Does Automating Your Trading Actually Remove Emotion?

No. It removes emotion from the fill, not from you. Handing execution to a rule set takes the decision out of the one moment you are least equipped to make it, and then moves the feeling somewhere else. Two places, mostly: the day nothing qualifies, and the loss you sit and watch and disagree with. Both need a rule of their own.

MasterVP · Systematic Trading Journey

Written by the founder of KenKem, a software engineer who traded gold by hand for years before building the research stack behind it. As a self taught quant trader I spent about two of those years answering the wrong question, and nobody told me, because in trading education almost everybody is answering the same one. This is what I found on the other side of it. It is knowledge sharing, not coaching, and none of it is a recommendation to trade anything.

Where does the emotion actually go when you automate?

It moves off the trade and onto the process. That is the honest version of the claim, and it is a smaller claim than the one usually sold.

What automation genuinely removes is the improvised decision at the moment of highest arousal: the entry you take because price just moved, the stop you widen because the loss has not been realised yet, the size you double because the last one worked. Those are gone, and they were most of my early losses.

What it does not remove is you. You still watch. And the watching produces three new moments where a person can undo a system, quietly, in a way that looks reasonable at the time.

Why is the day when nothing qualifies the hardest day?

Because the honest answer, for years, was that I found something. I always found something, and the somethings I found on those days are where the account went.

For about two years my question was some version of "what is a good entry". I read everything and I got measurably better at it. The question that actually moved my results was duller: what do I do on a day when nothing qualifies? Not as a discipline slogan. As an actual specified behaviour, written down while calm.

Once abstaining became a real, defined, first class outcome instead of the absence of one, everything downstream got easier. Sizing got easier, because there were fewer and better trades. Review got easier, because a skipped day is a data point rather than a blank. The software got simpler, because a system permitted to do nothing needs far fewer rules than one that must always have an opinion.

FOMO, read this way, is not a feeling problem. It is a specification problem. The feeling arrives either way. What decides the outcome is whether you wrote down, in advance, what you are going to do while feeling it.

How do you give a system permission to do nothing?

By making "is today like the days the edge was measured on" the first question, before any entry logic runs.

In my own system there is a floor on how alive the market has to be. Below it, nothing participates. I will not publish where the floor sits, because that number is part of the product, but the useful half is why it exists: a method built around price leaving an area of agreement needs the agreement to have formed in the first place. In a market too flat to build one, the premise is absent and what you get is a stream of technically valid signals about nothing.

There is a second gate, and it is about cost rather than movement. Trading cost is a toll, so the system measures the broker's spread against how far the market is actually travelling and stands aside for the day when the toll is too large a share of what a trade could earn. No prediction involved. It is an affordability check, and it skips roughly one trade in three. That figure is specific to a low spread venue. On a wider spread feed the same rule would stand aside nearly every day, and saying so is part of the number being worth anything.

I found the volatility floor by studying the years the system lost money, which is exactly the kind of fix that should make a reader suspicious. So I tested it the harder way and let a blind search choose from a menu that included having no floor at all. Across all nine blocks where it had a choice it picked a floor, and never once picked zero.

What do you do when the system loses and you disagree with it?

Nothing. And I want to be honest that nothing is much harder than it sounds.

The setup is familiar to anyone who has run rules with money on them. The system takes a loss, you watched it happen, you had a view, the view was right and the rules were not. The trade you would have taken by hand would have worked. That moment is the whole test. Everything before it was preparation for it.

The reason I do nothing is not stubbornness. It is that in the moment I cannot tell the difference between a genuine flaw in the rules and a normal loss that happens to sting. Both feel identical from the inside. The only thing separating them is a sample size I do not have while I am sitting there.

So the disagreement goes into the research queue, not into the live account. If it is real it survives being tested properly, on data that includes all the times my instinct was wrong, which my memory conveniently omits. Most of them do not survive.

Why write the exit before the entry?

Because an exit written in advance is a decision made by someone who is not in the trade, and an exit decided during the trade is made by someone who is. Those are not the same person, and the second one moves stops.

There is a research reason as well. The entry is the part that feels like insight, so it gets all the attention, but an entry only decides which trades you are in. The exit decides what those trades are worth. I ran a broad head to head that changed nothing about entries or filters and varied only the profit taking geometry, and it produced a version that beat the previous configuration on every axis I measured, over a longer and harder window. Same ideas about the market. Different arithmetic about when to leave.

Why write the decision rule before you see the result?

Because the human mind is extremely good at generating a reason why a disappointing result is actually encouraging, and it generates that reason instantly and sincerely. You do not experience it as rationalising. You experience it as noticing something.

In my quant development process every experiment gets its pass and fail criteria written down before it runs. Then it runs. Then the answer is read against what was already written, not against what I now wish I had written. As of the August 2026 write up, 137 experiments have been registered that way and 122 of them did not become a product. I am fairly sure a meaningful number of those 122 would have survived if I had been allowed to decide what success meant after seeing the outcome.

This is the same defence as trading a system instead of a feeling. Same problem, different hour of the day.

How do you tell a frozen configuration from a fitted one?

Ask what was allowed to change between the periods, and whether the losing ones are shown.

The hardest test I know how to run on my own work is this. Freeze the shipped configuration, then score it on every consecutive six month block of the record in turn, so that no one gets to choose where the line is drawn. Then run a second version that is allowed to re choose its settings before each block from a menu of candidates, using only data available before that block. A simulated optimiser with no knowledge of the future.

As of August 2026, on release 1.19: eleven half year blocks, and eight of the eleven were positive. I lead with the three that were not, because they are the informative ones. They are 2021 and the first half of 2023, and the reason is visible rather than mysterious. This is a trend following breakout method and gold went sideways in exactly those blocks. A method that wins in every single period is either extraordinary or, far more likely, has been fitted to the periods it was tested on.

The stronger half of that evidence is not the eight. It is that the frozen configuration finished ahead of the optimiser that was allowed to re tune at every boundary. Beating a re optimiser is much harder to fake than beating a buy and hold line, because the re optimiser is allowed to cheat with hindsight and the frozen one is not. These are backtest results on real tick data, not live returns, and I keep watching the forward test so I can correct whatever shows up next.

Why does one rule have to mean the same thing at eight in the morning and nine at night?

Because otherwise it is not one rule. It looks like one rule and it behaves like several.

Gold does not act like a single instrument across the day. Asian hours are thin and orderly. The London open changes the character entirely. The New York overlap has the widest range and the busiest tape, which is where the opportunity is and where a wrong read costs the most. A rule expressed as a fixed distance is a completely different rule at each of those times, because the distance means something different once the range has tripled.

That is most of what "adaptive" means in my system, and it is much less clever than it sounds. Almost every condition it reads is normalised against the market's own recent behaviour rather than stated in absolute price, so the same rule carries the same meaning in a quiet session and a violent one. That normalisation is what lets a configuration stay frozen instead of being re tuned every quarter, and a frozen configuration is the only kind whose backtest means anything.

The clearest proof of the idea is a failure. Across 2024 the spread consumed about 8.6% of the market's average range, against roughly 3 to 4% in the surrounding periods. Nothing about the entries had broken. The edge was the size it had always been and the toll had more than doubled. The fix was not a smarter model, it was permission to stand aside, and together with the drawdown de risk it cut the worst peak to trough loss from about 57% to about 22% at identical sizing.

Why publish the failures instead of burying them?

Because in a field where everybody's marketing looks identical and none of it can be checked from outside, the only thing carrying information is what someone is willing to say against themselves.

So the rejection ledger is published rather than hidden. The out of sample years that lost money. The presets I withdrew. The headline figure I pulled the same day I published it, when the measurement basis turned out to be wrong. A bug in my own export that overstated results by about 11.5% on a commission charging broker, found and fixed on 20 August 2026. Three risk ideas that failed their own pre registered study, including the intuitive one: taking profit earlier made drawdown worse, not better, because the large winners are what climb out of a drawdown.

An operation that only ever reports its winners is not showing you a track record. It is showing you a filter, and you have no way to know how coarse the filter was.

What should you ask before believing anyone, including me?

Take the checklist rather than my word. These questions cost nothing and they are portable.

Ask how many configurations were searched before the one being shown to you was found, and whether the result has been discounted for that search. Ask where the out of sample window is, and whether it was defined before or after the result was known. Ask what the worst drawdown was, how long it lasted, and what account it is a percentage of. Ask where the chart starts and where it stops, and why those dates. Ask what has been withdrawn, and where that list is kept.

They work on me, they work on the next person, and most importantly they work on the strategy you built yourself, which is the one you are least equipped to be sceptical about.

Frequently asked questions

Does automated trading remove emotion completely? No. It removes emotion from execution. The feeling moves to the moments where you are watching rather than acting, mainly the day nothing qualifies and the loss you disagree with. Those two moments need written rules of their own, or the automation gets overridden by hand.

Is it normal for a systematic strategy to have losing periods? Yes, and a record without them should raise your suspicion rather than lower it. My own record has eleven half year blocks with eight positive and three negative, and the losing ones are 2021 and the first half of 2023, when gold ranged and a breakout method had nothing to work with.

Why does writing the decision rule before the test matter so much? Because after the fact you will find a sincere reason why a bad result is encouraging. Pre registering the pass and fail criteria removes that option. Of 137 experiments registered this way as of August 2026, 122 never became a product.

What is the difference between a strategy that is adaptive and one that is re tuned? Re tuning changes the settings. Adaptive, in the sense I use it, means the settings never change but every condition is measured relative to the market's own recent volatility, so one frozen configuration keeps meaning the same thing across quiet and violent sessions.

Should a trading system be allowed to do nothing? In my experience that is the most valuable thing it learned. Most retail systems have no vocabulary for abstaining, which is why they are always in the market, and why their worst losses cluster in the hours when nothing was really going on.

How do you know a backtest result is not just luck or curve fitting? Look at what was allowed to vary. A configuration held frozen across every consecutive block of the record, which then finishes ahead of an optimiser given free choice and honest access to all prior history, is much harder to fake than a single flattering window.

Does removing emotion mean removing risk? No. It removes the argument, not the risk. Drawdown still happens and it is still uncomfortable. What changes is that the response to it was decided in advance rather than at three in the morning.

What can I take from this without buying anything? The two rules. Write down what you do on a day when nothing qualifies, and write down what you do when the system loses and you disagree with it. Both are free, both are portable, and in my case they mattered more than any entry I ever found.

The method itself is worth having whether or not you ever use my tools. If you do want to see it running, MasterVP is my own intraday method built on top of public volume profile theory, and it ships as the Master Volume Profiler indicator on TradingView and MetaTrader 5 and as an Expert Advisor that executes on the settings you configure.

Educational purpose only. Not financial advice. Backtests and forward tests do not guarantee future performance, and every trader remains responsible for their own risk settings and decisions.