5 / 5 · 7 min
When to declare an idea dead
A loss by itself proves nothing: a run of five losers happens almost always. Death conditions are set in advance — otherwise the decision is made by mood.
A drawdown and a breakdown are different things
A working system regularly goes into the red, and that is not a fault but its ordinary state. Measured over a hundred trades at a 50% win rate: a run of five losses appears in 95% of cases, a run of seven in half of them. In other words almost everyone who trades systematically will see five losses in a row within a hundred trades. If such a run makes you close the idea, you will be closing it constantly — including all the times when it is perfectly fine.
Death conditions are set before you start
The only way not to decide under pressure is to write down in advance the signs by which the idea counts as dead. The condition must be a number and a date, just like the idea itself. For example: I close it if expectancy is negative after a hundred trades, or I close it if the drawdown exceeded the figure computed for this system and has not recovered within fifty trades. A written condition protects in both directions: it stops you abandoning a working rule after three failures and stops you clinging to a broken one for years.
What a losing run does and does not mean
| Observation | What it means |
|---|---|
| 5 losses in a row | ordinary, happens almost always |
| 7 losses in a row | occurs for half of systems within a hundred trades |
| Drawdown deeper than computed | a reason to check position size, not the idea |
| Negative expectancy over 100+ trades | now this is a conversation about the idea |
| The entry condition stopped appearing | the market changed — the idea is not broken but inapplicable |
Three reasons ideas die
First: the effect was never there and the result came from fitting. Such an idea dies quickly — within the first hundreds of forward trades. Second: the effect was real but disappeared because many people found it. That is the fate of all simple rules: the more obvious a regularity, the faster it is used up. Third, the most common and the most peaceful: the market changed. Fees rose, volatility fell, the instrument lost liquidity — the rule remained true, but the conditions under it are no longer met.
Repairing the idea after every failure
When a rule stops working, the temptation to refine a parameter is nearly irresistible: add a filter, move the stop, exclude the awkward instrument. Every such edit looks like an improvement, but it has a price — the sample resets to zero. After the edit you are working with a DIFFERENT rule, and all accumulated trades no longer belong to it. Hence a simple limit worth writing down alongside the death conditions: how many edits are allowed and after how many trades. A rule repaired five times in a month has never been tested at all.
How to bury it properly
A closed idea is not wasted time if a conclusion remains after it. Write down three things: under what conditions the idea worked, under what it stopped, and what you now consider the cause. The second part is especially valuable: it turns a failure into a boundary of applicability, and the next idea starts beyond that boundary. Note separately whether the idea was testable at all — if it turns out it was not, that is the most important conclusion of all, and it will save more time than any working strategy.
The idea is dead — what now
The right next step depends on the cause of death. If the effect was never there, go back to the formulation and count the variants you searched; the answer is usually right there. If the effect was used up by the crowd, look for the same thing where fewer people look: another instrument, another scale, another market. If conditions changed, the idea goes into the archive with a note about the conditions under which to return to it — and that is not a figure of speech: volatility comes back, fees change, liquidity comes and goes.
The paper trading journal computes the result of every trade, and from it you can see both expectancy and the length of runs. That is the material for a decision: not a feeling that something is off, but the number of trades, current expectancy and the longest losing run set next to the one you deemed acceptable in advance.
About paper tradingWrite the obituary in advance
For the idea you are trading now, write down three closing conditions BEFORE they are needed: the number of trades after which you look at expectancy; the drawdown at which you stop; and the sign that the entry condition has stopped occurring. Put the note where you will see it on a bad day. Then test yourself with a question: if five losses in a row happen tomorrow, does that fall under any of the three? If it does, the conditions are written too softly and must be rewritten today rather than on that day.
Seven losses in a row. Is the idea broken?
Most likely not. At a 50% win rate a run of seven losses occurs for about half of systems over a hundred trades — that is an ordinary event, not a signal. What to look at is not the run but expectancy over the whole distance and whether the current drawdown fits the boundaries computed in advance. If expectancy is positive and the drawdown is within the calculation, you are simply in a normal bad stretch, and any action other than following the rule will make the result worse.
A rule worked for a year and then stopped. Should I repair it?
First establish the cause — it suggests the answer. If the entry condition stopped appearing, there is nothing to repair: the rule stayed true but the market no longer offers occasions for it, so it should be shelved rather than rebuilt. If the condition still appears while the result went negative, check external things: have fees risen, has the instrument's travel fallen, has the primary venue changed. Only if all of that is unchanged are we talking about the rule itself — and then any edit means a new sample and a new test from scratch.