Most investors have the same information. The difference isn't insight—it's process. HARVEST PROTOCOL is a complete educational investment research framework — a quantitative scoring methodology, deployment-gate logic, and behavioral circuit breakers — delivered as a structured publication you run inside your own LLM, backed by a deterministic scoring engine: same ticker, same data, same number, every run. Optimized for Claude; runs at high fidelity on any frontier model. No account to create; your portfolio is never stored on our side.
Educational research framework · Not investment advice · Not individual recommendations
Behavioral finance research documents an average 1.5–2% annual return loss from preventable decision errors — selling at panic lows, deploying at FOMO peaks, holding positions whose thesis broke months ago.
Morningstar Mind the Gap 2024; DALBAR QAIB 2024 — investor returns lagged fund returns by 1.68% annualized over 10 years.
Over a decade, a 1.5–2% annual drag compounds into a meaningful share of total return — not from bad picks, but from bad process. (Published Morningstar/DALBAR research averages — context for why a disciplined process matters, not a projection of any individual's outcome.)
S&P's SPIVA scorecards have measured active fund managers against their benchmarks for over two decades. The latest results: 79% of active large-cap U.S. equity funds underperformed the S&P 500 in 2025, and over a 20-year horizon roughly 92% of domestic funds underperformed their benchmarks.
S&P Dow Jones Indices, SPIVA U.S. Year-End 2025 Scorecard — 79% of active large-cap U.S. equity funds underperformed the S&P 500 in 2025; ~92% of domestic funds underperformed their benchmarks over 20 years.
That is not an argument that you will do better than the professionals. It is the argument for the opposite posture: a disciplined, low-cost, self-run process that treats broad index exposure as the default, holds every individual candidate to a high evidentiary bar, and keeps an honest written record of how its own past verdicts resolved. (Published S&P Dow Jones Indices research — context for why the framework defaults to the index sleeve and demands self-measurement, not a projection of any individual's outcome.)
SPIVA figures are published third-party index research (S&P Dow Jones Indices, SPIVA U.S. Year-End 2025 Scorecard; SPIVA U.S. Persistence Scorecard, Year-End 2024 edition), cited as context for why a disciplined, low-cost, index-defaulting process matters — not a guarantee, projection, or representation that this framework or any user will outperform any benchmark. We make no performance claims. Educational research framework. Not financial advice. Not a registered investment adviser.
Retrospective, hypothetical application of the screening criteria, shown for educational illustration only. This is not investment performance, not a track record, and not indicative of any result any buyer obtained or will obtain. Names were selected to illustrate trigger types and are not a representative or complete sample of names the criteria would flag; the criteria shown are simplified. The publisher exemption (Lowe v. SEC) concerns adviser status, not performance presentation — the format of any outcome-referencing content remains subject to securities-counsel review under SEC Marketing Rule 206(4)-1 before any performance figures are reintroduced. Not financial advice. Not a registered investment adviser. Your decisions are your own responsibility.
THE AI IS THE REASONING ENGINE. THE PROTOCOL IS THE PRODUCT.
HARVEST PROTOCOL gives Claude a 500-page protocol—with a Deployment Gate, 10-factor scoring, Expected Value Framework, and 9 exit triggers—that converts analysis into a sized, routed, behaviorally-enforced research output. And the load-bearing math doesn't run on the model's arithmetic at all: a hosted scoring engine computes the intrinsic-value anchor, expected value, verdict routing, and scoring gates server-side and returns fixed values — same ticker, same data, same number every run, auditable, never improvised. The AI is the reasoning engine. The protocol is the product.
| CAPABILITY | AI ALONE | WITH PROTOCOL |
|---|---|---|
| Analysis quality | High—but unfocused | 40+ quantitative checks, structured |
| Reproducible, auditable numbers | ✗ A different figure each run | ✓ Load-bearing math server-computed — same data, same number every run |
| Position sizing | ✗ None | ✓ EV-weighted, risk-adjusted |
| Market timing gate | ✗ None | ✓ VIX/CAPE/sentiment deployment gate |
| Exit criteria | ✗ None | ✓ 9 pre-committed triggers |
| Behavioral circuit breakers | ✗ None | ✓ VP disconfirmation, EV decay |
| FEATURE | HARVEST | Automated Services | Tip / Recommendation Services | Asking Claude/GPT Directly |
|---|---|---|---|---|
| Deterministic, server-computed scoring | ✓ Auditable, same number every run | Black box | ✗ | ✗ Non-reproducible |
| Deployment Gate (VIX/CAPE) | ✓ | ✗ | ✗ | ✗ |
| 9 pre-committed exit triggers | ✓ | Auto rebal only | ✗ | ✗ |
| EV-weighted position sizing | ✓ | Fixed % | ✗ | ✗ |
| 40+ quantitative checks | ✓ | ✗ | Varies | Unstructured |
| Behavioral circuit breakers | ✓ | ✗ | ✗ | ✗ |
| FIRE / tax routing integration | ✓ | ✗ | ✗ | ✗ |
| Your portfolio stays yours — no account, never stored | ✓ | Full account access | Cloud | Cloud |
| Cost | $127 once (Starter) | 0.25% AUM/yr | $99–299/yr | $0–20/mo |
| Decisions or just analysis? | Sized + routed framework output | Managed decisions | Ideas only | Essays only |
| You stay the operator | ✓ | ✗ | Partial | ✓ |
HARVEST PROTOCOL is a paid educational research publication and methodology — not a tip service. It is a complete investment research framework with three integrated components that work together to convert analysis into repeatable, behaviorally-enforced research outputs — all applied by Claude inside your own session, following the published method.
34 chapters. Knowledge layer spanning Graham to Dalio. Compiled mental models, frameworks, and decision heuristics from 8 decades of investment literature.
10-sweep BEST research methodology with 40+ quantitative checks. Composite scoring, expectation gap analysis, Beneish M-Score, Piotroski F-Score, and VP modeling — with the load-bearing math (intrinsic-value anchor, verdict routing, scoring gates) computed by the deterministic Harvest engine, not improvised by the model.
FIRE planning integration, research cycle management, tax-advantaged routing, and market-condition deployment gates that prevent bad timing.
QC_PROTOCOL.md forces calculation scratchpads, NULL-on-unsourced data, and source citations on every workflow — so you can audit every number.
Every number below is server-computed — same inputs, same result, every run. This is a real quick screen PEP triage, with an honest wrinkle worth seeing: PepsiCo's dividend wasn't in the quick-triage feed, so the framework refused to certify a reject on a missing input — it won't call a HOLD on incomplete data. Supply PEP's real $5.75 dividend and the verdict is firm: even crediting it in full, expected return is 4.4% against the 13% bar, so RESEARCH_HOLD — a quality business declined at today's price.
Real quick screen PEP output (PEP's dividend supplied to the triage, which the quick feed didn't carry), shown for educational illustration; price and estimates as of the run. Figures reflect methodology application, not investment recommendations. Not financial advice.
A free, standalone preview of the scoring methodology on any ticker — 3 per day, no account. (This preview is a separate hosted tool; the Starter Edition runs in your own LLM, optimized for Claude, backed by the deterministic Harvest engine — see below.)
An illustration of how the methodology's design differs from unstructured investing. It describes the process, not a promised personal result — individual outcomes depend on your own execution and market conditions.
Verified buyer testimonials will appear here after launch. As a matter of policy, Harvest Protocol does not publish fabricated, composite, or "representative" quotes, and does not publish performance, return, or dollar figures in any testimonial — see Disclosures.
Harvest Protocol — Testimonial Policy
No testimonials are currently published. Not financial advice. Not a registered investment adviser.
Behavioral-finance research (Morningstar Mind the Gap 2024; DALBAR QAIB 2024) documents an average 1.5–2% annual return drag from preventable decision errors. That published research is the reason a disciplined process matters — it is context, not a benefit promised to any buyer. The Starter edition costs $127 once.
What you are buying is process discipline — fewer panic exits and FOMO entries, and a repeatable monthly cycle. Whether that is worth $127 to you is your decision; we make no projection of any dollar outcome.
The 1.5–2% behavior-gap figure is published third-party research (Morningstar Mind the Gap 2024; DALBAR QAIB 2024), cited as context for why process matters — not a guarantee, projection, or representation of any individual outcome. Not financial advice. Not a registered investment adviser.
Buy-side research desk: $2,400+/mo · Seeking Alpha Premium: $299/yr · Automated services on $250K AUM: $625/yr · HARVEST PROTOCOL STARTER: $127 once
START_HERE.html — setup, a command lookup, and troubleshooting on one pageNo Claude subscription required to start. Begin with a free Claude account. Upgrade to Pro ($20/month) when your workflow demands it.
30-day money-back guarantee · Personal use license · Not financial advice
QC_PROTOCOL.md forces a calculation_scratchpad (named operands, source-tagged values, visible arithmetic), a NULL-and-DATA_DEFICIENT rule instead of estimating an unsourced number, and a source citation on every figure. You remain the final auditor against the primary sources (10-K, 10-Q, earnings transcripts).
MODEL_ROUTING.md tells you which model fits which task: most daily workflows run on Sonnet; the adversarial workflows (Bear Case Stress Test, VP Self-Audit, full Monthly BEST Cycle) benefit from Opus's deeper judgment. The workflow set is deliberately economical for free-tier accounts: a fast Quick Screen triage (1–3 min, server tools carry the math) decides whether a ticker even earns the longer Deep Screen session, so heavy analyses are spent only on names that can actually qualify. Your research conversation runs in your own session, with the load-bearing math (fair-value anchor, verdict routing, scoring gates) computed by the deterministic Harvest engine — same inputs, same number, every run.
QC_PROTOCOL.md requires a calculation_scratchpad — named operands, source-tagged values, visible arithmetic — for any figure the model itself works, and each workflow prompt carries verification cues to spot-check (e.g., "F-Score total should equal the sum of 9 sub-checks", "EV must equal the weighted sum of bull/base/bear returns"). When something looks off, the recovery prompt is built in: "Re-derive that score from these named operands and cite the source for each input." A server composite is always an integer — a stray decimal is your tell that a number was hand-computed and should be re-run with the connector.
START_HERE.html — eight checkboxes that walk you through creating a Claude Project and uploading the 8 core methodology files; (2) register the Harvest engine connector (one-time, ~2 min) — no data keys required; optionally add a free key (FMP, Finnhub, Massive) for more depth; (3) type quick screen MSFT and read your first verified card. The same page carries a command lookup for everything else — Morning Watchdog, Sunday Brief, the monthly cycle — so there is nothing to memorize.
When markets fall 25%, most investors panic. The ones with a Deployment Gate, pre-committed exit-review criteria, and a VP disconfirmation signal don't panic—they execute their own framework.
Run one full monthly BEST cycle. If it doesn't change how you approach your next investment decision — full refund, no questions, you keep the Playbook.
That's the deal. We don't promise returns. We promise the methodology will change how you research — or we don't keep your money.
30-day money-back · Personal use license · All future v7.x updates included · Not financial advice. All investments involve risk.