Guardrails
Hunter's constraint step: the combined-DD cap and what it honestly can't promise, correlation and stop-out caps, entry-timing rules, and the expert panel.
Step 3 collects the limits the solve must respect. The step's own hint sets the contract: "Type a number to apply one; leave it empty for no limit." All four main guardrails are off by default — an empty field means the constraint does not exist.
Combined DD cap (% of BP)
The most powerful guardrail, and the one with the most honest disclosure in the product. It caps the true combined-curve max drawdown — the worst peak-to-trough of the whole portfolio's daily equity, read from the picks' real daily series, not an average of per-strategy percentages — at your chosen percentage of the buying-power budget.
What it cannot do is bind the future, and the field says so whenever the cap is active:
Applies to the reference window the search optimises — not a forward limit. Realized drawdowns after selection can exceed it.
The full help text goes further and quantifies it: in the 2026-07 walk-forward study, one low-DD preset replayed over 3 years exceeded its 1.5% cap in all 11 quarters tested, with a stitched max drawdown about 6.3× the cap. The cap is an in-sample shaping constraint: tightening it makes the optimizer shed exposure and bring in wider, less-correlated spreads — useful — but the number itself describes the past window, nothing more. To see the measured out-of-sample record for your configuration, run the walk-forward panel.
If the cap is below what's achievable, Hunter returns the lowest-drawdown portfolio it found and says so rather than failing silently.
Max pairwise correlation
Forbids selecting any two strategies whose daily-P/L correlation exceeds your threshold. This is an exact constraint, and its practical value is killing near-duplicates — two 10:05 50-wides fifty cents apart are one bet, not two. Lower = stricter. The average pairwise correlation of the result is always reported in the live analytics whether or not the cap is set.
Both this and the combined-DD cap read daily series, adding roughly 1–2 seconds each to the run.
Max stop-out day (% of BP)
Caps the theoretical worst day — if every trade stops out — as a percentage of the budget. Loss when a spread stops equals premium × stop-multiple, so this bounds Σ(premium × SL) across the picks. It's structural and regime-independent: computed from the strategy definitions, no daily series. The help text carries the one caveat that matters: it assumes stops fill — in a gap or dislocation, stops can fail and losses approach the full spread width.
Max premium / strategy ($)
A universe filter rather than a portfolio constraint: excludes any strategy whose per-contract credit exceeds this dollar amount. Higher premium means the short strike sits closer to the money — higher win rate, but deeper drawdowns when it's reached. Capping premium keeps only the further-out-of-the-money, calmer strategies in the pool. Catalog premiums span roughly $1–6.
Entry timing
How entries must spread across the session:
- 60-min window max (required) — within any rolling 60-minute window of the day, at most this share of the picks may have their entry time. This is the anti-clustering rule; 100% means no restriction, and the form makes you say so explicitly rather than assuming it.
- 120-min window max (optional) — same idea over rolling 120 minutes.
- Last entry (HH:MM) (optional) — excludes strategies entering later than this time. Empty = no cutoff.
- No duplicate entry time (toggle) — no two picks at the exact same minute.
Advanced rules (expert)
Collapsed by default, for rules most runs don't need:
- Result tolerance (%) — controls alternates in the results menu. Besides each size's exact optimum, the menu also lists other found portfolios within this percentage of it — but only if they are materially different: an alternate must differ from the optimum in at least a third of its picks and stand at least 15% apart on the objective from every row already listed. Near-identical rows are noise, not choice — which also means a tolerance below 15% produces no alternates at all. Enumeration is compute-bounded (up to 3 alternates per size), and a warning appears if the bound cut the search short.
- Per-strategy DD cap (%) — a global pool filter: any strategy whose reference-period max drawdown exceeds this never enters the solve. Acts as the fallback for families without their own cap below.
- Max premium sold (% of BP) — a sizing throttle capping total credit collected per day. The help text is blunt about what it is not: "This is NOT a loss cap — premium is the max GAIN, not the max loss; use 'Max stop-out day' to bound losses."
- Wide (≥100) min share / Narrow (<50) max share — structural mix rules on spread width: wide spreads have more room before max loss, narrow ones are cheap but hit max loss quickly.
- Family mix — a per-family table over MEIC / METF / CALL / PUT / RATIO: minimum pick count, minimum share of the portfolio, and a per-family DD cap that overrides the global one. The pattern from the help text: keep MEIC tight at 1.3% while letting METF — which hedges — run looser at 5%.
Reading this page as a whole
Each guardrail states its own limits: the DD cap describes the past, the stop-out cap assumes fills, the premium cap bounds gains not losses. That honesty is the point. Constraints shape what the search returns; they are not promises about what the market will do to the result.