The three components a system cannot run without
Strip away the indicators and the dashboards and a working trading system reduces to three parts. Leave any one out and the machine does not run — you have a habit wearing a uniform.
Why exactly three, and why these
A search for “what a trading system needs” returns long checklists — indicators, timeframes, market filters, journals, psychology. Most of those are refinements. The irreducible core is smaller, and it is structural: a testable edge gives the system a reason to act, a sizing rule keeps it alive long enough for that reason to pay, and a coded exit closes each trade on the plan instead of on nerve. Remove any one and the other two stop working. An edge you cannot size is a gamble; a sizing rule with no specified exit is sizing against an unknown loss; a coded exit on an unspecified edge protects nothing, because its levels are arbitrary. They are not a checklist of independent boxes — they are one structure that holds itself up.
The same three components, the dishonest way and the honest way
Each component has a counterfeit that looks like the real thing until you ask one question. The honest test for all three is the same: can a stranger re-check the claim after the outcome is known? The table sorts the common ways a record is presented by exactly that.
| Where the claims live | Locked before outcome? | Full count shown? | Why it sits there |
|---|---|---|---|
| A scrolling chat (Telegram, Discord) | No | Rarely | The operator can add, edit or delete posts, so losers simply never appear and timing is unprovable. |
| A wall of winning screenshots | No | No | An image proves an image exists; it says nothing about when the call was made or what was cropped out. |
| A self-reported spreadsheet | No | Sometimes | A count may be shown, but the author owns the file and can revise any row after the outcome is known. |
| A platform leaderboard | Rarely | Sometimes | Ranks accounts but seldom timestamps each call or shows the full losing set behind a headline figure. |
| A timestamped, graded ledger | Yes | Yes | Each call is anchored on-chain before it resolves and the full count, losers included, is published. |
Only the bottom row survives the test, and it is the standard the finished system here is built to. The three pages below take each component apart in turn — what it is, what a bad version looks like, and how a codified, A-to-D graded model supplies it.
The three components, one per page
A testable edge
Why a system must specify its edge precisely enough to be tested - and how a graded model puts a number on that edge.
A sizing rule
How a system caps the damage from any one trade, and how conviction-based sizing leans harder on the stronger calls.
A coded exit
Why the exit has to be written before the entry, and how levels fixed on-chain remove the temptation to improvise.