Pybroker Strategy Creator

PyBroker Strategy Creator

Overview

Create practical PyBroker strategy code from user intent while preserving backtest hygiene, including no lookahead leakage, explicit sizing, clear risk controls, and locally valid PyBroker API usage.

Workflow

1. Extract the strategy spec: universe, data source, date range, timeframe, long/short permissions, entry and exit rules, sizing, stops, ranking, rebalancing cadence, model training needs, and desired output file/notebook.
2. Ask only for missing blockers. If details are absent but noncritical, make conservative assumptions and state them in the final answer or code comments.
3. Read references/wiki-index.md to choose the smallest relevant wiki page. For nontrivial strategy work, also read references/pybroker-patterns.md.
4. Build a complete runnable strategy surface:
   • start scripts with pybroker.disable_progress_bar() and pybroker.enable_data_source_cache("<name>")
   • create a StrategyConfig when cash, fees, delays, exits, margin, or returned signals/stops/positions matter
   • define indicators with highest, lowest, returns, the built-in factories in pybroker.indicator (such as atr), or indicator with vectorized NumPy/Numba functions
   • define model sources with pybroker.model only when training or loading predictions is part of the request
   • write execution functions that use completed-bar arrays such as ctx.close[-1], guard lookbacks with ctx.bars or warmup, and set at most one order side per symbol per bar
   • add executions with Strategy.add_execution, passing hyperparams= when optimization is involved, intervals= for higher-timeframe bars, and .intervals(...)-bound indicators/models for per-interval computation, and cap positions with strategy.set_max_long_positions / set_max_short_positions
   • run backtest for a single train/test pass, walkforward for model/walk-forward evaluation, or optimize for hyperparameter search; pass timeframe= whenever an execution declares intervals= or binds a model/indicator to an interval
5. Validate the produced code as far as the environment allows. At minimum, run syntax checks for created Python files. Run tests or a small local-data backtest when the repo and data make that practical.

Implementation Rules

• Treat PyBroker as a backtesting framework, not a source of financial advice. Make strategy assumptions explicit and avoid performance claims that are not supported by the produced backtest.
• Use completed historical bar data only. Do not use future prices, future indicator values, or shuffled time series unless explicitly doing a model training split that PyBroker supports. An indicator value at bar i may depend only on inputs at index i and earlier: no backward shifts such as shift(-1), and no negative indexing into full-length arrays inside indicator functions (a negative index silently wraps to the end of the series — the future).
• Start generated scripts with pybroker.disable_progress_bar() so backtest progress output does not flood agent context, and pybroker.enable_data_source_cache("<strategy_name>") (or pybroker.enable_caches) so repeated runs do not refetch data. Add pybroker.disable_logging() when running many backtests, such as parameter optimization.
• Report result.metrics_df as the human-readable summary. When structured output is needed (agent parsing, saved report files, downstream tools), use result.to_json() / result.to_json_str(): the default payload serializes metrics, trades, orders, and bootstrap capped at max_rows=100 rows per table, symbols= filters to specific tickers, and include= opts into portfolio/positions/metrics_df/signals/stops (positions needs StrategyConfig(record_position_bars=True)). Do not replace the metrics_df print outright: the default JSON payload (trades plus orders) is usually larger than the metrics table.
• Never use pandas to implement indicator or execution logic. BarData and ExecContext price fields are NumPy arrays: operate on them directly, prefer the vectorized helpers (highv, lowv, sumv, returnv, cross, atr) when they fit, and JIT-compile explicit loops with Numba @njit. Never construct a pd.Series or pd.DataFrame and never call pandas methods such as .rolling, .ewm, .shift, or .apply inside an indicator function or a per-bar execution function.
• If a Numba @njit function fails to compile or raises a cryptic error such as a TypingError, re-run once with the NUMBA_DISABLE_JIT=1 environment variable to get a readable Python traceback, fix the underlying code, then remove the variable so the backtest runs compiled.
• Self-test novel indicator logic for lookahead with the bump-last-bar check in references/pybroker-patterns.md: recompute after changing only the final input bar and assert every earlier output is unchanged.
• Use ctx.calc_target_shares(target_size) for allocation-based sizing and ctx.set_target_shares(target, dir="long") to rebalance toward a target allocation. Use fixed ctx.buy_shares or ctx.sell_shares only when the user asks for fixed share sizing.
• Check ctx.long_pos() or ctx.short_pos() before entering or exiting positions. Use ctx.sell_all_shares() and ctx.cover_all_shares() for full exits.
• Set entry-time stops on the same bar as the entry order: hold_bars, stop_loss_pct, stop_profit_pct, or stop_trailing_pct.
• Orders fill at PriceType.MIDDLE — the midpoint of the low and high of the execution bar, which under the default buy_delay/sell_delay of 1 is the bar after the signal, so PriceType.CLOSE means the next bar's close. Override with ctx.buy_fill_price / ctx.sell_fill_price, which take a PriceType (OPEN, HIGH, LOW, CLOSE, MIDDLE, AVERAGE), a number, or a (symbol, bar_data) callable, and read back as None rather than MIDDLE until set. A limit price only gates the fill: the order still fills at the fill price, never at the limit, so ctx.buy_limit_price = 200 against a midpoint of 108 books 108.
• StrategyConfig.exit_on_last_bar defaults to False, which leaves any position still open when the data ends. That position never becomes a Trade, so trade_count, win_rate, total_pnl and every other trade-level metric silently exclude it while its P&L sits in unrealized_pnl. Set exit_on_last_bar=True whenever trade statistics are reported; exits fill at exit_sell_fill_price / exit_cover_fill_price, both PriceType.MIDDLE, and in walkforward the liquidation fires only on each symbol's true final bar, never at window boundaries. Bar-level metrics (sharpe, max_drawdown) are computed from per-bar market value and barely move either way.
• calc_bootstrap is a backtest/walkforward/optimize parameter defaulting to False, not a StrategyConfig field. Pass calc_bootstrap=True to populate result.bootstrap with conf_intervals (BCa — bias corrected and accelerated — intervals for profit factor and Sharpe; 6x2, MultiIndexed on name then conf, columns lower/upper) and drawdown_conf (percentile bounds on max drawdown; 4x2, indexed on conf, columns amount/percent). It leaves metrics_df unchanged, costs roughly bars x StrategyConfig.bootstrap_samples (default 10_000, so lower it on intraday data), and needs StrategyConfig.bars_per_year or the Sharpe intervals are per-bar rather than annualized.
• Rank symbols with ctx.long_score and ctx.short_score and cap positions with strategy.set_max_long_positions(n) / strategy.set_max_short_positions(n); the StrategyConfig fields of the same names are deprecated. Scores rank descending on both sides: short orders go to the symbols with the highest short_score, so negate a lowest-wins short signal (for example ctx.short_score = -roc to short the most negative momentum). For rank-and-rotate portfolios call strategy.enable_rotation(worst_rank_held=...); rotation is exclusive, so execution functions then only set scores and any order fields they set are ignored.
• For parameter search, register values with pybroker.hyperparam(name, default=..., low=..., high=..., step=...), attach them via indicator kwargs or add_execution(..., hyperparams=[...]), read them with ctx.hyperparam("name"), and run strategy.optimize(score_fn, sampler="grid") (or "tpe"/"random" with n_trials=). Optimization does not support trainable models; use pretrained models or indicator-based rules.
• For multi-timeframe logic, declare add_execution(..., intervals="weekly") and read compressed bars with ctx.interval("weekly"). intervals= provides bars only; to compute an indicator or train a model per interval, bind it with indicator.intervals("weekly") / model_source.intervals("weekly") in indicators=/models= — bound intervals are available through ctx.interval without declaring them again. Binding is exhaustive: include "base" (e.g. .intervals("base", "weekly")) to keep the base-timeframe variant; unbound sources default to base. When any execution declares or binds intervals, backtest/walkforward require timeframe= for the base data.
• Model trading costs with strategy.set_slippage_model(...): FixedSlippageModel(bps=...) for constant costs, VolatilitySlippageModel for ATR-scaled slippage, VolumeSlippageModel for volume-capped fills. Only enable margin (StrategyConfig(leverage=..., interest_rate=...), which requires bars_per_year) when the user asks for it.
• result.positions is empty unless StrategyConfig(record_position_bars=True); enable it only when the user needs per-bar position output. The per-bar result.portfolio equity curve is always populated.
• Use strategy.set_before_exec or strategy.set_after_exec for cross-symbol portfolio logic instead of hiding global state inside a per-symbol execution function.
• Optional packages are not PyBroker dependencies: name the required pip install for any data source or library the script imports (for example pip install yfinance for YFinance) and never assume one is importable. When the network or a data-source package is unavailable, validate with a tiny local DataFrame passed to Strategy instead.
• If exact API names, constructor parameters, or methods matter, read references/api-public-surface.md.
• For exact type signatures — the writable ExecContext order/stop attributes, property and parameter types, enum members — read the matching references/pybroker_*.pyi stub (pybroker_context.pyi for ExecContext and slippage).
• If the user wants a standalone file, copy and adapt assets/strategy_template.py.

Common Deliverables

• Standalone .py backtest script.
• Notebook-ready PyBroker cells.
• Refactor of an existing strategy file.
• Debugging notes and patches for invalid ExecContext usage.
• Focused tests using local DataFrame data when live data sources are unavailable.

Resources

• references/wiki-index.md: start here for topic routing across the bundled PyBroker wiki.
• references/wiki-01-getting-started-with-data-sources.md: Yahoo Finance, Alpaca, Alpaca Crypto, AKShare, data caching, and data source setup.
• references/wiki-02-backtesting-a-strategy.md: defining execution rules, adding executions, running backtests, and filtering data.
• references/wiki-03-evaluating-with-bootstrap-metrics.md: evaluation metrics, confidence intervals, bootstrap metrics, and drawdown.
• references/wiki-04-ranking-long-and-short-signals.md: ranking long/short signals by score and max positions.
• references/wiki-05-writing-indicators.md: custom indicators, vector helpers, TA-Lib, built-in indicators, and indicator sets.
• references/wiki-06-training-a-model.md: model training, model predictions, caching, and walkforward analysis.
• references/wiki-07-creating-a-custom-data-source.md: extending DataSource, DataFrame inputs, CSV inputs, and custom columns.
• references/wiki-08-applying-stops.md: stop loss, take profit, trailing stops, limit prices, stop exit prices, and stop cancellation.
• references/wiki-09-rebalancing-positions.md: equal weighting, before/after execution hooks, and portfolio optimization.
• references/wiki-10-rotational-trading.md: rotational strategy examples, universe rotation, and custom position sizing.
• references/wiki-11-configuring-parallelization.md: worker counts, parallel indicators and model training, and the Ray backend.
• references/wiki-12-parameter-optimization.md: hyperparams, grid/TPE/random samplers, and walkforward optimization.
• references/wiki-13-margin-trading.md: leverage, buying power, margin interest, and shorting on margin.
• references/wiki-14-modeling-slippage.md: fixed, volatility, and volume slippage models plus custom slippage.
• references/wiki-15-multiple-time-intervals.md: interval types, compressing bars, and multi-timeframe strategies.
• references/wiki-16-time-series-models.md: GARCH volatility forecasting and models on lagged returns.
• references/wiki-17-multi-symbol-models.md: pooled models trained across multiple symbols.
• references/wiki-18-dynamic-symbol-selection.md: SymbolSelector and per-window universe selection.
• references/wiki-faqs.md: common PyBroker usage questions and edge cases.
• references/api-public-surface.md: generated public API signatures and first docstring sentences from local source.
• references/pybroker_context.pyi: generated type stubs for ExecContext (including its writable order/stop attributes), IntervalContext, RotationContext, ExecResult, and the slippage models.
• references/pybroker_strategy.pyi: generated type stubs for Strategy, StrategyConfig, TestResult, and the optimization types.
• references/pybroker_types.pyi: generated type stubs for enums, BarData, Portfolio, order/trade/position records, and evaluation result types.
• references/pybroker_model.pyi: generated type stubs for model(), indicator(), vector helpers, data sources, and top-level module functions.
• references/pybroker-patterns.md: load when writing nontrivial strategy code, debugging PyBroker API usage, or adding indicators, stops, models, ranking, rebalancing, or walkforward analysis.
• assets/strategy_template.py: copy and adapt when creating a new standalone strategy script.