Your strategy works in backtest but you lose live. Here is why.

You ran your strategy over the last five years. The backtest is clean. The curve climbs. The R:R is solid. You start live. And then, for three weeks, you lose.

If you are asking yourself why your strategy works in backtest but you lose live, you are not alone. It is probably the most common mismatch I see in the traders I talk with. And in the large majority of cases, the answer is not in the strategy. It is in the person executing it.

I am going to show you why a backtest always lies a little. And what actually happens, in real time, between the moment you see your setup and the moment you click.

A backtest is a trader with no body. You are not.

A backtest is a simulation where the strategy runs without you.

No fatigue. No doubt after two losses in a row. No memory of yesterday's winning trade pushing you to double your size. No tweet that makes you question your setup while you wait for your entry. No rough 6am wake-up. No message from a friend asking "what are you trading right now?".

The backtest is neutral, mechanical, clean. You are human. That is the whole difference.

This gap, between the perfect simulation and the real trader, has a name worth keeping. The execution gap. It is what makes the same strategy produce two completely different performance curves depending on who is at the controls.

Backtest. Testing a trading strategy on historical data to estimate a hypothetical past performance. The backtest does not include the human factor: hesitation, deviation, fear, overconfidence. That is its main limitation.

Why does my strategy work in backtest but not live?

Because a backtest executes the strategy perfectly, with no hesitation, no deviation, no emotion. You hesitate, you exit too early, you change your size depending on your mental state. The strategy has not changed. Your execution has.

That mismatch is what most traders read, wrongly, as a "strategy problem".

The execution gap: what the backtest never measures

Your backtest assumes three things that are false in live trading.

First, it assumes you will take 100% of the signals. In reality, you will skip some. Either because you doubt the context (when your strategy does not ask for your opinion on the context). Or because you just lost two trades and you prefer to "calm down a bit". Or because you happen to be on the phone at that exact moment.

Second, it assumes you will exit exactly where your plan says to exit. Live, you will exit before. At 1R instead of 2.5R. Because you are afraid it will reverse. Because you want to "secure". Because last time, your trade that was at 2R came back to zero, and you never want to feel that again.

Third, it assumes your sizing is constant. Live, your sizing will climb after a run of winners (overconfidence) and shrink after a run of losers (fear). Except your edge is calculated on constant sizing. You break your edge without noticing.

None of these three elements show up in a backtest. And yet, that is exactly where everything plays out. Between a strategy that resembles your backtest and a strategy that destroys you live.

Psychological slippage. The gap between the execution your plan prescribed and the execution you actually performed, caused by your emotional state. Not to be confused with technical slippage (latency, spread). Psychological slippage comes from you, not from the market.

The five places where your execution drifts from the plan

When a trader tells me "my strategy works in backtest but not live", I ask them to describe their last five trades in detail. Five places of drift come back almost every time.

1. The delayed entry. The signal appears. You hesitate. You wait for "one more confirmation". When you finally get in, the move is already halfway done. Your R:R is degraded. On the backtest, the entry was instant. For you, it took thirty seconds of doubt.

2. The moved stop. You place your stop, you watch price approach it, you move it "just a little further out". Your plan does not say that. You are refusing the loss.

3. The premature exit. You are at 1R, your plan says 2.5R, you exit. You secure. The market goes and grabs 3R without you. The backtest had captured those 3R. You took half of them.

4. The variable sizing. After a run of losses, you cut your size "to limit the damage". The winning trade finally arrives, except you are risking 0.5% instead of your usual 1%. You miss half of what your edge was supposed to return.

5. The off-plan trade. You see a setup that is not in your strategy, but that "speaks to you". You get in. That is discretion, not strategy. Often it is a need for control, or a small piece of revenge after a loss whispering "this one I feel, I am going to make it back". And the backtest never saw it.

Each of these five places is a leak. Strung together, these leaks can turn a strategy that is profitable on history into a losing strategy live. Without a single line of the strategy itself having changed.

What makes the diagnosis so hard is that the leaks are not spectacular. None of them ruins you in a single trade. Each one costs a small fraction of your edge. Added up, they are the difference.

A word on slippage, costs and overfitting

I focus on the behavioral gap because it is the one almost everyone underrates. But honesty requires naming the technical reasons too, because they are real and they stack on top.

A backtest often fills you at the mid price. Live, you cross the bid-ask spread on every entry and exit, and you pay commissions. On a high-frequency or scalping strategy, that friction alone can turn a positive backtest negative. The faster you trade, the more this matters. Add the occasional technical slippage during a volatility spike, and the gap widens further.

Then there is overfitting, the trap of a backtest that looks perfect because it was tuned to look perfect. If your edge only survives with one very specific set of parameters, on one specific period, it probably did not find a real pattern. It memorized noise. A strategy that breaks the moment you nudge a single parameter was never solid to begin with, it was decorated.

So the sequence I trust is simple. Rule out the obvious technical leaks (are your costs modeled, is your fill realistic). Sanity-check for overfitting (does the edge survive on data the strategy never saw). And only then, if the strategy still holds up on paper, accept that what remains to read is your execution. In most discretionary accounts I have seen, the human gap is bigger than the technical one. But you do not have to choose, you check both.

Overfitting. When a strategy is tuned so tightly to historical data that it captures random noise instead of a real pattern. It looks excellent in backtest and collapses on new data. A common sign: the edge disappears as soon as you change one parameter or test a different period.

What I understood in 2020

In 2020, I had a swing strategy I had backtested over three years. The curve was clean. I launched it live. After two months, I was losing.

My first reflex was the one every trader has: question the strategy. I backtested again. I changed a parameter. I tested on another instrument. Always profitable in backtest, always losing for me.

Then I did something I had never done seriously before. I logged every one of my trades. Not just the entry, the stop and the target. I logged what I felt before clicking. I logged where exactly I had deviated from the plan. I logged why I had exited at 1R when my plan said 2R.

After a month of honest journaling, the pattern jumped out at me. I was skipping almost half of my signals. I was exiting, on average, well before my target. And after every small loss, I skipped the next signal "to calm down".

My strategy had never been the problem. My execution was. And I had no way to see it without the journal.

That experience, among others from the same period, is what made me build a very different kind of journal afterward. It ended up becoming the central brick of Hyvirtus.

The only way to close the gap: read your execution

If you recognise what I am describing, you have two options.

The first is to keep switching strategies at every losing stretch. That is what the large majority of pre-profitable traders do. I did it for my first years. It does not work. The new strategy will produce the same execution gap, because the person executing has not changed.

The second is to read your execution. Meaning: to put what your plan said side by side, in cold blood, with what you actually did. Not to beat yourself up. Not to tell yourself "I should have". Just to see.

That reading takes a journal. Not a journal that logs only the entry price and the exit price (your broker already gives you that). A journal that logs the state you were in before, where you drifted during, and what you felt at the exit.

Without those three layers, you read numbers. With those three layers, you read your execution.

This is exactly what Hyvirtus tries to make legible during the session. Dojo Live captures those drift points trade by trade, in real time, and Hyvirtus reads back what keeps coming up when you break from your plan. Not a notebook. A behavioral reading that shows you where your execution moves away from the plan.

Trading plan. A document that fixes, ahead of time, your rules for entry, exit, stop, sizing and filters. The plan exists so you decide in cold blood what you would do under pressure. Without a plan, the execution gap is not even measurable.

What changes when you start reading your execution

Three things, in this order.

First, you stop changing strategy every two months. You understand that the strategy is almost never the problem, except for the extreme cases you will identify quickly with the checks above.

Then, you identify one or two recurring execution patterns. Often it is the premature exit. Or the moved stop. Or the delayed entry. You do not have ten of them. You have two or three. Far simpler than you thought.

Finally, you can work on one precise point at a time. Not "improve my strategy". Not "become a better trader". Something precise like "stop moving my stop". Over two weeks. And you watch the curve straighten up.

It is less glamorous than "switch strategy for the umpteenth time". It is more effective.

The honest way to run this, once you trust the strategy on paper, is forward testing, small. Take it live with the smallest size you can stomach, and treat the first dozens of trades as data on you, not on the strategy. The point of that small size is not to make money. It is to make the execution gap cheap enough to study while your real reflexes show up.

Personally, I think this is the most important turning point of my early years. The day I accepted that the execution gap was my real subject, I stopped looking for the magic strategy. If you are still wondering why you lose live while your strategy works in backtest, that is almost always where the answer sits.

FAQ

How long does it take to close the gap between backtest and live?

There is no fixed timeline. What matters is the number of trades you execute and log with attention. In my experience, an honest reading of your execution over a few dozen trades starts to surface the patterns. It is not a question of calendar. It is a question of volume of behavioral data.

Is my strategy necessarily good if it works in backtest?

No, but it is a good start. A backtest on few trades, on a single market period, or with obvious overfitting, can be misleading. That said, in most of the cases I see, the problem is not the strategy, it is the execution. Checking your execution before questioning the strategy is the right order.

Why do I always exit my winners too early?

Very often, it is loss aversion talking. Watching an unrealized gain fall back to zero hurts more than never having had it, so you secure the position. The problem is that your edge is calculated on trades that run all the way to their target. If you cut half of your winners at 1R, you break that edge.

Do slippage and costs explain the difference on their own?

Partly. A backtest filled at the mid price ignores the spread you actually cross, plus commissions and the occasional technical slippage. On a high-frequency strategy this alone can turn a positive backtest negative. But for most discretionary traders, the technical friction is smaller than the behavioral gap. Both matter, the human one usually more.

Why did my backtest never predict my live losses?

Because the backtest never simulated you. It simulated a strategy. As long as you see the strategy as the main variable, the backtest looks unbeatable. The day you realize that the main variable is you, you understand why the backtest could not see what was coming.

In short

The backtest validates a strategy, never a trader. The gap between the two plays out in the execution: anticipated entries, moved stops, variable sizing, emotional exits, off-plan trades. Slippage, costs and overfitting can widen it, and they deserve a quick check first. But once the strategy holds up on paper, what remains to read is the practice. According to ESMA, the European securities regulator, between 74% and 89% of retail investor accounts lose money trading CFDs. A large part of that number lives in the execution gap, not in the strategy.

Reading your execution is what Hyvirtus is built to structure: The Ritual sets the frame before the session, Dojo Live captures the drifts during it, The Timeline shows what repeats. If the strategy holds in backtest, what is left to read is you executing it.

← All articles