Consider a strategy that wins about a third of the time and makes roughly three times as much on a winner as it loses on a loser. Over a long record that is profitable, comfortably so. Now look at what it feels like to trade.
Losing four in a row happens constantly. Losing eight in a row will happen more than once. Twenty trades can pass with two or three winners in them, which on those numbers is roughly break-even or worse, and stretches like that are not rare events. In a system with this shape they are a routine part of the record, appearing over and over across a long history.
Nothing is wrong in any of those stretches. The edge is intact. That is simply what a one-in-three win rate looks like from the inside when you are living through it one trade at a time.
A trader who abandons a working strategy during a normal drawdown has destroyed a real edge with a decision that felt like prudence. This happens far more often than a strategy being wrecked by a subtle statistical error.
Concentration, and when to worry about it
Systems with this shape tend to earn most of their profit from a small minority of trades. It is not unusual for a large majority of the total return to come from a small fraction of the entries, with everything else roughly cancelling out.
Here it gets genuinely subtle, and you will see both of the following claims made confidently by different people.
The first is that concentration is a warning sign. Delete your five best trades, and if the whole result disappears, you may be looking at five lucky events rather than a strategy. That is a real and useful check.
The second is that concentration is expected. A design that cuts losses quickly and lets winners run is deliberately built to produce many small losses and occasional large gains. Removing its best trades and finding nothing left is not a scandal. It is the design working as intended.
Both are true, and the difference is whether the concentration matches what the strategy was built to do. If your system is supposed to grind out many similar small wins and instead its entire result rests on three outliers, something is wrong. If your system is supposed to catch rare large moves and its result rests on the rare large moves it caught, that is the system doing its job. What you cannot do is apply the check without knowing which kind you built.
The practical version: look at what remains after removing the best trades, and ask whether it looks like the strategy you designed. A trend system should leave behind a long tail of small losses, because that is what it is supposed to produce between the moves it is waiting for.
A mechanical version of the same concentration question is the regime analysis test, which splits a record by market condition and reports how much of the result rests on a single slice of it.
Decide the difference in advance
The problem with knowing that flat stretches are normal is that it also excuses every flat stretch, including the ones that mean your edge is gone. If your only tool is patience, you will hold a dead strategy indefinitely.
The fix is to decide in advance what a genuine breakdown looks like, while you are calm and have no position. Write down a specific threshold: a number of consecutive losses, a drawdown depth, or a win rate over a set number of trades that would be unusual enough to demand a review. Choose something that would happen rarely if the edge were intact, so that when it fires it means something.
Then write down what you will do when it fires, and make it a review rather than an exit. Cutting risk while you investigate is a decision you can make calmly in advance. Deciding what a bad month means while you are in one is not.
The value of writing it down first is that afterwards, every explanation sounds reasonable. Markets changed, volatility is different, it was an unusual period. Some of those will even be true. None of them can be evaluated by someone who is currently losing money and wants a reason to keep going.
It also cuts the other way. A bad stretch sometimes has a specific, checkable cause that is neither an ordinary run of losses nor a vanished edge, and the only way to find out is to decompose it rather than interpret it. One worked example is on the blog: the bad years were a toll booth.
Last lesson: what to do before you risk money.