The order most people use
It usually goes: find an indicator that looks convincing, put it on the chart, trade its signals, lose money, find another one. After enough rounds of this the conclusion is that indicators do not work.
The conclusion is wrong, but the method that produced it was always going to produce that conclusion. An indicator is the last piece of a method, not the first. Choosing it first is like buying a ladder before knowing which wall you are climbing.
The order that works is: constraints, style, timeframe, tools. Each one narrows the next. By the time you reach tools, the choice is usually obvious and there are very few candidates.
Why there are so many tools
Partly because the underlying calculations are few and the possible packagings are endless.
Strip the branding off most indicators and you find one of a handful of things: an average of past prices, a pivot high or low, a range measured from an average, a split of volume into two sides, a relationship between two timeframes. Almost everything on any platform is one of those, or two of them combined, with a name on top.
I should be specific about my own side of this. I have twenty-nine published scripts. They are not twenty-nine ideas. Underneath they are about twelve, fitted to fourteen strategies I have actually traded — the same calculation adjusted for a different timeframe, a different instrument, a different amount of patience. That is less impressive than twenty-nine and more honest. It is a workshop floor rather than a product line, and I am currently going through it one script at a time.
So when a catalogue has a hundred tools, the honest reading is not that there are a hundred edges. It is that packaging is cheap.
Start with constraints, not preferences
Before style, before charts: when can you actually watch a screen, and for how long without your attention degrading?
This is not a motivational question. It is the hard constraint that determines everything downstream, and it is the one most people skip because the honest answer is inconvenient.
If you have ninety uninterrupted minutes at a fixed time of day, that is a different method from four hours of fragmented attention, which is different again from checking the chart three times a day between other work.
Two more constraints worth writing down before anything else:
How much can you lose on one trade without it affecting the next decision. Not what you can afford — what you can lose and still think clearly twenty minutes later. These are different numbers and the second one is smaller.
This is also the honest answer to why the same strategy makes money for one person and loses it for another. The rules are identical. The patience is not, and neither is the ability to take a loss without it becoming a bad decision five minutes later.
A method that waits three days for a setup is profitable only for someone who can wait three days. Give it to someone who needs activity and they will take the near-misses on day two, and the method they are running is no longer the method that was tested. A method with a wide stop works only for someone who can watch the position go against them by a planned amount and leave it alone. The plan says that is normal. Whether it feels normal is a different question, and it is the one that decides the outcome.
So when a strategy that works for its author fails for you, the first thing to check is not the rules. It is whether the strategy assumes a tolerance you do not have. That is not a character flaw. It is a specification, and the fix is to find a method whose specification matches you — not to try to become a different person while holding a position.
How many decisions per week you want to make. Some people need activity to stay engaged. Others make worse decisions the more they make. Neither is wrong, but they lead to different methods.
Three honest answers to those questions eliminate about eighty percent of the possible approaches before you have looked at a single chart.
The timeframe follows
A timeframe is not a preference. It is the consequence of how long you can hold attention and how much you can risk per trade.
A five-minute chart demands presence. Setups form and die in minutes, and if you step away mid-trade you are not managing it, you are hoping. It suits a block of uninterrupted time, and it produces many small decisions.
An hourly chart is forgiving of absence. A setup that forms over six hours will still be there if you look away for twenty minutes. It produces few decisions and asks you to wait, which is harder than it sounds.
The stop distance scales with the timeframe, and so does the position size you can take for the same risk. A wider stop on an hourly chart with the same risk means fewer contracts — which is arithmetic, not preference, and it is why timeframe and risk cannot be chosen separately.
Pick the one your constraints allow, not the one that looks most interesting. Then stay there long enough to learn its character. Every timeframe has one, and it takes weeks of watching to pick up.
What a tool is actually for
A tool can answer one of a small number of questions. You need at most three on a chart, and they must answer different questions.
| Question | What answers it |
|---|---|
| Which way is the bigger picture pointing? | A smoothed average, usually from a higher timeframe |
| Where has price reacted before? | Pivots, levels, gaps |
| Is something happening right now? | Volatility, volume, a break of a recent level |
| How much room is there? | A range measured from an average |
Three tools that answer three different questions give you a reading. Three tools that answer the same question give you a vote — and you will unconsciously weight the vote toward whatever you already wanted to do.
That is the most common chart mistake I see, and it does not look like a mistake. It looks like confirmation.
The entry is the smallest part
A trade has four decisions: where you get in, where you are wrong, where you get out, and how much you take. Most attention goes to the first. Most of the outcome comes from the other three.
In the first note on this site I measured one of my own tools: a twenty-nine percent hit rate, consistent across three timeframes. At a 1.5R target that loses money. At 2.5R it barely produced signals. The entry was the same in both cases. What decided the result was the distance to the target and the distance to the stop.
A better entry would have helped. Better sizing and a realistic target would have helped more, and they do not require finding anything.
Choosing a tool, concretely
Once the style and timeframe are set, the choice is narrow:
One tool per question, maximum three. If a new tool answers a question you already have covered, it is not an addition, it is a replacement — and you should run them side by side for a while before deciding.
Run it on your timeframe for a month before trusting it. Not backtested. Watched. You are not checking whether it is profitable; you are learning what it does in a range, in a trend, at the open, in a dead hour. That knowledge is what lets you ignore it at the right moments.
Check that it does what it says. There are nine questions on this site for exactly that, and the most useful one takes two minutes: turn a filter off and count whether the number of signals changes. In three of my own scripts it did not.
Count the times you did not take the signal, and why. After a month that list is more valuable than the signals themselves. It is the written form of the reading you are developing, and it is the only part of this that compounds.
What I use
Little, and on one timeframe. Levels from a higher timeframe to know where the day's boundaries are, something that tells me what has happened in the current move, and a written note of the setups I take and skip. I trade one session, on futures, and I stopped trading another session because it kept costing me — a decision that came from damage rather than measurement, which is its own kind of honesty.
That is not a recommendation. It is the shape my constraints produced. Yours will produce a different one, and the point of all of the above is that the difference should come from your constraints, not from which indicator you happened to find first.