edge-candidate-agent

Name: edge-candidate-agent
Author: mphinance

15stars3forks

How to install edge-candidate-agent?

mphinance/alpha-skills/edge-candidate-agent

$npx -y skills add mphinance/alpha-skills --skill edge-candidate-agent

Installs into the current project.

Summary

Generate and prioritize US equity long-side edge research tickets from EOD observations, then export pipeline-ready candidate specs for trade-strategy-pipeline Phase I. Use when users ask to turn hypotheses/anomalies into reproducible research tickets, convert validated ideas int

Files · 1

View on GitHub

SKILL.md

1	# Edge Candidate Agent
2
3	## Overview
4
5	Convert daily market observations into reproducible research tickets and Phase I-compatible candidate specs.
6	Prioritize signal quality and interface compatibility over aggressive strategy proliferation.
7	This skill can run end-to-end standalone, but in the split workflow it primarily serves the final export/validation stage.
8
9	## When to Use
10
11	- Convert market observations, anomalies, or hypotheses into structured research tickets.
12	- Run daily auto-detection to discover new edge candidates from EOD OHLCV and optional hints.
13	- Export validated tickets as `strategy.yaml` + `metadata.json` for `trade-strategy-pipeline` Phase I.
14	- Run preflight compatibility checks for `edge-finder-candidate/v1` before pipeline execution.
15
16	## Prerequisites
17
18	- Python 3.9+ with `PyYAML` installed.
19	- Access to the target `trade-strategy-pipeline` repository for schema/stage validation.
20	- `uv` available when running pipeline-managed validation via `--pipeline-root`.
21
22	## Output
23
24	- `strategies/<candidate_id>/strategy.yaml`: Phase I-compatible strategy spec.
25	- `strategies/<candidate_id>/metadata.json`: provenance metadata including interface version and ticket context.
26	- Validation status from `scripts/validate_candidate.py` (pass/fail + reasons).
27	- Daily detection artifacts:
28	- `daily_report.md`
29	- `market_summary.json`
30	- `anomalies.json`
31	- `watchlist.csv`
32	- `tickets/exportable/*.yaml`
33	- `tickets/research_only/*.yaml`
34
35	## Position in Split Workflow
36
37	Recommended split workflow:
38
39	1. `skills/edge-hint-extractor`: observations/news -> `hints.yaml`
40	2. `skills/edge-concept-synthesizer`: tickets/hints -> `edge_concepts.yaml`
41	3. `skills/edge-strategy-designer`: concepts -> `strategy_drafts` + exportable ticket YAML
42	4. `skills/edge-candidate-agent` (this skill): export + validate for pipeline handoff
43
44	## Workflow
45
46	1. Run auto-detection from EOD OHLCV:
47	- `skills/edge-candidate-agent/scripts/auto_detect_candidates.py`
48	- Optional: `--hints` for human ideation input
49	- Optional: `--llm-ideas-cmd` for external LLM ideation loop
50	2. Load the contract and mapping references:
51	- `references/pipeline_if_v1.md`
52	- `references/signal_mapping.md`
53	- `references/research_ticket_schema.md`
54	- `references/ideation_loop.md`
55	3. Build or update a research ticket using `references/research_ticket_schema.md`.
56	4. Export candidate artifacts with `skills/edge-candidate-agent/scripts/export_candidate.py`.
57	5. Validate interface and Phase I constraints with `skills/edge-candidate-agent/scripts/validate_candidate.py`.
58	6. Hand off candidate directory to `trade-strategy-pipeline` and run dry-run first.
59
60	## Quick Commands
61
62	Daily auto-detection (with optional export/validation):
63
64	```bash
65	python3 skills/edge-candidate-agent/scripts/auto_detect_candidates.py \
66	--ohlcv /path/to/ohlcv.parquet \
67	--output-dir reports/edge_candidate_auto \
68	--top-n 10 \
69	--hints path/to/hints.yaml \
70	--export-strategies-dir /path/to/trade-strategy-pipeline/strategies \
71	--pipeline-root /path/to/trade-strategy-pipeline
72	```
73
74	Create a candidate directory from a ticket:
75
76	```bash
77	python3 skills/edge-candidate-agent/scripts/export_candidate.py \
78	--ticket path/to/ticket.yaml \
79	--strategies-dir /path/to/trade-strategy-pipeline/strategies
80	```
81
82	Validate interface contract only:
83
84	```bash
85	python3 skills/edge-candidate-agent/scripts/validate_candidate.py \
86	--strategy /path/to/trade-strategy-pipeline/strategies/my_candidate_v1/strategy.yaml
87	```
88
89	Validate both interface contract and pipeline schema/stage rules:
90
91	```bash
92	python3 skills/edge-candidate-agent/scripts/validate_candidate.py \
93	--strategy /path/to/trade-strategy-pipeline/strategies/my_candidate_v1/strategy.yaml \
94	--pipeline-root /path/to/trade-strategy-pipeline \
95	--stage phase1
96	```
97
98	## Export Rules
99
100	- Keep `validation.method: full_sample`.
101	- Keep `validation.oos_ratio` omitted or `null`.
102	- Export only supported entry families for v1:
103	- `pivot_breakout` with `vcp_detection`
104	- `gap_up_continuation` with `gap_up_detection`
105	- Mark unsupported hypothesis families as research-only in ticket notes, not as export candidates.
106
107	## Guardrails
108
109	- Reject candidates that violate schema bounds (risk, exits, empty conditions).
110	- Reject candidate when folder name and `id` mismatch.
111	- Require deterministic metadata with `interface_version: edge-finder-candidate/v1`.
112	- Use `--dry-run` in pipeline before full execution.
113
114	## Resources
115
116	### `skills/edge-candidate-agent/scripts/export_candidate.py`
117	Generate `strategies/<candidate

People who liked this also viewed

6 picks

Type

edge-candidate-agent

15stars3forks

View on GitHub

How to install edge-candidate-agent?

mphinance/alpha-skills/edge-candidate-agent

$npx -y skills add mphinance/alpha-skills --skill edge-candidate-agent

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Edge Candidate Agent
2
3	## Overview
4
5	Convert daily market observations into reproducible research tickets and Phase I-compatible candidate specs.
6	Prioritize signal quality and interface compatibility over aggressive strategy proliferation.
7	This skill can run end-to-end standalone, but in the split workflow it primarily serves the final export/validation stage.
8
9	## When to Use
10
11	- Convert market observations, anomalies, or hypotheses into structured research tickets.
12	- Run daily auto-detection to discover new edge candidates from EOD OHLCV and optional hints.
13	- Export validated tickets as `strategy.yaml` + `metadata.json` for `trade-strategy-pipeline` Phase I.
14	- Run preflight compatibility checks for `edge-finder-candidate/v1` before pipeline execution.
15
16	## Prerequisites
17
18	- Python 3.9+ with `PyYAML` installed.
19	- Access to the target `trade-strategy-pipeline` repository for schema/stage validation.
20	- `uv` available when running pipeline-managed validation via `--pipeline-root`.
21
22	## Output
23
24	- `strategies/<candidate_id>/strategy.yaml`: Phase I-compatible strategy spec.
25	- `strategies/<candidate_id>/metadata.json`: provenance metadata including interface version and ticket context.
26	- Validation status from `scripts/validate_candidate.py` (pass/fail + reasons).
27	- Daily detection artifacts:
28	- `daily_report.md`
29	- `market_summary.json`
30	- `anomalies.json`
31	- `watchlist.csv`
32	- `tickets/exportable/*.yaml`
33	- `tickets/research_only/*.yaml`
34
35	## Position in Split Workflow
36
37	Recommended split workflow:
38
39	1. `skills/edge-hint-extractor`: observations/news -> `hints.yaml`
40	2. `skills/edge-concept-synthesizer`: tickets/hints -> `edge_concepts.yaml`
41	3. `skills/edge-strategy-designer`: concepts -> `strategy_drafts` + exportable ticket YAML
42	4. `skills/edge-candidate-agent` (this skill): export + validate for pipeline handoff
43
44	## Workflow
45
46	1. Run auto-detection from EOD OHLCV:
47	- `skills/edge-candidate-agent/scripts/auto_detect_candidates.py`
48	- Optional: `--hints` for human ideation input
49	- Optional: `--llm-ideas-cmd` for external LLM ideation loop
50	2. Load the contract and mapping references:
51	- `references/pipeline_if_v1.md`
52	- `references/signal_mapping.md`
53	- `references/research_ticket_schema.md`
54	- `references/ideation_loop.md`
55	3. Build or update a research ticket using `references/research_ticket_schema.md`.
56	4. Export candidate artifacts with `skills/edge-candidate-agent/scripts/export_candidate.py`.
57	5. Validate interface and Phase I constraints with `skills/edge-candidate-agent/scripts/validate_candidate.py`.
58	6. Hand off candidate directory to `trade-strategy-pipeline` and run dry-run first.
59
60	## Quick Commands
61
62	Daily auto-detection (with optional export/validation):
63
64	```bash
65	python3 skills/edge-candidate-agent/scripts/auto_detect_candidates.py \
66	--ohlcv /path/to/ohlcv.parquet \
67	--output-dir reports/edge_candidate_auto \
68	--top-n 10 \
69	--hints path/to/hints.yaml \
70	--export-strategies-dir /path/to/trade-strategy-pipeline/strategies \
71	--pipeline-root /path/to/trade-strategy-pipeline
72	```
73
74	Create a candidate directory from a ticket:
75
76	```bash
77	python3 skills/edge-candidate-agent/scripts/export_candidate.py \
78	--ticket path/to/ticket.yaml \
79	--strategies-dir /path/to/trade-strategy-pipeline/strategies
80	```
81
82	Validate interface contract only:
83
84	```bash
85	python3 skills/edge-candidate-agent/scripts/validate_candidate.py \
86	--strategy /path/to/trade-strategy-pipeline/strategies/my_candidate_v1/strategy.yaml
87	```
88
89	Validate both interface contract and pipeline schema/stage rules:
90
91	```bash
92	python3 skills/edge-candidate-agent/scripts/validate_candidate.py \
93	--strategy /path/to/trade-strategy-pipeline/strategies/my_candidate_v1/strategy.yaml \
94	--pipeline-root /path/to/trade-strategy-pipeline \
95	--stage phase1
96	```
97
98	## Export Rules
99
100	- Keep `validation.method: full_sample`.
101	- Keep `validation.oos_ratio` omitted or `null`.
102	- Export only supported entry families for v1:
103	- `pivot_breakout` with `vcp_detection`
104	- `gap_up_continuation` with `gap_up_detection`
105	- Mark unsupported hypothesis families as research-only in ticket notes, not as export candidates.
106
107	## Guardrails
108
109	- Reject candidates that violate schema bounds (risk, exits, empty conditions).
110	- Reject candidate when folder name and `id` mismatch.
111	- Require deterministic metadata with `interface_version: edge-finder-candidate/v1`.
112	- Use `--dry-run` in pipeline before full execution.
113
114	## Resources
115
116	### `skills/edge-candidate-agent/scripts/export_candidate.py`
117	Generate `strategies/<candidate

People who liked this also viewed

6 picks

Type