Portfolio Hunter
Hunter solves your rule-set exactly: you type every rule, it returns one optimal portfolio per size — a beta research preview, not a recommender.
The Portfolio Hunter turns portfolio construction inside out. In the Strategy Browser you pick strategies one by one; in Hunter you write down the rules a portfolio must satisfy — how much buying power, which families, what drawdown you'll tolerate, how entries must spread across the session — and the engine finds, for every portfolio size you asked about, the combination that satisfies all of them best. Not a suggestion list: an exact solve of the program you wrote.
Read this first: it's a beta, and it says so
Hunter ships as a research preview, and its own header carries the full disclaimer — worth reading in the product and worth repeating here verbatim:
Preview — beta. No correctness guarantee. Hunter is an experimental research preview. Its output may be incomplete, inaccurate, or flat-out wrong, and carries no guarantee of correctness of any kind. It is not advice, a recommendation, a signal, or a solicitation to trade. You are solely and fully responsible for any decision you make and any action you take based on it. Verify everything independently. Use entirely at your own risk.
That framing is not boilerplate — it describes how the tool actually behaves. Hunter never marks a "best" row, never pre-fills a single field, and grades its own results for selection luck. What it optimizes is your definition of good, over the same backtested catalog every other view reads.
Where it lives, who has it
Hunter sits in the sidebar under Catalog, at /workbench/hunter. It is
included with the Blueprint plan. If your account doesn't have access, the menu
item simply isn't there and the page redirects to the Strategy Browser — there
is no locked teaser.
Everything starts blank — deliberately
The settings form has no defaults. Buying power, objective, families, stop-loss levels, periods — every field starts empty and carries an amber ✳ until you set it, and the run bar reads "Still your choice to make" with the list of what's missing. The saved-settings bar says it outright: "your own named rule-sets — nothing is pre-filled for you".
That is a design position, not laziness: Hunter executes a program you wrote. There are no risk presets, no house recommendations, and no composite score ranking the output. If you run it twice with different rules, you get different tables — and the results page tells you exactly that.
What a run does
Press Hunt and three stages run, live on screen:
- Fetching candidates from the catalog — the same backtested variants the Browser shows, measured over your two chosen windows.
- Applying your filters — stop-loss levels, premium caps, eligibility.
- Exact solve under your rule-set — one portfolio per size — for each portfolio size between your min and max, an exact optimization finds the single best combination under every rule at once.
A typical run takes 5–40 seconds; the combined-drawdown and correlation guardrails each add a short daily-series fetch. The result is a menu: one row per solved size (plus optional near-optimal alternates), each opening into full KPIs, live analytics, a Luck Check, and a walk-forward panel. Any row can be loaded straight into the Portfolio Builder, where every position stays editable.
The three tabs
- Settings — the rule-set form, in three numbered steps: objective, capital & search and guardrails. Rule-sets can be saved, imported and exported as JSON.
- Results — the menu of solved portfolios and everything inside an opened row. See Results & history.
- History — every run you've made, with its settings and KPIs, recallable and comparable side by side.
The honest-numbers machinery
Two things distinguish Hunter from a typical "optimizer" and both get their own pages:
- Every opened result carries a Luck Check — a statistical grade of whether the portfolio's daily P/L is distinguishable from zero on data the search never saw, with the caveat printed right next to it that searched extremes can grade well partly by selection.
- The walk-forward panel replays your exact rule-set at past re-selection points and reports how much of the in-sample number survived out of sample — including how often your own drawdown cap was exceeded after selection.
A search across thousands of rule-satisfying candidates finds extremes, and extremes are partly luck. Hunter's job is to make that visible, not to hide it.