Documentation.

The OHLC Table ships with a complete Context Pack — a set of focused reference files that define the indicator, its settings, its data, and the two code systems used to customize it.

What the documentation is for

The Table turns named trading windows into structured data: exact highs and lows, timestamps, ranges, direction, raids, averages, volume, and range anatomy. The documentation gives those values a stable definition and shows you how to work with them without relying on guesswork.

It does three jobs:

• Learn — what every row, timestamp, marker, and state means, with arithmetic you can check for yourself.
• Configure — sessions, metrics, display, heat, and live states, set correctly for your routine.
• Build — custom pcr-1 rosters, pcp-1 palettes, alert workflows, and an AI assistant whose reading of the Table stays grounded in what is actually printed.

How to use the pack

The focused way

Start a conversation with llm.md, then attach the one reference file that matches the task. This is the most reliable approach for reading data, explaining settings, or building a code.

The quick way

Open prompts.md, copy the prompt that matches the situation, and attach a clear screenshot of the Table. Prompts that ask for a roster or palette also require the matching grammar file.

The one-file way

Attach CONTEXT-PACK.md when you want every definition and grammar available in one place. Send either the combined pack or the individual files — not both.

The installed way

Every major assistant now has a persistent workspace — ChatGPT Projects, Claude Projects, Gemini Gems. Put PINNED.md in the workspace's instructions field and add the reference files as knowledge; the pack's Start-here file walks through it. One setup, and the rules ride every conversation.

Using it alongside the indicator

1. Configure the Table for the sessions and metrics relevant to your routine.
2. Show the hours beside the session names when a screenshot will be read later. The names are yours, so printing each window's clock keeps a reader — human or model — from guessing hours from a name.
3. Capture the Table clearly at a size where every value and timestamp is legible.
4. Attach the matching documentation and ask the question from prompts.md, or write your own.
5. Save useful configurations as TradingView indicator templates so a full setup can be restored without rebuilding it.

For a past session, use TradingView's Bar Replay. The Table always describes the most recent bar it can see; scrolling backward by itself does not move the Table to an earlier day.

Questions a record can answer

A question is only worth logging if a column can answer it. These can — each one is a filter and a count, one spreadsheet formula away:

• Which window prints the day’s extreme most often — and does it hold into the close? Needs the high and low timestamps, plus hold flags.
• After an overnight level is raided before the open, what does the cash session usually do? Needs raid times and direction.
• How wide is a normal London for your market — and how often does 150% of normal actually print? Needs range and Avg R Consumed.
• Does the opening range set the morning’s scale — or is that just a story? Needs range across two windows, plus Relative Volume.
• How long does the Manipulation leg usually run before the real move? Needs the anatomy leg durations.
• Which sessions are you actually good in? Needs the window column and your own outcome tags.

None of these predict anything — they just tell you what your market tends to do. recipes.md turns a question into a setup; prompts.md asks it the same way every day.

Custom roster and palette codes

The indicator accepts two portable configuration formats. A pcr-1 code carries session names and windows. A pcp-1 code carries the Table's colour system.

pcr-1 · the roster code

A roster carries up to six named windows. Times are a bare four-digit, 24-hour clock in New York time; overnight wrapping is native.

pcr-1-London~0200~0500|New York~0930~1615

The code fills slots in the order written. The enable switches stay yours — a code never turns a slot on or off — and the day-scale entries such as Daily and Previous Day are separate from the six roster slots.

pcp-1 · the palette code

A palette code carries twenty ordered six-character hex colours, plus an optional heat token. The order is positional and the positions carry no labels, so build or check a code with pcp.md open.

pcp-1-<colour-01>-<colour-02>-…-<colour-20>[-<heat>]

A valid code carries the base Table colours and its direction, forming, raid, and flash states. Layout, enabled rows, sessions, and dimensions remain separate settings.

Ready-made codes live on this site — palettes and rosters, each with a copy button, plus the Composer for writing a roster from six rows.

What is included

For reference — each file has one job. Attach the focused file that matches the question, or the complete Context Pack when you want the whole system at once.

What the files do not provide

• They do not make the Table predictive or turn a measurement into a trading instruction.
• They do not contain the indicator's engine or source code.
• They do not recover the path price travelled inside a window; that remains visible on the chart.
• They do not turn one session into a tested tendency. A screenshot is one observation.

The purpose of the pack is narrower and more useful: define what the Table prints, show how to configure it, and make the same information portable between the chart, your own process, and another reader.