What a signal actually is
An arrow on a chart means one thing: a set of conditions written by someone, at some point, in some language, evaluated to true on this bar. That is the whole of it.
It does not mean the move is likely. It does not mean the author tested it. It does not even mean the conditions are the ones the description says they are — I found three cases in my own scripts where a filter was described, drawn on the chart, reported as active in the panel, and had no effect on a single signal.
So the question is never "did the indicator fire". The question is what the market is doing, and whether this particular instance of the pattern is one of the ones that works.
The part that can be written down
In the previous note I measured one of my own tools and found it did not hold up. The more useful finding was not the number. It was why the mechanical version underperformed the way I actually trade.
Reading my own rules back, most of them cannot be written as conditions.
"I take more size on a stronger setup." How much stronger? There is no threshold behind that word. It is an assessment made from the whole screen at once.
"Order flow is optional confirmation." A rule cannot have an optional input. Either it is in the condition or it is not, and both versions are wrong.
"If the open shows a large one-sided push, I go with it." Large compared to what? I know it when I see it. That sentence is not a specification.
The script implements the part of my process that can be written down, and that turns out to be the smaller part. The rest is reading — and reading is the thing worth building.
Confirmation, not trigger
There is a version of indicator use that works, and it is narrow.
You look at the chart. You form a view about what is happening: where the move came from, where it stalled, what the session has been doing, where price would have to go for your idea to be wrong. You decide what would have to happen for you to act.
Then the signal arrives, or it does not.
Used this way the indicator is a second opinion on a view you already hold. It adds a little, and it is allowed to add a little, because the work was already done. The signal is checking you, not leading you.
The failure mode is the opposite order: the arrow appears, and you build a reason for it afterwards. That reason will always be available. The chart is large and something on it will always support whichever direction you are being pointed in.
The test is simple. If the signal had not appeared, would you have been looking at this instrument at all? If the answer is no, you are being led.
When the signal makes you hesitate
This is the part almost nobody writes about.
Sometimes you have a view, the signal arrives, and something in you does not want to take it. The setup matches, the conditions are met, and you are uneasy.
Most of the time that unease is information. It is the part of your reading that has not been put into words yet — the gap between the conditions the script checks and the much larger set of things you have seen before. Session timing. The shape of the last failure. How the instrument has behaved all morning. The fact that the move has already run.
The right response is not to override it with discipline, and not to obey it blindly either. It is to write it down. "I did not take this one because the move had already run twenty points before the signal." Three weeks of those notes and you will have found a condition you did not know you were using.
A trade not taken costs nothing and teaches something. That is a better deal than most trades.
How to turn reading into something testable
Vague intuition is not a method, and a rule invented at a desk is not one either. The way across is slow and it works:
Observe. Notice something that keeps happening. Not a theory — an observation. "Price often stalls at the previous day's high before the cash open."
Define. Write it so that another person could point at it on a chart and agree it happened. If you cannot, the observation is not ready. This step kills most ideas, and it should.
Expect. Before looking at any data, say what you expect to find. "It stalls more often than it goes straight through." Writing the expectation down first is what stops you from reading the result you wanted.
Measure. Count it. By hand is fine — fifty instances counted honestly beats a backtest of a rule you have already tuned.
Result. Write what came out, including when it is nothing. Most observations do not survive this step. That is not failure, that is the point of the step.
Decide. Keep it, drop it, or narrow it. Then go back to the first step.
This is slower than installing an indicator. It is also the only part of this work that accumulates. An indicator you downloaded teaches you nothing when it stops working, because you never knew why it worked.
The useful paradox
An indicator is most useful to someone who could do without it.
If you can read the chart, the signal saves you attention: it watches twenty instruments while you watch one, and tells you where to look. That is real value.
If you cannot read the chart, the same signal is a way of making decisions without having to understand them — which feels like a shortcut and works like a tax.
Every tool on this site is built for the first case. That is also why each one has a section about where it falls short, and why one of them is not published at all.