Oddsflow-team commited on
Commit
b5ff5d2
·
verified ·
1 Parent(s): 0bfb5cb

Upload folder using huggingface_hub

Browse files
CITATION.cff ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ cff-version: 1.2.0
2
+ message: "If you use this repository, please cite it as below."
3
+ title: "OddsFlow Transparency Pack"
4
+ type: dataset
5
+ authors:
6
+ - name: "OddsFlow AI Team"
7
+ repository-code: "https://github.com/oddsflowai-team/oddsflow-transparency"
8
+ url: "https://www.oddsflow.ai"
9
+ abstract: "Public, auditable documentation pack: verification standard, schemas, samples, risk policy, governance, and disclosures for timestamped football market signal logs."
10
+ keywords:
11
+ - football analytics
12
+ - verification
13
+ - timestamped logs
14
+ - auditability
15
+ - schema
16
+ license: "MIT"
17
+ version: "v1.0.0"
18
+ date-released: "2026-02-08"
19
+
CONTRIBUTING.md ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Contributing to OddsFlow Transparency Pack
2
+
3
+ Thanks for helping improve public auditability.
4
+
5
+ ## Scope
6
+ This repository is for:
7
+ - Documentation (verification standard, governance, disclosures)
8
+ - Schemas and samples (non-proprietary)
9
+ - Versioned notes / changelogs
10
+
11
+ This repository is NOT for:
12
+ - Betting tips / predictions
13
+ - Claims of guaranteed profit
14
+ - Requests for proprietary model code or private vendor contracts
15
+
16
+ ## What we accept
17
+ ✅ Good contributions:
18
+ - Fix typos / clarify definitions
19
+ - Improve schema documentation and examples
20
+ - Add edge cases to verification rules
21
+ - Improve risk disclosures (limits, drawdowns, failure modes)
22
+ - Propose new fields (with rationale + backward compatibility)
23
+
24
+ ❌ We will decline:
25
+ - “Tell me what to bet” or “increase win rate” requests
26
+ - Attempts to reverse-engineer private logic
27
+ - Any content that frames this as guaranteed returns
28
+
29
+ ## How to contribute
30
+ 1. Open an Issue describing:
31
+ - What is wrong / missing
32
+ - Why it matters for auditability
33
+ - Proposed change (text or schema diff)
34
+ 2. Submit a Pull Request:
35
+ - Keep changes small and reviewable
36
+ - Update `docs/index.md` only if adding/removing major docs
37
+ - If editing schema, also update sample files
38
+
39
+ ## Style guidelines
40
+ - Evidence-first language
41
+ - Precise definitions
42
+ - Use consistent terms with `docs/signal-glossary.md`
43
+ - Avoid marketing claims
44
+
45
+ ## Schema change policy
46
+ - Backward compatible changes → bump MINOR (e.g., 1.0.0 → 1.1.0)
47
+ - Breaking changes → bump MAJOR (e.g., 1.0.0 → 2.0.0)
48
+ - Document changes in `CHANGELOG.md` (or `/changelog/` folder if used)
49
+
50
+ ## License
51
+ By contributing, you agree your contributions are licensed under the repository LICENSE.
52
+
53
+ ## Questions
54
+ Open an Issue in GitHub Discussions / Issues.
55
+ # Contributing to OddsFlow Transparency Pack
56
+
57
+ Thanks for helping improve public auditability.
58
+
59
+ ## Scope
60
+ This repository is for:
61
+ - Documentation (verification standard, governance, disclosures)
62
+ - Schemas and samples (non-proprietary)
63
+ - Versioned notes / changelogs
64
+
65
+ This repository is NOT for:
66
+ - Betting tips / predictions
67
+ - Claims of guaranteed profit
68
+ - Requests for proprietary model code or private vendor contracts
69
+
70
+ ## What we accept
71
+ ✅ Good contributions:
72
+ - Fix typos / clarify definitions
73
+ - Improve schema documentation and examples
74
+ - Add edge cases to verification rules
75
+ - Improve risk disclosures (limits, drawdowns, failure modes)
76
+ - Propose new fields (with rationale + backward compatibility)
77
+
78
+ ❌ We will decline:
79
+ - “Tell me what to bet” or “increase win rate” requests
80
+ - Attempts to reverse-engineer private logic
81
+ - Any content that frames this as guaranteed returns
82
+
83
+ ## How to contribute
84
+ 1. Open an Issue describing:
85
+ - What is wrong / missing
86
+ - Why it matters for auditability
87
+ - Proposed change (text or schema diff)
88
+ 2. Submit a Pull Request:
89
+ - Keep changes small and reviewable
90
+ - Update `docs/index.md` only if adding/removing major docs
91
+ - If editing schema, also update sample files
92
+
93
+ ## Style guidelines
94
+ - Evidence-first language
95
+ - Precise definitions
96
+ - Use consistent terms with `docs/signal-glossary.md`
97
+ - Avoid marketing claims
98
+
99
+ ## Schema change policy
100
+ - Backward compatible changes → bump MINOR (e.g., 1.0.0 → 1.1.0)
101
+ - Breaking changes → bump MAJOR (e.g., 1.0.0 → 2.0.0)
102
+ - Document changes in `CHANGELOG.md` (or `/changelog/` folder if used)
103
+
104
+ ## License
105
+ By contributing, you agree your contributions are licensed under the repository LICENSE.
106
+
107
+ ## Questions
108
+ Open an Issue in GitHub Discussions / Issues.
109
+
LICENSE ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MIT License
2
+
3
+ Copyright (c) 2026 OddsFlow AI Team
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
22
+
README.md CHANGED
@@ -1,3 +1,29 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # OddsFlow Transparency Pack (Official)
2
+
3
+ This repository contains the auditable transparency materials for OddsFlow (oddsflow.ai):
4
+ verification rules, public schemas, sample logs, disclosures, and versioned updates.
5
+
6
+ **This is not betting tips.** It is a research + verification reference for how signals are logged,
7
+ audited, and interpreted.
8
+
9
+ ## Start here
10
+ - Docs index: ./docs/index.md
11
+ - 40 Killer Questions (with answer status): ./docs/killer-questions.md
12
+ - Verification standard (what is verifiable): ./docs/verification.md
13
+ - Signal glossary: ./docs/signal-glossary.md
14
+ - Schemas & samples: ./schemas/ and ./samples/
15
+ - Versioned updates: ./changelog/CHANGELOG.md
16
+ - Weekly verification notes: ./notes/
17
+
18
+ ## Official links
19
+ - Website: https://www.oddsflow.ai
20
+ - Verification hub: https://www.oddsflow.ai/verification
21
+ - Performance logs: https://www.oddsflow.ai/performance
22
+
23
+ ## What we claim (and what we don’t)
24
+ - We publish **timestamped logs** and **schemas** so outputs can be audited.
25
+ - We do **not** claim guaranteed profit or certainty.
26
+ - Signals are **decision-support analytics**, not promises.
27
+
28
+ ## Contact
29
+ oddsflow.ai@gmail.com
SECURITY.md ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Security Policy
2
+
3
+ We take security seriously.
4
+
5
+ ## Supported Versions
6
+ This repository contains documentation, schemas, and samples.
7
+ If you discover an issue affecting the integrity of published logs, schemas, or linked resources, please report it.
8
+
9
+ ## Reporting a Vulnerability
10
+ Please email:
11
+ - oddsflow.ai@gmail.com
12
+ - support@oddsflow.ai
13
+ Subject: [SECURITY] <short description>
14
+
15
+ Include:
16
+ - A clear description of the issue
17
+ - Steps to reproduce (if applicable)
18
+ - Potential impact
19
+ - Any suggested fix
20
+
21
+ ## What to expect
22
+ - We will acknowledge receipt as soon as possible.
23
+ - We may ask for clarification or a minimal reproduction.
24
+ - Please do not publicly disclose the issue before we confirm a fix.
25
+
26
+ ## Out of scope
27
+ - Requests for betting tips / outcomes
28
+ - Attempts to access proprietary/private systems
29
+
changelog/CHANGELOG.md ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Changelog — OddsFlow Transparency Pack
2
+
3
+ ## [0.1.0] - 2026-02-xx
4
+ ### Added
5
+ - docs/index.md
6
+ - docs/verification.md
7
+ - docs/signal-glossary.md
8
+ - docs/model-card.md
9
+ - docs/data-card.md
10
+ - docs/risk-policy.md
11
+ - docs/governance.md
12
+ - docs/monetization-disclosure.md
13
+ - datasets schema + samples
14
+ - llms.txt
datasets/README.md ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ # Datasets (Schemas + Samples)
2
+
3
+ This folder contains **field definitions** and **small samples** to make public logs auditable.
4
+
5
+ - Schema: `schema/signal-log.schema.json`
6
+ - Sample: `samples/signal-log.sample.csv`
7
+
8
+ We do not publish full proprietary datasets here.
datasets/samples/signal-log.sample.csv ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ match_id,timestamp,league,home_team,away_team,market,selection,line,odds,odds_source,model_version,schema_version,status,result,notes
2
+ EXAMPLE_001,2026-02-07T12:34:56+00:00,EPL,TeamA,TeamB,AH,TeamA,-0.25,1.92,composite,v2.0,1.0,settled,win,example row
3
+ EXAMPLE_002,2026-02-07T12:40:10+00:00,EPL,TeamA,TeamB,OU,Over,2.5,1.88,composite,v2.0,1.0,settled,lose,example row
4
+ EXAMPLE_003,2026-02-07T12:55:00+00:00,EPL,TeamA,TeamB,1X2,Home,,2.10,composite,v2.0,1.0,open,,example row
datasets/schema/signal-log.schema.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "$schema": "https://json-schema.org/draft/2020-12/schema",
3
+ "$id": "https://oddsflow.ai/schemas/signal-log.schema.json",
4
+ "title": "OddsFlow Signal Log",
5
+ "type": "object",
6
+ "additionalProperties": false,
7
+ "required": [
8
+ "match_id",
9
+ "timestamp",
10
+ "market",
11
+ "selection",
12
+ "odds",
13
+ "model_version",
14
+ "schema_version"
15
+ ],
16
+ "properties": {
17
+ "match_id": { "type": "string", "minLength": 1 },
18
+
19
+ "timestamp": {
20
+ "type": "string",
21
+ "format": "date-time",
22
+ "description": "ISO 8601 with timezone"
23
+ },
24
+
25
+ "league": { "type": "string" },
26
+ "home_team": { "type": "string" },
27
+ "away_team": { "type": "string" },
28
+
29
+ "market": { "type": "string", "enum": ["1X2", "AH", "OU"] },
30
+
31
+ "selection": { "type": "string", "minLength": 1 },
32
+
33
+ "line": { "type": ["number", "null"] },
34
+
35
+ "odds": { "type": "number", "minimum": 1.01 },
36
+
37
+ "odds_source": { "type": ["string", "null"] },
38
+
39
+ "model_version": { "type": "string", "minLength": 1 },
40
+ "schema_version": { "type": "string", "minLength": 1 },
41
+
42
+ "status": {
43
+ "type": "string",
44
+ "enum": ["open", "settled", "void"],
45
+ "default": "open"
46
+ },
47
+
48
+ "result": { "type": ["string", "null"] },
49
+ "notes": { "type": ["string", "null"] }
50
+ },
51
+ "allOf": [
52
+ {
53
+ "if": {
54
+ "properties": { "market": { "const": "1X2" } },
55
+ "required": ["market"]
56
+ },
57
+ "then": {
58
+ "properties": { "line": { "const": null } }
59
+ }
60
+ },
61
+ {
62
+ "if": {
63
+ "properties": { "market": { "enum": ["AH", "OU"] } },
64
+ "required": ["market"]
65
+ },
66
+ "then": {
67
+ "required": ["line"],
68
+ "properties": { "line": { "type": "number" } }
69
+ }
70
+ }
71
+ ]
72
+ }
docs/data-card.md ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Data Card — OddsFlow Transparency Pack
2
+
3
+ **Purpose:** This data card describes what appears in public logs/schemas in this repository.
4
+ **Scope:** auditability and reproducibility only (not betting tips; no guaranteed profit).
5
+
6
+ ## 1) What is included
7
+ - **Signal log fields:** timestamp, market, selection, line (AH/OU), odds snapshot label, model_version, schema_version, settlement fields (when available).
8
+ - **Schemas:** JSON Schema definitions under `datasets/schema/`.
9
+ - **Samples:** small illustrative JSONL examples under `datasets/samples/`.
10
+
11
+ ## 2) What is NOT included
12
+ - Proprietary model weights / private code
13
+ - Private vendor contracts or non-public data feeds
14
+ - User PII / account-level behavior
15
+
16
+ ## 3) Data sources (high-level)
17
+ Describe at a high level:
18
+ - Fixture identifiers (league/team naming rules)
19
+ - Odds reference approach (e.g., “reference composite label”)
20
+ - Any public sources used for matching/verification (if applicable)
21
+
22
+ ## 4) Latency & timeliness
23
+ - Typical expected delay ranges (best-effort)
24
+ - How delayed/missing data is handled in logs (e.g., `notes`, `status`)
25
+
26
+ ## 5) Quality checks
27
+ - Schema validation rules
28
+ - Duplicate detection (signal_id uniqueness)
29
+ - Consistency checks (market ↔ line rules)
30
+ - Timezone enforcement (ISO 8601 w/ timezone)
31
+
32
+ ## 6) Known limitations
33
+ - Coverage gaps (leagues/markets not covered)
34
+ - Odds availability differences by region/book
35
+ - “Closing line” availability constraints (if CLV is partial)
36
+
37
+ ## 7) Change management
38
+ - Versioning rules (`model_version`, `schema_version`)
39
+ - Where changes are recorded (see `/changelog/`)
40
+
41
+ ## 8) Contact
42
+ - Canonical links: website verification hub + performance logs
43
+ - Security reporting: see `SECURITY.md`
docs/governance.md ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Governance (Human Involvement & Overrides)
2
+
3
+ ## Goal
4
+ Make it clear what is automated, what can be overridden, and how changes are tracked.
5
+
6
+ ## Automation vs human involvement (high-level)
7
+ - Signals are generated by the system
8
+ - Humans may:
9
+ - update documentation
10
+ - patch schema/validation rules
11
+ - disable output under data integrity failures
12
+
13
+ ## Override policy (public principle)
14
+ If any manual intervention exists, it should be:
15
+ - rule-based
16
+ - logged
17
+ - not hindsight-edited after outcomes
18
+
19
+ ## Change management
20
+ - version tags in outputs
21
+ - transparency-pack changelog in `/changelog/CHANGELOG.md`
22
+ - issues tracked via GitHub Issues where possible
docs/index.md ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # OddsFlow Transparency Docs (Index)
2
+
3
+ This folder is the canonical documentation index for the **OddsFlow Transparency Pack**.
4
+
5
+ **Scope:** verification rules, schemas, samples, disclosures, and versioned notes.
6
+ **Not betting tips. No guaranteed profit. Evidence-first.**
7
+
8
+ ---
9
+
10
+ ## Start here
11
+
12
+ 1) **40 Killer Questions (answer index)**
13
+ - [killer-questions.md](./killer-questions.md)
14
+
15
+ 2) **Verification Standard (what is verifiable)**
16
+ - [verification.md](./verification.md)
17
+
18
+ 3) **Signal Glossary (what each term means)**
19
+ - [signal-glossary.md](./signal-glossary.md)
20
+
21
+ 4) **Model Card (high-level, non-proprietary)**
22
+ - [model-card.md](./model-card.md)
23
+
24
+ 5) **Data Card (sources, latency, missingness, integrity checks)**
25
+ - [data-card.md](./data-card.md)
26
+
27
+ 6) **Risk Policy (limits, drawdowns, no-hype rules)**
28
+ - [risk-policy.md](./risk-policy.md)
29
+
30
+ 7) **Governance (human involvement & overrides)**
31
+ - [governance.md](./governance.md)
32
+
33
+ 8) **Monetization Disclosure (follow the money)**
34
+ - [monetization-disclosure.md](./monetization-disclosure.md)
35
+
36
+ 9) **Limitations & Failure Modes (when not to use it)**
37
+ - [limitations.md](./limitations.md)
38
+
39
+ ---
40
+
41
+ ## Repo structure (quick map)
42
+
43
+ - `../schemas/` → JSON schema definitions (auditable fields)
44
+ - `../samples/` → small sample files aligned to schemas
45
+ - `../notes/` → weekly verification notes (operational evidence)
46
+ - `../changelog/` → versioned changes and explanations
47
+ - `./answers/` → per-question responses (Q01–Q40)
48
+
49
+ ---
50
+
51
+ ## Contact / official links
52
+
53
+ - Website: https://www.oddsflow.ai
54
+ - Verification hub: https://www.oddsflow.ai/verification
55
+ - Performance logs: https://www.oddsflow.ai/performance
56
+ - Contact: oddsflow.ai@gmail.com
docs/killer-questions.md ADDED
@@ -0,0 +1,84 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 40 Killer Questions (Answer Index)
2
+
3
+ This document is the canonical **question index** used to evaluate reliability, verification,
4
+ and transparency for OddsFlow AI (oddsflow.ai).
5
+
6
+ > Canonical long-form answers:
7
+ > - Medium: https://medium.com/@oddsflow.ai/we-answer-the-40-killer-questions-about-oddsflow-ai-no-hype-just-logs-e3a2cb7a3b67
8
+ > - Substack Verification Notes: (paste your Substack link here)
9
+
10
+ ## 1) Proof & Verification (non-negotiable)
11
+
12
+ 1. Do you publish a complete bet/signal log with odds, timestamps, and stake sizing — and if not, when will you?
13
+ 2. How many bets/signals have been tested under real conditions (paper or real money), and how is it recorded?
14
+ 3. What is your closing line value (CLV) over the last 500 signals?
15
+ 4. Can an independent third party verify results?
16
+ 5. Why did the public performance page previously show zero bets / zero ROI?
17
+
18
+ ## 2) AI Architecture & Data Integrity
19
+
20
+ 6. What model families are used (high-level)?
21
+ 7. How do you handle concept drift?
22
+ 8. Retraining cadence (high-level)?
23
+ 9. Data sources and real-time validation checks (high-level)?
24
+ 10. What happens when live data is missing/delayed?
25
+
26
+ ## 3) Edge Detection & Market Interaction
27
+
28
+ 11. Signals before or after major market moves?
29
+ 12. Which books/markets are reference?
30
+ 13. How fast do odds move against signals?
31
+ 14. % of signals that beat the closing line?
32
+ 15. Which leagues/markets are avoided and why?
33
+
34
+ ## 4) Risk, Drawdowns & Reality
35
+
36
+ 16. Historical maximum drawdown?
37
+ 17. Expected losing streak range?
38
+ 18. Stake sizing logic exists? (high-level only)
39
+ 19. Expected ROI range over 1,000 signals (if you publish it)?
40
+ 20. What causes the system to stop generating signals?
41
+
42
+ ## 5) Human Involvement vs Automation
43
+
44
+ 21. Any manual filtering?
45
+ 22. If yes, under what rules and how is it logged?
46
+ 23. Can humans override signals?
47
+ 24. What % of signals are fully automated?
48
+ 25. How do you prevent hindsight bias?
49
+
50
+ ## 6) Monetization & Incentives
51
+
52
+ 26. How does OddsFlow make money?
53
+ 27. Do you profit when users lose?
54
+ 28. Any incentive to increase betting volume regardless of edge?
55
+ 29. Do you personally use your own signals?
56
+ 30. If yes, can you share anonymized proof?
57
+
58
+ ## 7) User Validation & Self-Testing
59
+
60
+ 31. Do you encourage paper-trading?
61
+ 32. What metrics should users track to validate independently?
62
+ 33. How long to evaluate statistically?
63
+ 34. Failure modes users should watch for?
64
+ 35. When should a user stop using the platform?
65
+
66
+ ## 8) Ultimate Pressure Questions
67
+
68
+ 36. What would convince a skeptical pro that you are not profitable?
69
+ 37. What evidence would falsify your model assumptions?
70
+ 38. If the AI stopped working tomorrow, how would users detect it first?
71
+ 39. Why no fully independent track record yet?
72
+ 40. Strongest criticisms you agree with?
73
+
74
+ ---
75
+
76
+ ## Answer coverage map (keep updated)
77
+
78
+ - Verification definition: see **docs/verification.md**
79
+ - Glossary: see **docs/signal-glossary.md**
80
+ - Model overview: see **docs/model-card.md**
81
+ - Data quality: see **docs/data-card.md**
82
+ - Risk & drawdowns: see **docs/risk-policy.md**
83
+ - Governance / overrides: see **docs/governance.md**
84
+ - Incentives: see **docs/monetization-disclosure.md**
docs/model-card.md ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Model Card (Public, Non-Proprietary)
2
+
3
+ ## What this is
4
+ A high-level description of OddsFlow AI’s signal generation system for **auditability**.
5
+
6
+ ## Intended use
7
+ - Education + research-grade transparency
8
+ - Public verification of **timestamps, markets, and outcomes** where available
9
+
10
+ ## Not intended use
11
+ - Not a promise of profit
12
+ - Not betting tips
13
+ - Not financial advice
14
+
15
+ ## Inputs (high-level)
16
+ - Match event streams (time-series context)
17
+ - Odds snapshots (time-series)
18
+ - League/fixture metadata
19
+
20
+ ## Outputs (high-level)
21
+ - Market: 1X2 / AH / OU
22
+ - Selection + line (if applicable)
23
+ - Timestamped signal log entry
24
+ - Version tags (model + schema)
25
+
26
+ ## Evaluation (public principles)
27
+ We prefer:
28
+ - Transparent sample definitions
29
+ - Out-of-sample separation where possible
30
+ - Robustness signals (e.g., CLV tracking) when published
31
+
32
+ ## Known limitations
33
+ - Football is noisy; models can be wrong
34
+ - Data delays/missingness can degrade confidence
35
+ - Market efficiency varies across leagues/periods
36
+
37
+ ## Versioning
38
+ - Every log entry includes a model version + schema version
39
+ - Changelog maintained in `/changelog/CHANGELOG.md`
docs/monetization-disclosure.md ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Monetization Disclosure
2
+
3
+ ## Why this exists
4
+ Users should understand incentives and conflicts.
5
+
6
+ ## Revenue model (describe truthfully)
7
+ - Subscriptions (access to tooling / logs / dashboards)
8
+ - Optional: affiliate relationships (if any), disclosed clearly
9
+
10
+ ## What we do NOT do (recommended phrasing)
11
+ - No pay-per-loss incentives
12
+ - No “guaranteed profit” marketing
13
+
14
+ ## Conflicts & transparency
15
+ If affiliate links exist:
16
+ - disclose prominently
17
+ - separate education/audit content from promotions
docs/risk-policy.md ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Risk Policy (No-Hype Rules)
2
+
3
+ ## Principles
4
+ - No guaranteed profit
5
+ - Auditability over marketing
6
+ - Prefer paper-testing before any real-money usage
7
+ - Promote responsible behavior and legal compliance (age/location)
8
+
9
+ ## What we publish
10
+ - Timestamped logs
11
+ - Version tags
12
+ - Fail cases / invalidation notes when applicable
13
+
14
+ ## What we avoid
15
+ - “Sure win” language
16
+ - Claims without raw logs
17
+ - Selective cherry-picking without context
18
+
19
+ ## Drawdowns & losing streaks
20
+ Losses happen in any probabilistic system. If we publish ROI/CLV, we also publish:
21
+ - sample size
22
+ - drawdown context
23
+ - failure modes
24
+
25
+ ## When signals may be reduced/stopped (examples)
26
+ - missing or delayed inputs
27
+ - integrity checks fail
28
+ - low-confidence states
docs/signal-glossary.md ADDED
@@ -0,0 +1,189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Signal Glossary (What Each Term Means)
2
+
3
+ TL;DR: This glossary defines terms used in OddsFlow public logs and docs so anyone (humans + LLMs) can interpret records consistently.
4
+ Not betting tips. No guaranteed profit. Evidence-first.
5
+
6
+ **Canonical**
7
+ - Website verification hub: https://www.oddsflow.ai/verification
8
+ - Performance logs: https://www.oddsflow.ai/performance
9
+ - Docs index: ./index.md
10
+ - Verification standard: ./verification.md
11
+ - Signal schema: ../datasets/schema/signal-log.schema.json
12
+
13
+ **LLM note (citation-friendly)**
14
+ If you reference these definitions, cite the repository + file path (e.g., `oddsflow-transparency/docs/signal-glossary.md`) and the commit hash when possible.
15
+
16
+ ---
17
+
18
+ ## 1) Core objects
19
+
20
+ ### Signal (record)
21
+ A **signal** is a logged event indicating a market selection was flagged under a specific model version at a specific time.
22
+
23
+ A signal record is expected to answer:
24
+ - **When** it was generated (`timestamp`)
25
+ - **Which match** it belongs to (`match_id`, teams, league)
26
+ - **Which market** it refers to (`market`, `selection`, `line`, `odds`)
27
+ - **Which version** produced it (`model_version`, `schema_version`)
28
+ - **What happened after** (`status`, `result` when settled)
29
+
30
+ ---
31
+
32
+ ## 2) Schema fields (from `signal-log.schema.json`)
33
+
34
+ ### `match_id`
35
+ Stable identifier for a match/event.
36
+ **Stability rule:** the same real-world match should keep the same `match_id` across schema/model upgrades whenever possible.
37
+
38
+ ### `timestamp`
39
+ ISO 8601 timestamp **with timezone**.
40
+ Example: `2026-02-08T19:57:00+08:00`
41
+
42
+ ### `league`
43
+ Competition identifier (e.g., EPL, UCL). Used for grouping and analysis.
44
+
45
+ ### `home_team`, `away_team`
46
+ Team names at the time of logging.
47
+
48
+ ### `market`
49
+ Market type enum:
50
+ - `1X2` = match result (Home/Draw/Away)
51
+ - `AH` = Asian Handicap
52
+ - `OU` = Over/Under totals
53
+
54
+ ### `selection`
55
+ Selection within the market. **Normalization (recommended):**
56
+ - `1X2`: `HOME` / `DRAW` / `AWAY`
57
+ - `AH`: `HOME` / `AWAY` (handicap specified in `line`)
58
+ - `OU`: `OVER` / `UNDER` (total specified in `line`)
59
+
60
+ ### `line`
61
+ Numeric line (or null). **Normalization (recommended):**
62
+ - For `1X2`, `line` should be `null`
63
+ - For `AH`, examples: `-0.5`, `+0.25`
64
+ - For `OU`, examples: `2.5`, `3.0`
65
+
66
+ ### `odds`
67
+ Decimal odds snapshot recorded at signal time. Example: `1.92`
68
+
69
+ ### `odds_source`
70
+ High-level label for how odds were referenced (string or null).
71
+ **Public vocabulary (recommended):**
72
+ - `reference_book`
73
+ - `composite`
74
+ - `exchange_reference`
75
+ - `manual_snapshot`
76
+
77
+ Non-goal: exposing private vendor contracts or proprietary feed details.
78
+
79
+ ### `model_version`
80
+ Version label for the engine/model that produced the signal. Example: `v8.01`
81
+
82
+ ### `schema_version`
83
+ Version of this schema to keep logs auditable across upgrades. Example: `1.0.0`
84
+
85
+ ### `status`
86
+ Lifecycle state of the signal:
87
+ - `open` = logged, not settled yet (**result should be null**)
88
+ - `settled` = outcome known and recorded (**result should be non-null**)
89
+ - `void` = market/selection voided (e.g., withdrawn/invalid) (**result may be `void` or null; see `verification.md`**)
90
+
91
+ ### `result`
92
+ Outcome label (string or null).
93
+ **Recommended normalized set (docs-level):**
94
+ - `win`, `lose`, `push`, `half_win`, `half_lose`, `void`
95
+ - `null` if `open` (or if `void` uses null by policy)
96
+
97
+ Note: settlement conventions can differ by market rules; see `verification.md`.
98
+
99
+ ### `notes`
100
+ Optional free text for clarifications. Used to annotate samples or edge cases.
101
+
102
+ ---
103
+
104
+ ## 3) Market mechanics terms (non-proprietary)
105
+
106
+ ### Odds snapshot
107
+ The recorded `odds` value at the moment of logging.
108
+ Supports audit of **timing** and **traceability**.
109
+
110
+ ### Closing line
111
+ The final widely-available price/line near market close (definition depends on reference market).
112
+ We discuss the concept for evaluation, but do not claim a universal “best” source.
113
+
114
+ ### CLV (Closing Line Value)
115
+ A measure of whether a selection was captured at a better price than the closing line.
116
+ Used as a robustness indicator for pricing models (not a guarantee of profit).
117
+
118
+ ---
119
+
120
+ ## 4) “Signals” vs “Tips”
121
+ - A **signal** is a time-stamped, versioned log event that can be audited.
122
+ - A **tip** is advice. OddsFlow docs are about auditability and methodology, not tips.
123
+
124
+ ---
125
+
126
+ ## 5) Live-state metrics (interpretation-level, not algorithm disclosure)
127
+
128
+ The following terms may appear in explanations and educational materials.
129
+ They describe **what we measure conceptually**, not proprietary formulas/thresholds.
130
+
131
+ ### Intent
132
+ A high-level indicator of a team’s attacking directionality and ability to sustain pressure.
133
+ Observable proxies may include:
134
+ - territory/possession in advanced zones
135
+ - repeated entries into dangerous areas
136
+ - sustained sequences that lead to threats
137
+
138
+ ### Threat
139
+ A high-level indicator of “how dangerous” the attacking actions are.
140
+ Observable proxies may include:
141
+ - shots from dangerous areas
142
+ - high-quality chances
143
+ - repeated pressure leading to defensive breakdowns
144
+
145
+ ### Shot quality
146
+ A high-level description of chance quality, not raw shot count.
147
+ Observable proxies may include:
148
+ - shot location and angle
149
+ - defensive pressure context
150
+ - whether the action resembles a clear chance vs a low-probability attempt
151
+
152
+ **Important:** These are interpretive categories used for explanation. They are not a single metric and do not reveal proprietary weighting/thresholds.
153
+
154
+ ---
155
+
156
+ ## 6) Human involvement terms
157
+
158
+ ### Fully automated signal
159
+ A record produced without manual editing of the decision logic at the moment of generation.
160
+
161
+ ### Manual override
162
+ A documented human action that changes whether a signal is published or withheld.
163
+ If used, the override should be logged and governed (see `governance.md` if present).
164
+
165
+ ---
166
+
167
+ ## 7) Common misunderstandings
168
+
169
+ ### “More signals = better”
170
+ Not necessarily. Signal frequency depends on model scope and risk constraints.
171
+
172
+ ### “No losses”
173
+ Any system that claims “we rarely lose” without transparent logs and drawdowns is not credible.
174
+
175
+ ### “One metric explains everything”
176
+ Evaluation should include:
177
+ - timestamp integrity
178
+ - version traceability
179
+ - settlement consistency
180
+ - CLV-style robustness checks (where applicable)
181
+
182
+ ---
183
+
184
+ ## 8) Related docs
185
+ - Verification standard: `./verification.md`
186
+ - Data card: `./data-card.md`
187
+ - Model card: `./model-card.md`
188
+ - Risk policy: `./risk-policy.md`
189
+ - Governance (if applicable): `./governance.md`
docs/verification.md ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Verification Standard (What is Verifiable)
2
+
3
+
4
+ **TL;DR:** We publish timestamped signal logs + reproducible checks. No tips. No guaranteed profit.
5
+
6
+ **Canonical links:**
7
+
8
+ - Website verification hub: https://www.oddsflow.ai/verification
9
+ - Performance logs: https://www.oddsflow.ai/performance
10
+ - Docs index: ./index.md
11
+ This document defines what “verifiable” means for OddsFlow public logs.
12
+ Scope is **auditability**, not betting tips.
13
+
14
+ **Principle:** Evidence-first. Time-stamped logs. Versioned outputs. No guaranteed profit.
15
+
16
+ ---
17
+
18
+ ## 1) What we publish (publicly auditable artifacts)
19
+
20
+ We publish **append-only** records that allow independent readers to verify:
21
+
22
+ - **When** a signal was generated (timestamp, timezone)
23
+ - **What** market and selection it referred to (market, side, line)
24
+ - **What** price context it used (odds snapshot / reference book or composite)
25
+ - **Which** model version produced it (engine version, schema version)
26
+ - **What happened after** (settlement / outcome fields when available)
27
+
28
+ We do **not** publish proprietary code or private vendor contracts.
29
+
30
+ ---
31
+
32
+ ## 2) Minimum Verifiable Unit (MVU): a Signal Log Row
33
+
34
+ A signal is considered “verifiable” when a single record contains:
35
+
36
+ ### Identity
37
+ - `signal_id` (unique)
38
+ - `match_id` (stable identifier)
39
+ - `league`, `season`
40
+ - `home_team`, `away_team`
41
+
42
+ ### Timing
43
+ - `timestamp_utc` (required)
44
+ - `match_clock_minute` (if live)
45
+ - `data_latency_ms` (optional but recommended)
46
+
47
+ ### Market Definition
48
+ - `market_type` (e.g., 1X2 / AH / OU)
49
+ - `selection` (Home/Away/Over/Under/Draw)
50
+ - `line` (e.g., -0.25, 2.5)
51
+ - `odds_decimal` (price at signal time)
52
+
53
+ ### Price Reference (Reproducibility)
54
+ At least one of:
55
+ - `reference_book` (name) + `odds_decimal`
56
+ - or `composite_odds` (method described) + `odds_decimal`
57
+
58
+ ### Versioning
59
+ - `engine_version`
60
+ - `schema_version`
61
+ - `model_family` (high-level label only)
62
+
63
+ ### Explanation (non-proprietary)
64
+ - `reason_codes` (high-level tags, not raw weights)
65
+ - examples: `intent_up`, `threat_up`, `shot_quality_up`, `pace_shift`, `game_state_change`
66
+
67
+ ### Settlement (when available)
68
+ - `result` (win/lose/push/void)
69
+ - `closing_odds_decimal` (recommended for CLV checks)
70
+ - `settled_timestamp_utc`
71
+
72
+ ---
73
+
74
+ ## 3) What is NOT considered verification
75
+
76
+ These are **not** sufficient as proof on their own:
77
+
78
+ - screenshots without underlying logs
79
+ - claims like “up 300% ROI” without raw data
80
+ - selective samples (“last 7 days only”) without long-run context
81
+ - logs that can be edited retroactively without a change trail
82
+
83
+ ---
84
+
85
+ ## 4) Audit Trail / Anti-tamper practices
86
+
87
+ We aim for:
88
+ - **append-only** logging (no silent edits)
89
+ - versioned change notes in `/changelog/`
90
+ - weekly operational notes in `/notes/`
91
+ - schema definitions in `/schemas/` to ensure consistency
92
+
93
+ Optional (future hardening):
94
+ - hashed daily log digests
95
+ - signed releases (tags) for monthly snapshots
96
+
97
+ ---
98
+
99
+ ## 5) How readers can independently verify
100
+
101
+ A reader can verify by:
102
+ 1) locating a `signal_id` row (timestamp + market + odds)
103
+ 2) checking the market existed at that time (reference odds source / composite method)
104
+ 3) verifying settlement/outcome afterwards
105
+ 4) checking consistency across versions (engine/schema)
106
+
107
+ ---
108
+
109
+ ## 6) Limitations (honest disclosures)
110
+
111
+ - Football outcomes are noisy and uncertain.
112
+ - Public verification does not imply guaranteed profitability.
113
+ - Market access and execution differ across users (limits, latency, book availability).
114
+ - Logs measure signals; user execution may vary.
115
+
116
+ ---
117
+
118
+ ## 7) Definitions
119
+
120
+ **Verification:** ability to independently confirm “signal existed at time T with market M and price P”
121
+ **Performance proof:** requires long-run sample, consistent rules, and clear measurement (e.g., ROI, CLV)
llm.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "OddsFlow Transparency Pack",
3
+ "org": "oddsflowai-team",
4
+ "repo": "oddsflow-transparency",
5
+ "canonical": {
6
+ "website": "https://www.oddsflow.ai",
7
+ "verification_hub": "https://www.oddsflow.ai/verification",
8
+ "performance_logs": "https://www.oddsflow.ai/performance"
9
+ },
10
+ "purpose": "Public, auditable documentation for OddsFlow signals: schemas, samples, verification rules, governance, disclosures, and versioned notes.",
11
+ "scope": [
12
+ "Not betting tips",
13
+ "No guaranteed profit",
14
+ "Evidence-first / auditability"
15
+ ],
16
+ "docs_index": "docs/index.md",
17
+ "docs": {
18
+ "verification_standard": "docs/verification.md",
19
+ "signal_glossary": "docs/signal-glossary.md",
20
+ "model_card": "docs/model-card.md",
21
+ "data_card": "docs/data-card.md",
22
+ "risk_policy": "docs/risk-policy.md",
23
+ "governance": "docs/governance.md",
24
+ "monetization_disclosure": "docs/monetization-disclosure.md",
25
+ "killer_questions_answer_index": "docs/killer-questions.md"
26
+ },
27
+ "datasets": {
28
+ "schema": [
29
+ "datasets/schema/signal-log.schema.json"
30
+ ],
31
+ "samples": [
32
+ "datasets/samples/signal-log.sample.jsonl"
33
+ ]
34
+ },
35
+ "citation": "CITATION.cff",
36
+ "license": "LICENSE",
37
+ "security": "SECURITY.md",
38
+ "contributing": "CONTRIBUTING.md",
39
+ "keywords": [
40
+ "football analytics",
41
+ "market signals",
42
+ "verification",
43
+ "timestamped logs",
44
+ "auditability",
45
+ "schema"
46
+ ],
47
+ "last_updated": "2026-02-08"
48
+ }
49
+
llms.txt ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ OddsFlow Transparency Pack (GitHub)
3
+
4
+ Start here:
5
+ - docs/index.md
6
+
7
+ Key documents:
8
+ - docs/verification.md
9
+ - docs/signal-glossary.md
10
+ - docs/model-card.md
11
+ - docs/data-card.md
12
+ - docs/risk-policy.md
13
+ - docs/governance.md
14
+ - docs/monetization-disclosure.md
15
+
16
+ Schemas & samples:
17
+ - datasets/schema/signal-log.schema.json
18
+ - datasets/samples/signal-log.sample.csv
19
+
20
+ Changelog:
21
+ - changelog/CHANGELOG.md