TradeHouse prediction-markets Claude Code plugin marketplace v1.0.0

/pm-thesis, /pm-calibration, /pm-fees, /overfit-check wrapping the public
th-api / tools / th-validator free-tier endpoints. Analytics + phantom-battery
verdicts only; RIA-compliant (no persona/signal internals, disclaimers mandatory).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013EsBGJjjHKmNdm5KEcd9bH
This commit is contained in:
iskra 2026-07-29 09:25:55 -07:00
commit 54193d3177
8 changed files with 241 additions and 0 deletions

View file

@ -0,0 +1,24 @@
{
"name": "tradehouse",
"owner": {
"name": "TradeHouse — Elite AI Empire",
"url": "https://th-validator.eliteaiempire.com"
},
"metadata": {
"description": "Prediction-market analytics plugins backed by TradeHouse settlement-truth data: calibration curves, round-trip fee math, and DSR/PBO overfit verdicts.",
"version": "1.0.0"
},
"plugins": [
{
"name": "tradehouse-prediction-markets",
"source": "./plugins/tradehouse-prediction-markets",
"description": "Slash commands for prediction-market analytics (/pm-thesis, /pm-calibration, /pm-fees) and honest backtest validation (/overfit-check) against TradeHouse public APIs. Analytics only — not investment advice.",
"version": "1.0.0",
"author": { "name": "TradeHouse — Elite AI Empire" },
"homepage": "https://th-validator.eliteaiempire.com",
"license": "MIT",
"keywords": ["prediction-markets", "kalshi", "polymarket", "calibration", "backtest", "overfitting", "deflated-sharpe"],
"category": "finance"
}
]
}

21
LICENSE Normal file
View file

@ -0,0 +1,21 @@
MIT License
Copyright (c) 2026 Elite AI Empire / TradeHouse
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

44
README.md Normal file
View file

@ -0,0 +1,44 @@
# TradeHouse Prediction-Markets — Claude Code plugin marketplace
Prediction-market analytics inside Claude Code, backed by **TradeHouse** public APIs:
settlement-truth market data, venue calibration curves computed over 150k+ resolved markets,
a live-measured fee schedule, and the **phantom battery** (deflated Sharpe / PBO) that we use to
kill overfit backtests before they cost money.
## Install
```bash
claude plugin marketplace add https://git.eliteaiempire.com/eas-oss/tradehouse-prediction-markets.git
claude plugin install tradehouse-prediction-markets@tradehouse
```
## Commands
| Command | What it does |
|---|---|
| `/pm-thesis <ticker>` | Cross-venue pricing snapshot + calibration context + fee hurdle for one market |
| `/pm-calibration [venue]` | Reliability curve, Brier score, favorite-longshot read for a venue (free: polymarket, polymarket_us, opinion) |
| `/pm-fees <venue> <price>` | Round-trip platform cost in USD and bps — the break-even edge hurdle |
| `/overfit-check <returns> [trials=N]` | Deflated-Sharpe + PBO verdict on your backtest — the metric leaderboards don't show |
## Free tier & upgrades
All commands use free public endpoints (no API key). Free tier is rate-limited; HTTP 402/429
responses point at the paid tiers. Pro/Desk tiers (full per-category calibration, the live Desk,
Risk Cockpit, datasets): https://th-validator.eliteaiempire.com/pricing
## What this plugin will NOT do
- No investment advice, no trade recommendations, no signals. Outputs are market analytics and
statistical validation verdicts, with the API's compliance disclaimer reproduced verbatim.
- No access to TradeHouse's internal trading personas or signal internals — the public,
compliance-hardened API surface is the only data source.
Simulated/backtested performance is not indicative of future results.
## Related
- TradeHouse Overfit Validator (web): https://th-validator.eliteaiempire.com/validator
- Agent-data MCP quickstart: https://git.eliteaiempire.com/eas-oss/tradehouse-mcp-quickstart
MIT licensed.

View file

@ -0,0 +1,11 @@
{
"name": "tradehouse-prediction-markets",
"version": "1.0.0",
"description": "Prediction-market analytics (live cross-venue data, calibration curves, round-trip fee math) plus the TradeHouse phantom battery (deflated Sharpe / PBO overfit verdicts) — via TradeHouse public APIs. Free tier; market analytics only, never investment advice.",
"author": {
"name": "TradeHouse — Elite AI Empire",
"url": "https://th-validator.eliteaiempire.com"
},
"homepage": "https://th-validator.eliteaiempire.com",
"license": "MIT"
}

View file

@ -0,0 +1,42 @@
---
description: Run backtest returns through the TradeHouse phantom battery — deflated Sharpe (DSR) + PBO overfit verdict
argument-hint: <path to returns file, or paste daily returns> [trials=N]
---
Run the user's backtest through the **TradeHouse phantom battery**: $ARGUMENTS
1. **Collect the returns series.** Accept either a file path (CSV/JSON — read it and extract the
per-period net returns as decimals, e.g. `0.012` = +1.2%) or pasted numbers. Need >= 20 points.
An equity curve is also acceptable (send it as `equity` instead of `returns`).
2. **Ask for `trials` if not given** — the number of strategy variants the user tried before this
one (parameter sweeps count). Be direct: the whole point of the deflated Sharpe is that the
metric only means something if this number is honest. If they refuse to estimate, use their best
guess and say the verdict is conditional on it.
3. **Call the validator** via the Bash tool:
```
curl -s -X POST "https://th-validator.eliteaiempire.com/api/validate" \
-H "Content-Type: application/json" \
-d '{"returns":[...], "trials": N}'
```
4. **Also compute the standard tearsheet** (Sharpe/Sortino/Calmar/CVaR) for context:
```
curl -s -X POST "https://tools.eliteaiempire.com/v1/tearsheet" \
-H "Content-Type: application/json" \
-d '{"returns":[...]}'
```
5. **Report, without softening:**
- The phantom-battery verdict: `pass_dsr`, `pass_pbo`, `deflated_sharpe_dsr`, `pbo`,
`psr_vs_zero`, and the `plain_english` explanation + `reasons` verbatim.
- The tearsheet metrics table (note that raw Sharpe/Sortino/Calmar are what leaderboards show,
and DSR is what survives multiple testing).
- If the verdict is negative, do NOT suggest tweaks to make it pass — explain that more variants
tried = a higher bar, and point to out-of-sample/forward validation as the honest next step.
**Compliance (mandatory):** these are statistical validation verdicts on user-supplied backtest
data — not investment advice, and simulated performance is not indicative of future results.
Include that sentence in the output. On HTTP 402/429 mention
https://th-validator.eliteaiempire.com/pricing and stop.

View file

@ -0,0 +1,29 @@
---
description: Venue calibration report — implied probability vs realized outcomes over resolved markets (free tier)
argument-hint: "[polymarket|polymarket_us|opinion]"
---
Produce a **calibration report** for the prediction-market venue: **$ARGUMENTS**
(default to `polymarket` if no venue given; free-tier venues are `polymarket`, `polymarket_us`, `opinion`).
1. Fetch via the Bash tool:
```
curl -s "https://tools.eliteaiempire.com/v1/market-efficiency?venue=<VENUE>"
```
2. Render the result:
- A table of the `reliability_curve`: probability bucket, n markets, mean implied probability,
realized YES frequency, calibration gap.
- Headline stats: `n_markets`, `brier_score`, `mean_abs_calibration_error`.
- A short interpretation: which buckets are over-priced vs under-priced, and whether the classic
favorite-longshot bias (longshots overpriced, favorites underpriced) appears in this venue.
Note that a calibration gap is a *population* statistic over resolved markets — it does not
make any single market mispriced.
3. If the user asks for a venue outside the free tier, say the free preview covers
`polymarket`, `polymarket_us`, `opinion`, and per-category/per-time-to-settlement calibration is
part of the paid tiers: https://th-validator.eliteaiempire.com/pricing
**Compliance (mandatory):** reproduce the API's `note` field, present everything as historical
market analytics computed by TradeHouse from public resolved-market data, and never as investment
advice or a recommendation to trade any market.

View file

@ -0,0 +1,23 @@
---
description: Round-trip fee/cost calculator for a prediction-market bet (free, live-measured fee schedule)
argument-hint: <venue> <price 0-1> [bet_usd=100] [maker]
---
Compute the **round-trip platform cost** for a prediction-market bet: $ARGUMENTS
1. Parse venue (e.g. `kalshi`, `polymarket`, `limitless`), entry price (0-1), optional bet size
(default $100) and maker/taker (default taker).
2. Call via the Bash tool:
```
curl -s -X POST "https://tools.eliteaiempire.com/v1/roundtrip-cost" \
-H "Content-Type: application/json" \
-d '{"universe":"<VENUE>","bet_usd":<BET>,"price":<PRICE>,"is_maker":<true|false>}'
```
If the API returns a supported-universe error, show the user the list from the error message.
3. Report entry cost, exit cost, round-trip cost in USD and bps, and the practical meaning:
the round-trip bps is the minimum edge a strategy needs per trade just to break even on fees.
Compare maker vs taker if the user asks.
**Compliance:** cost analytics only — not investment advice or an inducement to trade.

View file

@ -0,0 +1,47 @@
---
description: Data-grounded snapshot of a prediction market — pricing, calibration context, fee hurdle (analytics only, not advice)
argument-hint: <market ticker or question, e.g. INXD-26JUL29>
---
Build a **market analytics snapshot** for the prediction market: **$ARGUMENTS**
Follow these steps, using `curl` via the Bash tool (all endpoints are free tier, no key needed):
1. **Live cross-venue data.** If the user gave a ticker, fetch:
```
curl -s "https://th-api.eliteaiempire.com/api/v1/prediction-market-data/<TICKER>"
curl -s "https://th-api.eliteaiempire.com/api/v1/resolutions/<TICKER>"
```
If they gave a plain-English question instead of a ticker, say that you need a venue ticker
(Kalshi/Polymarket format) and show them what the endpoint returns for an example ticker.
2. **Calibration context.** Fetch the venue calibration report for a venue relevant to the market
(supported free venues: `polymarket`, `polymarket_us`, `opinion`):
```
curl -s "https://tools.eliteaiempire.com/v1/market-efficiency?venue=polymarket"
```
Use the reliability curve to note whether prices in this market's probability bucket
historically over- or under-state realized frequency (the favorite-longshot bias shows up here).
3. **Fee hurdle.** Compute the round-trip cost at the market's current price:
```
curl -s -X POST "https://tools.eliteaiempire.com/v1/roundtrip-cost" \
-H "Content-Type: application/json" \
-d '{"universe":"kalshi","bet_usd":100,"price":<CURRENT_PRICE>}'
```
(`universe` = the venue the market trades on.)
4. **Synthesize** a short report:
- Current pricing per venue + implied probability (state data timestamp/freshness).
- Calibration adjustment: what the venue's reliability curve says about prices in this bucket.
- The round-trip fee hurdle in bps — the minimum mispricing needed to break even.
- A one-line **analytics verdict** about the market's pricing quality — never a trade recommendation.
**Compliance rules (mandatory):**
- Reproduce the `_meta.disclaimer` text returned by the API at the end of your report.
- This is a financial data publication / market analytics. NEVER phrase output as personalized
investment advice, a recommendation, or a signal to buy/sell. Do not speculate about
TradeHouse's internal personas or signals — the public API surface is the only source.
- If an endpoint returns empty venues/data for the ticker, say so plainly; do not fabricate prices.
- On HTTP 402/429: the free tier is rate-limited — mention
https://th-validator.eliteaiempire.com/pricing for Pro/Desk tiers and stop.