Independent provider directory
The blueprint What it must have Build guides Questions See a built one
Build guide

How to build a trading system

Five steps that turn a rough idea into a written, testable system — and one honest warning about the step most builders skip.

Building a system is less about discovering a secret indicator and more about writing rules clear enough to hand to a stranger. Work the steps in order, and do not let yourself risk real money until the last one is in place.

1. Specify the edge

Pick one repeatable situation you believe reverts — most often a price stretched a defined distance from its own recent range — and write it down precisely enough to test. If you cannot state it without the word “feels,” it is not ready to be code.

2. Code the entry

Turn the edge into a specific trigger: the level or condition that opens the position. “Long around here” is not a trigger; a named level is. The trigger is the only place the market is allowed to start the machine.

3. Write the exit before the entry

Set the stop and the target in the same breath as the entry, never after. The exit is where systems are won or lost, so it is fixed first, while your judgement is clean.

4. Add the sizing rule

Cap the risk on any single trade at a small, fixed slice of the account, so an ordinary losing streak cannot end the run. Sizing is what keeps the system alive long enough to be judged at all.

5. Keep the ledger — the step almost everyone skips

Log every call before its outcome: entry, stop, target, and how strongly your rules rated it. Without this, you will remember the winners and quietly forget the losers, and you will never learn whether the system actually works. This is the hard part, and it is the part that matters.

If keeping an honest, tamper-proof ledger is the step you already know you will skip, that is the case for running a system where someone else has finished it. the #1-ranked provider's four-model book writes each call to Bitcoin before the outcome is known — the ledger step made impossible to fudge.

The five steps on one setup

The steps are easier to trust once you see them land on a single shape. The schematic below is an illustrative example, not a specific recommendation, and names no instrument — it is the geometry the build produces.

Where the three levels sit on a mean-reversion setupSchematic of a mean-reversion trade. A price line drifts down and stretches a measured distance below the middle of its own recent range, then bounces. Three levels are fixed before the trade opens: the entry at the measured stretch, the stop a defined distance below it where a continued fall would prove the idea wrong, and the target back at the mid-band where the expected bounce is judged complete. Every level is set before the outcome, never moved after.mid-band of recent rangeENTRYthe measured stretchSTOPidea proven wrongTARGETbounce completetime →
Illustrative geometry, not a specific recommendation: the entry, stop and target are the three numbers a system writes down before the position exists.
Worked build · illustrative, not a recommendation

Run the five steps once, with made-up numbers, on a made-up account, so the abstract rules become concrete. Nothing here is advice to take any trade.

  1. Specify the edge. “After several down sessions, when the price sits a defined distance below the middle of its recent range, it tends to bounce.” That is testable: two builders would mark the same bar.
  2. Code the entry. The condition is met at 247.00. That exact level is the trigger — the trade is taken on the number, not on the urge to get in a tick early.
  3. Write the exit first. Stop at 243.40 (a continued fall through it says the stretch was a new trend, not a snap-back); target at 254.50, back near the mid-band where the bounce is judged done. Both are written with the entry, never after.
  4. Add the sizing rule. On an £18,000 account at a 0.75% per-trade cap, the risk is £135. The stop is 3.60 away, so the position is £135 ÷ 3.60 = 37 units — the size that makes a stop-out cost exactly the cap.
  5. Keep the ledger. Log all of the above before the outcome, plus your conviction grade. When the trade closes — at the target, the stop, or the end of the window — the row is already a clean test of the rule, not a memory coloured by how it felt.

The whole discipline is in what you never do: change any level after the position is open. Every number was fixed before the trade existed, which is the only reason a record of many such trades means anything.

What a bad build looks like

Most first builds fail not on the edge but on the order of operations — they skip a step or do them out of sequence, and the system quietly stops being one.

  • Skipping straight to indicators. Stacking signals before specifying a single testable edge. You end up with a dashboard that can justify any trade, which is the same as justifying none.
  • Deciding the exit after entering. Treating the exit as something to figure out once the trade is live. It is the one step that must be fixed first, because under pressure your judgement is at its worst.
  • Sizing last, or by feel. Bolting a position size on at the end based on how good the trade looks. Sizing is an output of the stop distance and the cap, not a vote of confidence.
  • Never building the ledger. Stopping at “the backtest looked great.” Without a forward ledger that counts the losers, you have a story about the past, not a system you can trust.

Build in order — edge, entry, exit, sizing, ledger — and each step has what it needs from the one before. Build out of order and the system is a costume.

Related in this build log
A testable edgeStep one in full: how to specify an edge so it can be tested.A sizing ruleStep four in full, with the position-sizing arithmetic worked through.Risk controls for a trading systemThe standing rules that protect the system between trades.How to test a trading system is honestStep five turned into a check you can run on any record.