Let the system propose a strategy
Tell the system how much money you have and how much risk you’ll accept. It searches the space of possible goals (how aggressively to bid, whether to use leverage, how much risk to tolerate), picks the best one for the live parcel pool, and reports: the top deals it would buy, what would break the strategy, what would win it, and how confident it is.
Every number below is tagged Real data, Best-guess (sourced), or Your input with the source. Nothing is fabricated. See the glossary for what each tag means.
This page shows the most-recent proposal baked at build time by scripts/wave3_advisor_demo.py. A live re-run is also available via the new POST /advisor endpoint when a Python backend is reachable (Phase 1).
Searched the Objective space for the best risk-adjusted strategy matching your goal (capital $800,000, risk=moderate, modes=cash,owner_finance,debt_leveraged). Adjustments from your input: acquisition_target_pct: 45% -> 32%; max_leverage_ratio: 3.0 -> None; risk_tolerance: moderate -> aggressive Best risk-adjusted score: mean P/L $579,969, P(ruin) 0.0%, expected ROI 140.3%. Allocated 49 of 60 candidates ($377,388 deployed).
| Parcel ID | Buy price | Profit (est.) | Annual return | $-for-$ score |
|---|---|---|---|---|
| 10104973B | $33,865 | $69,865 | 104.6% | 2.06 |
| 10115003B | $28,252 | $58,285 | 104.6% | 2.06 |
| 10114577 | $25,818 | $53,262 | 104.6% | 2.06 |
| 10114571 | $22,459 | $46,333 | 104.6% | 2.06 |
| 10104001C | $20,548 | $42,391 | 104.6% | 2.06 |
Combinations of bad markets (catastrophe rate, sale-price shocks) where the strategy goes negative or busts past the ruin ceiling.
No breaking scenarios found in the swept range.
Combinations where the strategy posts top-decile profit with the ruin chance still under your ceiling.
Top winning conditions (2 total): - cat_prob=0.09; sale_ratio_mean=1.04; sale_ratio_std=0.05 -> mean P/L $716,073, P(ruin) 0.0% - cat_prob=0.08; sale_ratio_mean=0.97; sale_ratio_std=0.13 -> mean P/L $645,037, P(ruin) 0.0%
The system flags every important number it uses with where it came from. Real data = fit from actual data the system has loaded. Best-guess = a sourced placeholder (we cite where it came from). Your input = you typed it on the Your-goal page. Nothing here is fabricated.
| What it’s used for | Source kind | Value | Where it came from |
|---|---|---|---|
| objective.capital | Your input | $800,000 | user-supplied |
| objective.acquisition_target_pct | Your input | 32% | user-supplied (advisor may have tuned) |
| simulator.regime_mix | Best-guess | 0.0 | FRED NBER recession indicator + ATTOM home-price index 2000-2024, https://fred.stlouisfed.org/series/USREC (free) |
| simulator.cat_prob_per_deal | Best-guess | 0.12 | NOAA Storm Events Database 2014-2024 county-year freq, https://www.ncdc.noaa.gov/stormevents/ (free) |
| lender_book.monthly_default_hazard | Best-guess | 0.005 | CFPB Mortgage Performance Trends 2018-2024 non-prime SDQ, https://www.consumerfinance.gov/data-research/mortgage-performance-trends/ (free) |
| lender_book.default_recovery_pct | Best-guess | 0.5 | MERSCorp / Fannie Mae REO disposition reports 2020-2024, median net-of-cost recovery on land-secured notes; free summary at FHFA House Price Index |
Want a fresher proposal? Run uv run python scripts/wave3_advisor_demo.py locally — it pulls the latest real parcels from Supabase and re-bakes this page’s data. Or hit the new POST /advisor endpoint for a live re-run (Phase 1).