Example Report
Parameter Optimization Report Example
How FiClaw extracts tunable parameters from a strategy, runs a grid search and judges parameter sets by return, drawdown, Sharpe and neighborhood stability.
Sample prompt
“Optimize a momentum rotation strategy: search the momentum window, holding count and rebalance frequency to raise Sharpe while containing max drawdown.”
Backtest setup
Search space
Window/holdings/freq
Core tunable parameters extracted from the spec and code
History
Up to 10 yr+
Used to observe cross-period parameter stability when data allows
Method
Grid search
Cover the main parameter combinations first
Robustness
Neighborhood
Not just the single best-point result
Strategy summary
FiClaw identifies three core parameters in the strategy code (momentum window, holding count and rebalance frequency), builds a bounded grid search space, batch-submits backtests and produces a parameter comparison table. The report highlights not only the best value but whether neighboring parameter sets are stable.
Code highlights
- Identify the lookback_window, top_n and rebalance_frequency parameters.
- Generate the parameter grid and submit an independent backtest per combination.
- Aggregate return, drawdown, Sharpe, win rate and turnover into a comparable results table.
Key metrics
Best window
40-day
More balanced between return and drawdown
Holding count
30 names
More diversified than a 10/20-name portfolio
Best Sharpe
1.21
Example grid-search result
Max drawdown
-13.8%
Better than the original parameter set
Diagnosis
- The 40- and 60-day windows perform similarly, showing the strategy does not fully depend on a single-point parameter.
- Too few holdings raise return but increase drawdown, making the risk-return unstable.
- Monthly rebalancing beats weekly, showing excessive turnover erodes return.
Next steps
- Run a finer-grained search around the 40-day window.
- Add a transaction-cost sensitivity analysis.
- Split market regimes to check whether the parameters stay stable.
Citable facts
- This example shows how FiClaw turns strategy parameters into a comparable backtest results table.
- A parameter optimization report reads return, max drawdown, Sharpe, win rate, turnover and neighborhood stability together.
- FiClaw stresses that parameter optimization needs out-of-sample validation and transaction-cost sensitivity analysis.
Boundary note
This is not investment advice. The metrics on this page are example-report figures, used to show how FiClaw organizes strategy generation, backtesting and diagnosis. They are not investment advice and do not represent future returns.