ml4t-point-in-time skillA
ml4t-point-in-time is agent-read markdown (skill) from ml4t/skills: Ensure data reflects what was known at each decision point, not revised or restated values. Use when joining fundamental, macro, or alternative data to price series..
Indexed from public GitHub and served as immutable, content-addressed versions. Install it pinned to an exact SHA-256 with the mdr CLI, and every file is verified against the hash recorded here before it reaches your agent. The deterministic audit below grades the latest version, and the same file always earns the same grade.
What the file says
# Point-in-Time Correctness
Joining data by event date instead of availability date uses information that did not exist yet, inflating backtest returns by 1-3% annually on fundamental strategies.
## The Problem
Fundamental data has two timestamps: when the economic event occurred (quarter-end) and when the data became publicly available (SEC filing date, press release, data vendor publication). Using the event date treats revised, delayed, or embargoed data as if it were available in real time.
Example: Apple's Q4 2024 revenue (period ending Dec 31) is filed with the SEC on Jan 31, 2025. A model that joins revenue on Dec 31 uses data 31 days before it existed.
Macro data has the same problem: GDP is released ~30 days after quarter-end and revised two to three times over the following months.
## The Pattern
### WRONG
```python
import polars as pl
# Join fundamentals on the quarter they describe
prices = pl.read_parquet("prices.parquet")
fundamentals = pl.read_parquet("fundamentals.parquet")
df = prices.join(
fundamentals,
left_on=["symbol", "timestamp"],
right_on=["symbol", "quarter_end"], # uses event date
how="left",
)
```
### CORRECT
```python
…Read the whole file at its exact version.
How to install
mdr add ml4t/skills/ml4t-point-in-time@git:20260901.c415df0mdr add ml4t/skills/ml4t-point-in-time@sha256:cff8e72b709d54bePin to a label to follow the author's releases, or to a sha256 to freeze the exact bytes forever. Either way the resolved hash is written to mdr.lock, and mdr install reproduces it on any machine.
[](https://markdownregistry.com/a/art_uvuoqpf3besgpi7w)
1 badge views in 30 days
Versions
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20260901.c415df0 latest | 2026-09-01 | c415df0 | 4,174 B | A | view · diff |
| git:20260528.303089e | 2026-05-28 | 303089e | 4,175 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (4174 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
ml4t/skills · 19 stars · license Apache-2.0 · pushed 2026-09-24 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_uvuoqpf3besgpi7w GET https://markdownregistry.com/api/v1/resolve?ref=ml4t/skills/ml4t-point-in-time GET https://markdownregistry.com/api/v1/blob/cff8e72b709d54bed7bbc1edc3387fafa6a8a4191326cd1844ff6dd65c9be697
Your agent does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.