☐ Eg FillOptimizer
    ✔ Ensure MarketTimeseries is not cutting out dividends or split factors for real time @done(26-02-17 07:19)
    ✔ Ensure make position id & signal id comes from one place (Signal id from strategy base, position id from create order) @done(26-02-17 07:20)
    ☐ Update get_risk_free_rate_helper to use RatesDataManager
    ☐ Look into alternatives for backfilling dividends with 0.0 in get_timeseries
    ☐ Move strategy files to db.
    ✔ Set evb default market_model = bsm, div = continuous for speed. Disable logging. @done(26-02-17 07:20)
    ☐ Replace populate_chain_cache with ChainDataManager which caches none-today data and reloads today data on each call.
    ☐ Remind zino about table diffing
    ✔ Update info in config dictionary in order picker to reflect new changes (eg min_total_price) @done(26-02-17 07:21)
    ☐ streamline all evb stream logging level as done in trade.datamanager
    ☐ Normalize everywhere to use the following for position:
        - fyi: This includes: events, signals, positions, etc. Anywhere we need to represent a position, we should use this format to avoid confusion and ensure consistency across the codebase.
        - Side int: 1 for long, -1 for short
        - Side str: "long" or "short"
        - Position effect: "open" or "close"
        - It should all come from eventdriven.types.position
    ☐ Clean up old todos in the codebase and move any relevant ones to this todo list.
    

Todo before going back to backtesting:
    ✔ For test order gen, it should use PREVIOU day's cash of order gen, not current cash. @done(26-02-13 14:52)
    ✔ Allow preset orders in Evb order getter @done(26-02-09 21:14)
    ✔ Clean up evb order getter + preset orders (speed up essentially) @done(26-02-11 21:44)
    ✔ Push config change to db (changed min_total_price to 0.5 from 0.95) @done(26-02-13 09:42)
    ☐ Move emefiele reports to the reports channel
    ☐ Populate get_strategy_instance with has_position (Optional)
    ✔ Cache checker should use date range info, with missing dates info saved @done(26-02-14 18:47)
        ✔ All ts loaders should return the data @done(26-02-15 21:28)
        ✔ Maybe cache timeseries object and use to return date range & at index to avoid doing multiple cache lookups @done(26-02-15 21:28)
        ✔ Maybe during the first get_timeseries, cache the entire date range for syms with class range, find a way to drop today as usual. @done(26-02-15 21:28)
        ✔ New methods in timeseries object: `covers_full_range(start, end)`, `get_at_index(date)`, `get_date_range(start, end)` @done(26-02-15 21:28)
    ✔ Evb should use strategy base class and not have any direct dependency on dataframe (except for timeseries data retrieval) @done(26-02-15 19:45)
    ✔ Update start date in Backtesting.py stats generated during PTBacktester. And general cleanup. @done(26-02-15 20:39)
    ✔ Implement tplusn execution in StrategyBase @done(26-02-13 16:29)
    ✔ Close signals added to signals table @done(26-02-09 16:34)
    ✔ Ensure backtesting a live strategy works seamlessly @done(26-02-15 21:28)
    ✔ Decipher the diff btwn backtest and live strategy performance @done(26-02-10 19:57)
    ✔ I don't like how limit sizing is going. Look into it. @done(26-02-10 21:25)

Future Todos:
    ☐ In the new order gen system:
        1. Orders will be generated straight from StrategyBase instead of being generated with a table
        2. Signals table will be generated by using StrategyBase class to make the model aware of positions, and then create a snapshot
            a. To achieve this, get_strategy_instance will need to be updated to store current positions information and pass it to the strategy instance. This will allow the strategy to make informed decisions about generating signals based on existing positions.
        3. This will allow us to have a more seamless integration between backtesting and live trading, as the same logic will be used for both. It will also allow us to have more complex strategies that can take into account existing positions when generating new signals.
        4. Signal table can actually show the decisions of inidividual tickers in the strategies.

    ☐ Backtesting.py doesn't have t-plus-n functionality. But we can effectively implement it with the wrapper around StrategyBase.