Settings Changes You Can Justify, And Ones You Cannot

"Should I change my indicator settings" is usually typed on a Friday, after a week that did not go the way you planned. The honest answer is that sometimes you should, and sometimes the urge to is the problem itself. Those two situations look identical from the outside. Same settings window, same keystroke, same five seconds of work. What separates them is where the reason came from.

This guide splits them. One kind of change comes from the structure of what you are measuring. The other comes from the chart you were looking at ten minutes ago. Learning to tell them apart is cheaper than learning it the slow way, over months, with a system that quietly changes shape every weekend.

Every setting change is either structural or cosmetic.

A structural change happens because the input no longer matches the thing it measures. You moved from the 15-minute to the 4-hour, and the lookback window you had is now covering a different amount of market. The number was right for one context and is wrong for another. Nothing about your last trade is involved.

A cosmetic change happens because the output disagreed with a chart you have already seen. You know where price went. You adjust the period until the indicator would have agreed with that outcome. This feels like research. It is not. It is memory wearing the clothes of analysis.

There is a single test, and it is blunt. Can you state the reason for the change without referring to a specific chart? If the justification contains the word "Tuesday", or a candle, or a trade you took, you are fitting. If the justification is about bars, sessions, volatility or timeframe, you are adapting.

It helps to drop one superstition first. Default values are conventions, not laws. Fourteen, twenty, fifty, two hundred — those numbers were chosen by people, for markets and timeframes that may have nothing to do with yours. They are not wrong. They are also not sacred. You are allowed to move them. You just need a reason that existed before you opened the chart.

A period is counted in bars, not in time. This is the part most traders know and then forget the moment they switch charts.

A 20-period average on the 5-minute looks back across a hundred minutes. The same 20-period average on the 4-hour looks back across more than three days. The number on the screen did not change. The window of market being described changed completely. So when you switch timeframe and leave every setting alone, you have already altered what the tool is telling you. Doing nothing is also a decision.

That is the clearest justified change there is. You moved timeframe and you want the lookback to cover roughly the same stretch of market it covered before. You can say that out loud to somebody who has never seen your chart, and they can check your arithmetic.

There are a few others in the same family. An instrument with a different volatility character may need a different distance-based setting — gold does not move like a quiet major pair, and anything calibrated in points or in volatility units has to account for that. A session-based tool may need a different window for the Asian range than for London. Each of these has a reason you could defend in a room.

Adapting a period also moves you along a trade-off that never goes away. Faster settings react sooner and flag more things that go nowhere. Slower settings confirm later and miss the start. You are not removing that cost by tuning. You are choosing which of the two errors you would rather make, repeatedly, in advance. Say which one you chose and why.

Here is the version that does the damage, and it is familiar.

You take a loss. You open the chart afterwards and look for what would have warned you. You see that if the oscillator period had been eleven instead of fourteen, it would have been stretched before the move against you. So you set it to eleven. It took four seconds, and it felt like learning.

It was not learning. You have shaped a tool using one sample, and that sample is a trade you already know the outcome of. Next week a different loss will suggest a different number. The month after that, you will not be able to say what your system is, because it has had nine versions and none of them ran long enough to tell you anything.

That is the real cost. Your results stop being evidence of anything, because the thing producing them keeps changing. You cannot review a week honestly if the week contains three different setups wearing the same name.

Three tells that a change is cosmetic. It was made after a losing trade rather than before a session. You can describe the chart it fixes but not the rule it follows. And you did not write it down — because writing it down would have made the reason visible, and the reason was embarrassing.

A fourth, quieter one: you changed two things at once. When two inputs move together you have given up the ability to attribute anything to either.

One rule covers most of this. No setting changes while a chart is open on the instrument you just traded.

When you do change something, the entry goes in the journal before the value moves, and it has five lines. The date. The old value and the new one. The reason, stated without naming a single candle or trade. The sample you will judge it over — a number of sessions, fixed now, not decided later. And the condition that sends you back to the old value.

One variable at a time. Then forward test it over the sample you committed to, with the setting left alone the whole way through. Hands off is the entire point. If you adjust mid-sample, the sample is gone and you start again.

Most of the value arrives before any of that. If you cannot fill in the reason line without naming a trade, you have your answer, and you have it in thirty seconds. A change you can justify survives a chart you have not seen yet.

The MMFX indicators sit inside a fixed routine, and that is what makes this manageable. The daily SOP taught in the course runs in a set order — mark, bias, classify, wait, confirm, execute, manage, journal. The indicator lives at a known step. It highlights potential zones and structural context on the chart. It does not predict anything, and it does not decide for you.

Because the step is fixed, a settings change has a describable consequence. You can say which part of the routine it affects, and you can watch that part across a defined sample instead of guessing from a handful of trades. Run one documented configuration at a time, through the whole workflow, and let the journal do the arguing. The settings serve the process. The process does not get rebuilt around a chart you have already seen.

Trading carries risk of loss. Every decision you make is your own, and nothing on this page is financial advice.