Every trader has intuitions. "This tends to bounce off that level." "Volume dries up before it moves." "It usually gives up the morning gain."
Some of these are real. Most are pattern-matching on a handful of memorable cases. The uncomfortable truth is that you cannot tell which is which while the idea remains a sentence in your head — because in that form it can quietly adjust itself to fit whatever happened.
Writing it as a rule removes that escape route. That is the entire point, and it is why the exercise is unpleasant.
Why vague ideas survive indefinitely
"Buy strong stocks on a dip" has no failure condition. Whatever occurs, the idea accommodates it:
- It worked — the rule is good.
- It failed — it was not really a strong stock.
- It failed again — that was not a proper dip.
Nothing can refute it. An idea that cannot be refuted cannot be improved either, which is why traders can hold the same instincts for a decade without getting better.
A rule that can fail is worth more than a rule that always seems right.
The four things every rule must state
Before an idea is testable, it needs four answers. If any is missing, you do not yet have a rule.
1. The universe. Which instruments? "Stocks" is not an answer. Liquid large-caps behave differently from thin small-caps, and a rule that works on one will often be meaningless on the other.
2. The entry condition. What exactly must be true? Not "when momentum is strong" but a comparison a machine could evaluate: this value above that value, on this timeframe.
3. The exit condition. What ends the trade, in both directions? An entry rule without an exit rule is half a strategy, and the missing half is where most of the outcome lives.
4. The timeframe. On what bars is this evaluated? The same comparison on five-minute bars and on daily bars are two unrelated strategies that happen to share a sentence.
Working an example through
Start with a typical hunch: "Stocks that break above their recent average with strong volume tend to keep going."
Untestable as written. Every term is elastic. Now interrogate it.
Which stocks? Say: liquid large-caps I would actually trade. That fixes the universe.
Break above what, exactly? "Recent average" could be anything. Pick one: the 20-period exponential moving average. Now it is a specific number that either is or is not exceeded.
How much above? Any amount, or a meaningful margin? A close above by any fraction is noise. Decide, and state it.
Strong volume compared to what? This is where the idea usually collapses, because "strong" was never defined. Either define it against a baseline, or drop the condition. Dropping it is honest; keeping it vague is not.
What timeframe? Daily closes, say.
When do I exit? The hardest question, and the one people skip. A price target? A close back below the average? A fixed stop? Decide before you know how it performs, because deciding afterwards is how you fit the exit to the result.
You now have something like: on liquid large-caps, on daily bars, enter when the close crosses above the 20-period EMA; exit when the close crosses back below it.
That is a rule. It might be a bad rule. But it can now be evaluated, which the original sentence never could.
What the PocketX cash strategy language supports
Knowing the available vocabulary shapes how you write the rule, so it is worth being concrete.
The cash strategy language supports one comparison, or a flat group of comparisons joined by ALL or ANY. The building blocks are:
- Indicators: EMA, SMA, RSI and VWAP.
- Price and volume fields: open, high, low, close, volume.
- Constants, for fixed thresholds.
- Five timeframes.
- An optional independent exit condition.
That is deliberately narrow, and the narrowness is useful. A language that cannot express a hundred nested conditions cannot be used to build a rule so complicated that it fits the past perfectly and predicts nothing.
If your idea does not fit in that vocabulary, that is worth knowing early. Sometimes it means the idea needs sharpening. Sometimes it means it is not a rule-based idea at all — which is a legitimate finding.
The strategy builder guide covers the mechanics of expressing this in the strategy workspace.
The exit is most of the strategy
Traders spend nearly all their effort on entries and almost none on exits. The ratio should be closer to reversed.
Entry determines whether you are in a move. Exit determines how much of it you keep, and it governs both the winners and the losers. A mediocre entry with a disciplined exit generally beats a clever entry with an improvised one.
Write the exit at the same time as the entry, before you have seen any results. An exit invented after looking at performance is not a rule — it is a description of what would have worked, which is a different and far less useful thing.
Note that the language allows an independent exit condition, so the exit need not simply be the entry reversed. Use that deliberately.
Stating what would make you abandon it
This is the step that separates a research process from a belief system.
Before testing, write down what result would make you discard the idea. A hit rate below some level. A drawdown beyond some depth. Fewer than some number of occurrences to judge on.
Write it down first, because afterwards every disappointing result acquires an explanation. The market regime was unusual. The period was unrepresentative. One outlier distorted everything.
Some of those explanations will even be true. That is exactly why the threshold has to exist before you need it.
Then, and only then, test it
With a written rule and a written failure condition, validation and backtesting become meaningful — you are asking a specific question with a pre-committed interpretation.
Two things worth carrying in:
- A backtest is a conditional statement, not a forecast. What a backtest actually proves sets out what it can and cannot support.
- Rules evaluate on closed bars. A condition is true when a bar closes, not the instant price touches a level intrabar. This shapes what a rule can realistically capture.
And when a strategy is armed, it produces alerts, not orders. Cash arming creates alerts only; it never places a trade. That boundary is explained in smart alerts and the execution boundary.
The one-page method
- Write the hunch in one sentence, however vague.
- Name the universe. Which instruments, specifically.
- Make the entry a comparison a machine could evaluate.
- Write the exit before seeing any results.
- Fix the timeframe. It is not a detail; it is half the strategy.
- Write the abandonment threshold in advance.
- Validate, backtest, and read the assumptions before the returns.
Most ideas do not survive step three. That is the process working. An intuition that cannot be written as a comparison was never going to be a strategy — and finding that out in ten minutes on paper is considerably cheaper than finding it out over six months in the market.