The brief decides the outcome
Whether your strategy is written by a freelancer, generated by a model, or built in a form, the same thing determines whether you get what you wanted: how completely you specified it. An incomplete brief produces a wrong strategy no matter who implements it, and the gaps are invisible until the code exists.
Here is what a brief has to answer, because Pine will answer it for you otherwise.
The nine questions
| Question | If you don't answer it |
|---|---|
| Which instrument, exactly? | slippage is counted in ticks, and a tick is a different size elsewhere |
| Which timeframe? | nothing in Pine records it; the chart decides, and anyone can change the chart |
| Do you act intrabar or on the confirmed close? | whoever writes it picks, and the two give measurably different results |
| What is the exit, fully? | you get the implementer's guess, usually the opposite signal |
| What happens when the entry condition fires again while a position is open? | pyramiding defaults to 1 [1], so it is silently ignored |
| What are the costs? | commonly omitted entirely, and the backtest becomes fiction |
| Fixed size or percent of equity? | percent-of-equity compounds and flatters a weak edge |
| Which date range was it validated on? | you cannot check the result later |
| What must not happen? | no maximum drawdown, no direction limit, no session restriction |
Note that six of those nine are settings in the strategy() declaration, not
logic. They are the part that decides your reported results and the part a brief
almost never mentions.
Write the rule as one sentence with no adjectives
Before the table above, do this:
Enter long when the 14-period RSI closes below 30 and price is above the 200-period simple moving average, exit at 2× ATR profit or 1× ATR loss, on EURUSD 1-hour, acting on confirmed closes only.
Every word in that sentence is a number, a named indicator, or a logical connector. If you cannot write it, the specification is not finished — and finding that out on paper costs an hour instead of an invoice. The full version of this exercise is in converting a manual strategy.
Words that must not appear in a brief
Each of these is a parameter you have not set, and the implementer will set it for you:
strong trend · near support · good volume · a clean break · when it looks extended · confirmed by momentum · reasonable risk
Replace each with a threshold and the indicator it is measured on. This is the single highest-value hour in the whole project.
Own the specification, not just the file
The specification is the asset. A Pine file is one expression of it, and if you later want the same idea on MetaTrader, a written spec ports cleanly while code does not — the two platforms have different execution models, so porting is a rebuild rather than a translation.
Keep the spec somewhere that is not the code, and update it when you change your mind. It is also the only thing that lets you tell, six months later, whether a strategy is behaving wrongly or behaving exactly as you asked.
Tactix AI — Studio vs Guide
Tactix AI is AlfaTactix’s product assistant brand (open Tactix AI).
- Tactix Studio turns a one-sentence strategy description into a draft across Timeframe, Signals, Filters, and Risk in the visual Strategy Builder. You review and edit every field before Code Generator writes MQL5 or Pine Script.
- Tactix Guide explains the step you are on — what to fill, what a control means, or how to phrase a rule — without dumping untested source code.
That is form-first automation: the LLM never replaces Code Generator, and you keep plan limits and real-time validation.