Files

hjnormey 36d263eb91 feat(config/scripts): GREEN config wiring + S6 recompute tooling

prod/configs/green.yml:
- asset_bucket_ban_set: [4] (B4 banned at selector level)
- s6_size_table: inline bootstrap multipliers (B0→0.4, B1→0.3, B3→2.0,
  B5→0.5, B6→1.5) matching CRITICAL_ASSET_PICKING S6 scenario
- esof_sizing_table: FAV→1.2, MILD_POS→0.6, UNKNOWN→0.25,
  MILD_NEG→0.0, UNFAV→0.0
- use_int_leverage: true (1x fixed pending winrate analysis)
- s6_table_path: pointer to generated YAML (recompute updates this)
BLUE (blue.yml) carries none of these keys → BLUE math unchanged.

prod/configs/green_s6_table.yml: bootstrap stub with frontmatter
(generated_at, source_branch, n_trades). Regenerated by recompute script.

prod/scripts/recompute_s6_coefficients.py:
  Queries trade_events, maps assets to KMeans buckets, derives per-bucket
  sizing mults. Variance guard: >20% net-PnL move flags bucket in
  dolphin.s6_recompute_log for manual review before promote.

prod/s6_recompute_flow.py:
  Prefect flow wrapping the recompute script. Cadence via
  S6_RECOMPUTE_INTERVAL_DAYS env (default 30). Kill-switch:
  S6_RECOMPUTE_DISABLED=1.

prod/scripts/analyze_leverage_winrate.py:
  Read-only walk of CH trade_events; bins trades by leverage_raw,
  emits per-bin WR/net-PnL/avg-MAE. Output informs the int-leverage
  rounding rule choice (Option 1 round-half-up vs Option 2 banker's round
  vs stay-at-1x). Does not auto-apply a rule change.

Plan refs: Tasks 3, 8, 10.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-04-22 06:08:08 +02:00

2.2 KiB

Raw Blame History

DOLPHIN Branching Convention

Established 2026-04-21. The DOLPHIN repo (Gitea: 100.119.158.61:3000/hjnormey/DOLPHIN) uses a fork-per-experiment pattern to support parallel multi-algo benchmarking on shared infrastructure (ClickHouse, HZ, MHS).

Branches

Branch	Role
`master`	Production baseline. Only merged-and-validated work lands here.
`exp/<slug>`	Experiment branches. Named `exp/<feature>-<YYYY-MM-DD>`.

Merge to master only after the experiment has passed staging validation (integration replay reproduces expected counterfactual, parity tests pass, 24h staging monitor clean).

Active / recent experiments

exp/green-s6-esof-aem-shadow-2026-04-21 — S6 asset picker, EsoF regime gate (NEUTRAL→UNKNOWN), AEM shadow completeness, integer-leverage gate.

Infrastructure suffixing (avoid collisions on shared CH / HZ / MHS)

When an experiment writes runtime state to shared infra, suffix the table or map name with the branch slug so parallel experiments don't stomp each other.

Examples:

ClickHouse tables: adaptive_exit_shadow_exp_green_s6, trade_events_exp_green_s6
HZ maps: DOLPHIN_PNL_GREEN_S6, DOLPHIN_FEATURES_EXP_GREEN_S6
Supervisord program names: dolphin-green-exp-s6

The suffix slug is the part of the branch name after exp/, with / → _ and - → _ as needed.

Workflow

Branch from master: git checkout master && git pull && git checkout -b exp/<feature>-<YYYY-MM-DD>.
All experiment code, configs, CH DDL, supervisord entries, and docs land on the experiment branch.
Staging deploy runs on the experiment branch only.
Validation + monitor period.
If the experiment is kept, open a merge request to master. If it's abandoned, leave the branch for reference — do not delete immediately (post-mortem value).

Commit identity

Commits should use the operator's Gitea-bound identity (hjnormey@gmail.com). Agents making commits should pass identity per-command via git -c user.email=... -c user.name=... rather than mutating global config.

2.2 KiB Raw Blame History