Lab / Systems
Make the process inspectable.
A personal proprietary research lab means building a process for Josh’s own decisions—not managing outside capital. This is the architecture, with its current limits exposed.
01 / Evidence
AvailableData and research intake
Research pages are published · no live data feed
Start with a question and inspect the source, including what is uncertain or missing, before a rule is treated as a finding. This is the data and research intake for the lab—not a market signal.
- Inputs
- A written research question, source URLs and timestamps, and public chart study used as education (including the technical examples drawn from public Kraken OHLC).
- Outputs
- Notes that separate direct evidence from interpretation, with uncertainty left in view. This stage does not issue a trade idea or a signal.
- Tools on this site
- The research standard and research pages, the educational technical-example charts, and the digital-assets study page. The public site publishes documents; it does not ingest private account balances or run a live market-data feed.
- Current status
- Those research pages and educational charts are on the site today. No research bot, model-powered journal, or live data collection is connected.
Research →Research standard →Technical examples →Digital-assets study →
Implemented on this site
Tools you can inspect today
Risk calculations
Position sizing from the stop, risk/reward and drawdown education. These are learning tools, not orders or personalized recommendations.
Open the calculators →Written research rules
Four explicitly untested playbooks, shared core rules and a publication checklist. None has a verified performance record here.
Inspect the playbooks →Public learning infrastructure
Ten lessons, a glossary, a practice lab and browser-local lesson completion. No member account is required; progress does not sync across devices.
Read the course →Static publishing
Astro pages, local search and a blog RSS feed. The public system publishes documents; it does not place trades or ingest private account balances.
Read the implementation note →Proposed · not connected
The AI and automation layer
“AI-driven” is the project direction, not a claim that these services are running. No research bot, model-powered journal or autonomous execution system is connected to the public site.
- Research assistant: collect source URLs and timestamps, separate direct evidence from interpretation, and flag missing or contradictory claims. Every output needs human source checks.
- Journal review: compare a future decision record with the rule version in force, then surface missing fields and possible process deviations. Never rewrite an earlier thesis to fit the outcome.
- First-principles tools: translate an idea into measurable conditions, cost assumptions and a falsification test. A model suggestion is not a validated trading rule.
Before any integration: define data permissions, retain source lineage, test failure behavior, protect private records and approve scope. No automatic trading is offered.
Design specification · not a populated journal
A decision record that can survive a loss
- Before
- Timestamp, source evidence, hypothesis, invalidation, planned risk in R, expected costs and rule version.
- After
- Actual outcome, fees, slippage, rule adherence, loss classification and the next test—not a revised past.
No trade records are supplied or invented on this page. Future real, paper and backtested records must stay distinctly labeled.
The full publication standard →Current boundaries
Accounts, collection, payments, community posting and automated execution are off. Membership plans are descriptions of a future product, not working entitlements. See the membership plan.