A provisional preset, with explicit evidence limits
The current Recommended · provisional preset retains ADT’s existing protection values. Its identifier is synthetic-retained-v1. An engineering harness exercised the actual protection functions across seeded synthetic paths and neighboring parameter packages. No historical securities dataset or broker paper/live performance sample was available for this selection.
The evidence supports deterministic mechanics: independent high-water marks, upward protection ratchets, configured reversal triggers and exact exit reasons. It does not establish optimal thresholds, expected returns, a win rate or resilience across actual market regimes. The absence of stronger evidence was a reason to retain the existing values rather than silently change users’ live risk settings.
Exact protection values
The initial protective distance is 2% below average entry, subject to the strategy’s tick rounding and actual broker acceptance. Later tiers use the best observed price since entry:
| Peak gain activates at | Entry-price floor | Trail below high-water mark |
|---|---|---|
| +1.50% | +0.10% | No trail at this tier |
| +2.50% | +0.20% | 1.00% |
| +4.00% | +0.20% | 0.75% |
| +6.00% | +0.20% | 0.60% |
| +10.00% | +0.20% | 0.50% |
These are protection activation levels, not promises of net profit and not hard take-profit instructions. The protective level advances when a higher stage or high-water mark requires it. Ordinary pullbacks do not reset the high-water mark or loosen the level.
A configured reversal through protection initiates the exit. A tighter tier can exit on a smaller pullback; it cannot know whether the price will subsequently recover. A gap may fill substantially beyond the stop. Fees and slippage can turn a nominally positive floor into a net loss.
What the evaluation tested
The repeatable evaluation compares 19 parameter packages over 34 seeded synthetic or explicit paths, producing 646 outcomes. It includes trend, chop, gaps and reversal shapes. Separate discovery and validation seeds use independent random draws from the same assumed distributions; this is not an out-of-sample historical-market study.
Default friction assumptions are 5 basis points of adverse slippage plus half of a 10-basis-point spread on each side, and $0.35 commission for each entry and exit order. These are configurable scenario assumptions, not current IBKR fee quotes. Outputs include exits by reason, favorable and adverse excursion, excursion retention, gain/loss distributions, quote-path drawdown and cost effects.
The parameter comparison includes activation at 2%, 3% and 4%, constant trails of 1%, 1.5% and 2%, and optional tightening after an 8% move. Wider trails survive some pullbacks but can surrender more of a move. Expansion tightening changes those trade-offs on the synthetic sample. The observed differences do not justify declaring one package reliably superior for real markets.
What was not established
The protection harness does not model actual research quality, market microstructure, auction queue position, correlated portfolio exposure, monitoring delays or broker order throughput. Separate engine scenarios exercise multiple lifecycles, reconciliation, partial fills, retries and injected failures. Neither suite substitutes for timestamped historical replay or a paper session on the operator’s actual broker environment.
Engineering can rerun scripts/evaluate_trading_defaults.py with different slippage, spread, commission and share assumptions. The source distribution includes the full methodology in docs/recommended-trading-defaults.md and machine-readable evaluation results. Future threshold changes require stronger data and deliberate configuration updates.
Applying or customizing the preset
In Settings → Trading, select Use recommended and confirm. This intentionally replaces protection thresholds for future sessions while preserving the operating envelope and live-account authorization. Existing custom settings are not overwritten automatically, and a session already in progress keeps its recorded settings.
The implemented exit model includes downside protection, ratcheting profit protection, configured session exits and optional observed-loss exits. Manual flatten remains an override. There is currently no selectable hard-take-profit mode. Advanced settings expose supported protection thresholds without changing the requirement for broker-confirmed completion.