Most trading strategies do not “die” due to psychology, market manipulation, or lack of discipline. They die earlier — at the level of probability.

Speaking narrowly and strictly on the topic: the problem is that a significant part of the strategies have a negative or too weak mathematical expectation, which means they are statistically doomed to distance.

Here we are not analyzing the costs, the scale of capital, or the trader's behavior. There is only one mechanism — the probabilistic construction of an unprofitable strategy.

Negative mathematical expectation: the main sign of a losing strategy

A popular question on this topic is: “why does the strategy drain the deposit if I do everything according to the rules?”

The answer may be unpleasant: because the model itself loses money on average.

The mathematical expectation is the average result of one trade with a large number of repetitions. Not the best result and not a successful month, but an average one.

Simplified:

if you repeat the same actions hundreds of times, will the result be positive?

Many strategies are created around a beautiful entry, but not around an average result. The result is a construction with:

1. high win rate,
2. small average profit,
3. rare, but large losses.

Psychologically, it's comfortable. Often right, it means everything is working. But the market does not consider the frequency of correctness. He counts the total on the balance sheet.

If the average loss statistically exceeds the average profit, the strategy will lose money even with “good accuracy". It's not a glitch. It's math.

Why does a high win rate not make the strategy profitable?

The query “is 70% of profitable trades enough to make money” appears regularly. Intuition says: yes, of course.

Mathematics answers: not necessary.

Let's imagine a model:

1. 7 out of 10 profitable trades,
2. average profit +1R,
3. the average loss is 4R.

The total for 10 transactions:

+7R − 12R = −5R.

The win rate is high. The result is negative.

The mistake here is simple: the win rate is a psychological indicator.

Mathematical expectation is economic.

The strategy can be profitable with 40-45% winning trades if the risk structure is balanced. And it can be unprofitable at 70% if the imbalance destroys the average result.

If the strategy is based around “being right more often” rather than “earning more than you lose”, the probability will sooner or later show the real result.

A series of losing trades: the norm of probability or a sign of a bad model

Another narrow query: “Is 5-7 cons in a row normal?”

Yes, it can be normal. Even a profitable strategy can result in a series of losses. This is called variance — the natural spread of results around the average.

But a nuance is important here.

If the mathematical expectation is positive and stable, the loss series remains a temporary deviation.

If the expectation is weak or negative, the series of losses becomes not a deviation, but a pattern.

Probability does not distribute profits evenly. She works in waves. And if the model does not have a statistical margin of safety, even the usual variance can destroy it faster than the trader has time to understand what is happening.

A weak strategy often “breaks down” not because the market has changed, but because it was initially based on a minimal advantage.

Overoptimization: why the strategy "works on history" but loses money in the future

The query “why is the strategy profitable in the test, but unprofitable in reality” is often related specifically to probability.

Overfitting is a situation where strategy parameters are adjusted to historical data so that the yield curve looks perfect.

But such a model does not have a real statistical advantage. She just remembered the past.

It's like a student who learned the answers to the last exam, but didn't understand the subject. It is lost in the new ticket.

If a strategy works only with precise parameters and falls apart with a small change in conditions, this is a sign of a weak edge.

A strong model maintains a positive expectation when the environment varies. Weak — loses it instantly.

Probability doesn't like fragility.

Why does the "zero" strategy eventually go into negative territory?

Sometimes the strategy really shows zero expectation in the tests. Neither a plus nor a minus. It seems that with perfect execution, you can “pull out".

But in a probabilistic system, the null model is already a problem.

If the expectation is 0, any small variation in the result, a deviation in the distribution of winnings and losses, or a shift in the structure of price movement moves the model into a negative zone.

Probability doesn't have to give you an even balance. It gives the distribution. And a distribution with a zero average easily goes into negative territory.

The zero strategy is not a safe zone. It's a shaky balance with no safety margin.

The broader picture of how negative expectations are reinforced by real market conditions and why most models fail to be tested by economics is discussed in detail in the material "The Real Economy of Trading: why most strategies do not work." Here we focus specifically on the probabilistic construction.

Weak statistical advantage: when the plus is too small

Let's say the strategy has a positive expectation of +0.05R per trade.

Technically— it's a plus.

Practically — minimal.

With this magnitude, any change in the distribution of winnings and losses can make the result negative for a significant period.

If a strategy wins only “on the edge”, it lives in a constant risk zone. One unfavorable series is enough, and the advantage disappears.

A strong advantage is not just a plus, but a plus with a margin.

The weak advantage is the illusion of stability.

Why are most strategies created without calculating probability?

The main system error is that strategies are built in reverse order.

First, the trader sees the pattern.

Then he comes up with an entrance.

Then he adds a stop.

And only then — if it gets there — does he count the result.

But the correct logic should be different.:

first, the probability structure and the average result,

then — resistance to a series of losses,

and only then — specific entry rules.

When the order is disrupted, the strategy may look neat and logical, but be mathematically weak.

It's like building a bridge without calculating the load. As long as people walk on it, everything is fine. As soon as the truck starts moving, the design will show the real limit.

Why 90% of trading strategies are mathematically doomed

If you put everything together in one structure, the picture looks like this:

1. the model is created without calculating the expectation,
2. The parameters are adjusted to the past,
3. The risk structure is unbalanced,
4. the advantage is either non-existent or minimal,
5. the usual variance ruins the result.

It's not a matter of talent. And it's not a question of a “bad market.”

It's a matter of ignoring probability when creating a strategy.

The market does not have to pay for beautiful logic. He only pays for the statistical advantage.

Practical conclusion: how to understand that the strategy is not doomed

If you remove everything unnecessary, the criterion is simple.

The strategy should:

1. have a stable positive mathematical expectation,
2. save it when changing the parameters,
3. to withstand a series of losing trades without destroying the model.

If the expectation is negative, discipline will not save.

If the expectation is zero, the probability will easily drag you into the negative.

If the expectation is minimal, the slightest shift in the distribution of results will destroy the advantage.

Trading is not a competition in the accuracy of forecasts.

This is probability management.

And if a strategy isn't built on math, it can't “fail.”

She'll just fulfill her statistical destiny sooner or later.