New to finance or coding
Learn what positions, exposure, stop-losses and short selling mean. Follow small worked examples, then give your coding agent a brief that asks you to choose your own settings.
Kronos / ForgeMesh
For humans · ebook + digital build package
The build kit that shows what broke.
Build your own Kronos market-intelligence and paper-trading system. Understand the settings, choose your rules, hand the brief to your coding agent, and learn from the failures in our own setup.
Read a free preview — Chapter 2 ↗Educational only. Not financial, investment, legal or tax advice. Trading can lose all committed capital; borrowing, shorting and derivatives can create additional liabilities. No forecast, stop or configuration guarantees protection or profit.
Two ways into the same system
Start with the knowledge you have.
Financially experienced
Define a mandate, risk budget, execution policy and benchmark. Examine cost drag, overlapping evidence, calibration and replay fidelity. Coding experience is not assumed.
Both paths include hardware and VM planning, a model-installation walkthrough, the official research-paper link, a paper-operation runbook and implementation acceptance scenarios. Model weights and upstream source are downloaded separately. The package is a specification and learning resource, not a finished trading application.
Frozen evidence · September 7, 2026
The losses and uncertainty stay in the record.
Our captured forecast outcomes show 49.8% directional accuracy, not a demonstrated profitable edge. The latest captured paper-account value was $94,840.62 versus an initial $100,000: a simple 5.2% difference below the start, not an audited or cash-flow-adjusted strategy return.
The raw log records a $40,367.32 peak-to-trough difference, but the lowest value rebounded the following hour. The trough remains unreconciled. It is not established as a war-caused realized loss. Lines break across observation gaps longer than six hours.
Simulated-performance disclosure: The account figures shown are recorded paper-account observations, not actual investment returns. Simulated and retrospective studies have material limitations: execution assumptions, missing liquidity effects, imperfect fees and slippage, hindsight, selection bias, changing parameters and dependence among observations. A successful simulation does not demonstrate that a real account would achieve similar results. Past observations do not predict future results.
The case studies explain the trailing-peak reset bug, rejected broker orders, position deadlocks and a replay that could not reproduce the recorded decisions. Changing your settings changes your build and its outcomes. Copying ours does not promise the same results.
Read before purchasing
A self-service educational purchase.
No support of any kind.
No installation help, debugging, configuration reviews, trading advice, calls, community access, maintenance, updates or response-time promise. You need to work independently or with your own coding agent. Third-party hosting, data and optional APIs may cost extra.
Personal, non-transferable use. No redistribution. Sold as-is, subject to non-excludable rights. Refunds follow the applicable checkout platform policy disclosed at purchase and mandatory law; no additional voluntary refund program is promised.
Read the complete personal-use license and purchase termsAffiliate disclosure: The guide contains optional DigitalOcean links through which GSD Contracts LLC / ForgeMesh may earn a commission. A non-affiliate alternative is provided. Infrastructure choices are optional and should follow your own requirements. No Alpaca referral relationship is represented.
Need the agent API instead? Return to Kronos market intelligence.