Research starts with a hypothesis
We treat trading ideas as hypotheses to be tested, not beliefs to be defended. A hypothesis must be expressed in rules that can be measured against historical behaviour and challenged under different conditions.
Strategy logic is one layer
A strategy defines participation logic—conditions for entry, exit and behaviour. TradeBoTicks does research strategy logic, but strategy is not treated as the whole system. A deployable system also needs validation, risk constraints, execution rules, monitoring and the ability to evolve when evidence changes.
Backtesting and validation
Historical testing is used to understand how defined rules behaved under known market conditions. We look beyond a headline win rate: drawdown, risk-reward behaviour, trade distribution, regime sensitivity and execution assumptions all matter. Backtests remain research evidence and do not guarantee future results.
Underlying-to-options translation
For the current OptionTurtle workflow, the strategy engine first defines entry, stop-loss and target levels on the underlying index. A separate option-selection layer then evaluates real-time option-market data and Greeks to identify an eligible option contract under predefined system rules and translate the index trade structure into option-level trade values.
Risk and execution infrastructure
Before an eligible trade reaches the broker, the risk layer uses available user capital and the maximum permitted exposure for that trade to calculate position size. The execution layer then applies predefined system-state and order checks before routing the entry. After successful entry, stop-loss and target orders are placed and monitored. The user retains the ability to exit an active trade. Real execution can still differ from intended execution because of slippage, liquidity, latency, broker/API behaviour or market conditions.
Observation and post-trade learning
Research does not stop at deployment. Trade logs and post-trade data create a feedback layer for investigating behaviour, execution quality and future research questions. We view deployment as part of the research lifecycle, not the end of it.
AI-assisted research
TradeBoTicks is exploring AI-assisted methods for research, parameter selection, trade analysis and system adaptation. Our use of AI is framed as assistance to structured research—not as a claim that AI can reliably predict markets or guarantee winning trades.
Walk-forward optimization & validation
Walk-forward optimization is part of how TradeBoTicks evaluates whether strategy behaviour remains robust across changing market conditions. Instead of optimizing a configuration once across the full historical dataset, we divide history into sequential rolling windows.
How the process works
Within each cycle, an earlier window is used to research and optimize eligible system parameters. The resulting configuration is then evaluated on the next unseen, out-of-sample period. The process moves forward through time and repeats across multiple windows.
This creates a sequence of train-and-test observations rather than one optimized backtest. We compare behaviour across windows using factors such as drawdown, trade frequency, risk-reward characteristics, stability of results and realistic execution assumptions.
What we are looking for
The objective is not to discover the single best historical parameter set. It is to identify system configurations whose behaviour remains sufficiently stable across different periods and market conditions to justify deeper validation. No individual walk-forward result is treated as proof of future performance.
Research boundaries
- No claim of guaranteed profitability.
- No assumption that historical behaviour must repeat.
- No strategy is exempt from risk controls.
- No automated system removes market, technical or execution risk.