Research tools, runners, validators, diagnostics, and proof-board generators.
Trading Lane Presentation
What We Actually Built In The Trading Lab
A public snapshot of a broader trading research and operations platform: runtime engines, lattice systems, adaptive controller research, proof boards, supervision infrastructure, and focused validation. The goal is not to sell a black-box trading product. It is to show the scope and quality of the research-and-systems work while keeping credentials, runtime state, and generated logs out of GitHub.
GitHub Actions verifies the highest-signal Kraken maker shadow unit surface.
Curated documentation for protocol, proof, quickstart, performance, and review.
Browser-readable presentation surface linked from the GitHub sidebar.
Public portfolio snapshot with clear reuse terms and explicit boundaries.
Core Achievements
The trading lane is a stack of research capabilities, not just a collection of strategy files.
Core Runtime
Built
Trading Engine And Bot Runtime
Built a full MT5 runtime with supervisor, worker, config, exit engine, learning loop, health monitor, and extracted bot modules.
- Supervisor + worker split
- Dedicated exit logic and health checking
- Learning system with hot-streak sizing and ATR-scaled profit logic
Penetration Lattice
Core Research
Stopless Lattice Research Program
The main lattice lane turned oscillation harvesting into a serious program with engine code, doctrine, live runners, and runner-modeled evidence.
- Tick-native lattice engine and live/shadow runners
- Universal step thinking around range and ATR
- Broker-truth emphasis over fantasy backtests
Adaptive Research
Original Methodology
Hungry Hippo And Adaptive Lattice
Built a layered adaptive research stack with regime reading, shape libraries, controller scaffolding, and explicit gap doctrine.
- Regime-aware geometry and posture ideas
- Transfer priors and candidate libraries
- Explicit admission that the controller is not fully validated yet
Competition
Operationalized
Competition And Promotion Framework
Systematically tested competing variants and promoted winners instead of managing a pile of one-off experiments.
- Anchor and close-style competitions
- Family-local benchmarks and breakthrough dockets
- Shadow-to-live graduation rules with anti-promotion firewalls
Supervision
24/7 Ops
Watchdogs, Registry, And Shared Price Infrastructure
Built the operations layer needed to keep many lanes alive without silent drift, duplicate processes, or stale runtime ambiguity.
- Watchdog groups and runner registry
- MT5 terminal guard and process singletons
- Shared price feeder and runtime drift reports
Operator Surfaces
High Visibility
Board And Dashboard Ecosystem
Generated thousands of board files so the lane can be operated through explicit authority surfaces instead of memory or guesswork.
- Health, PnL, seat, concentration, readiness, and proof boards
- Per-symbol live-seat truth
- Compact reports for real operator questions
Key Discoveries And Durable Lessons
Some of these are stronger than others. The point is to show disciplined findings, not to pretend they are all universal laws.
House Discovery
Strong Internal Evidence
ATR Threshold Discovery
Crypto lattice steps below 1.0x ATR repeatedly went negative EV in the warp family. That became a durable internal rule rather than another tunable preference.
Internal Finding
Validated In Lane
M5 Timeframe Dominance
M5 repeatedly outperformed M15 and H1 on money velocity and reset behavior in the strongest mean-reversion families.
Durable Lesson
Operational Rule
Spread-To-Step Ratio Matters
Spread friction regularly destroyed edges that looked acceptable in looser or replay-style evaluation.
Research Discipline
Strong Signal
Forward Shadow Beats Replay
Bar-replay leaders degraded or failed when moved to honest forward conditions. The lane treats forward shadow as the real gate.
Unsolved
Important
The Cover Problem
The repo repeatedly concludes that realized flow can still fail to stay ahead of floating burden. This remains one of the biggest open problems.
Unsolved
Important
Composition Is Hard
The best sleeves and the best local doctrines do not necessarily compose into the best whole. The lane discovered that repeatedly in ratio and universal-doctrine work.
Capability Timeline
This is the story arc you were missing: not just one controller, but the evolution from basic engines into a research organism.
Early lane
Built the core MT5 bot runtime
Supervisor, worker, exits, config loading, learning hooks, and health surfaces.
Lattice phase
Turned the penetration lattice into a real research family
Canonical engine, live and shadow runners, geometry doctrine, and step/close experiments.
Competition phase
Operationalized tournaments and promotion logic
Staged-anchor competitions, family-local benchmarks, and shadow-to-live rules.
Operator phase
Built supervision, registry, watchdog, and board infrastructure
The lane became operable as a multi-runner system rather than a code pile.
Adaptive phase
Started the state-aware control research program
Profit modes, unified objective, adaptive shape libraries, proof boards, and explicit foundational gap doctrine.
How The Lane Works
These diagrams now sit inside a broader achievement story instead of trying to carry the whole presentation on their own.
System Map
flowchart LR
subgraph Inputs
A[Bars and quotes]
B[Spread and range]
C[Open burst events]
D[First-path telemetry]
E[Runtime state files]
end
subgraph Controller
F[Profit mode classifier]
G[Adaptive lattice controller]
H[Unified objective]
I[Survival constraints]
J[Shape recommendation]
end
subgraph Runtime
K[Runtime overlays]
L[Shadow or live lane]
M[State and event logs]
end
subgraph Proof
N[Proof watch boards]
O[Incumbent vs challenger studies]
P[Promotion or parking decision]
end
A --> F
B --> F
C --> F
D --> G
E --> G
F --> G
G --> H
G --> I
H --> J
I --> J
J --> K
K --> L
L --> M
M --> N
M --> O
N --> P
O --> P
Proof Discipline
flowchart LR
A[Runtime events and state logs] --> B[Path-quality extraction]
B --> C[First-path verdict]
B --> D[Market-state hypothesis]
B --> E[Close and carry metrics]
C --> F[Proof watch board]
D --> F
E --> G[Unified objective validation]
E --> H[Joint control law study]
F --> I{Evidence supports promotion?}
G --> I
H --> I
I -->|No| J[Keep negative evidence visible]
I -->|No| K[Do not overclaim]
I -->|Yes| L[Promote challenger]
Representative Edges And Families
These rows are there to show breadth and experimentation, not to claim every family is currently dominant.
| Family | What Was Built | Representative Result Or Lesson | Why It Matters |
|---|---|---|---|
| Warp | High-throughput lattice family optimized for closes/hour | BTC and selected crypto lanes became major modeled candidates; ATR floor emerged as a durable rule | Shows the lane can discover and refine a family instead of just tuning one row |
| Hungry Hippo | Regime-aware adaptive lattice framework with per-symbol configs and audits | Some symbols improved strongly, others showed zero improvement, which sharpened the adaptive doctrine | Shows serious adaptive research instead of vague “AI” language |
| Ratio Lattice | Coinbase spot ratio-trading sleeve research with walk-forward checks | Good sleeves did not compose cleanly; denominator contention became a real finding | Shows the repo learns from composition failure, not just single-lane wins |
| Snake / Micro-Harvest | Extreme-frequency M1 family with bootstrap and margin realism | Raw ceiling looked attractive, but bootstrap and margin reality killed most honest rows | Shows the lane knows how to reject seductive but false edges |
| Kelly / Allocation | Portfolio sizing across modeled spot candidates | Allocation became part of the organism, not an afterthought | Shows portfolio-level systems thinking beyond strategy discovery |
Infrastructure Around The Research
The repo is stronger because the research lane is surrounded by operational infrastructure.
Operations
Supervision And Runtime Safety
- Watchdog groups and runner registries
- MT5 terminal guard and duplicate-launch suppression
- Shared price feeder and runtime drift reporting
- Closure firewall and affordability gate logic
Memory
Durable Truth Surfaces
- Current assumptions, modeled candidates, active risks, and lane state
- Session logs and review notes kept outside the public snapshot when they contain local state
- A real separation between public proof surfaces and private operational context
Coordination
Multi-Agent Research Support
- Switchboard communication layer
- Agent coordination surfaces and discovery logs
- Parallel research and monitoring support
Presentation
Board-First Operator UX
- Thousands of generated reports and status boards
- Authority-stack thinking for answering runtime questions quickly
- Readable planning and readiness surfaces for each seam
Why This Is Actually Impressive
The impressive part is not just the code count. It is the combination of research ambition and operational honesty.
Strong Signal
What Makes The Lane Strong
- Real doctrine instead of prompt-level slogans
- Negative evidence is kept visible
- Promotion is gated through shadow, benchmarks, and guardrails
- There is extensive operational infrastructure around the research
- The lane repeatedly turns unclear behavior into explicit rules and boards
Boundary
What Not To Claim
- Do not present it as a finished money-printing machine
- Do not claim the adaptive lane is already proven superior
- Do not market every internal finding as field-novel research
- Do present it as a serious research and control platform under uncertainty