A Losing Streak: Where Statistics End and a Broken System Begins

A losing streak proves nothing by itself. At any share of profitable entries other than one hundred per cent, runs of failure are built into the distribution of results: they are bound to happen, and the longer your trading history, the longer the streak you are guaranteed to find in it. So the first question after a run of losses is not "what should I fix" but "is this streak any different from the ones my system produces normally". As long as the answer is "no", the right action is to change nothing except position size. As soon as the answer is "yes", what needs changing is not the entry but the assumption the entry is built on.

Why streaks are inevitable and why that is no comfort

The share of profitable trades describes a long-run frequency, not the order of outcomes. Outcomes arrive in clusters — that is a property of a random sequence, not bad luck. A practical consequence follows: you need to know the length of a "normal" streak in advance, before it has begun. Pull your entire history out of the trading journal and look at the longest run of losses in past periods. That is your benchmark. A streak that does not exceed the historical maximum is working noise. A streak that has clearly gone beyond it is a reason for a check, but not yet a verdict.

There is little comfort in this, however. The statistical normality of a streak says nothing about whether your account and your nerves will survive it. The account survives it if the risk per trade is chosen so that the worst run observed in your history leaves the capital in working condition. Your nerves survive it if the decision to pause was made in writing and ahead of time — because inside a drawdown that decision never gets made.

What separates a drawdown from a breakdown

A broken system looks different from bad luck, and the difference shows in signs that are visible in the journal.

The market regime has changed. A breakout strategy stops working when the market moves from a trend into a range — and that is not a failure of the strategy but a failure of the environment to match its assumption. The check is simple: the losses are concentrated in trades of one type and one direction.

Execution liquidity has changed. If slippage and the spread in your instruments have widened, the return drains away into costs even though the signals are still correct. This shows up when you compare the intended entry price with the actual price in the trade tape.

You have changed. This is the most common case. The losing trades in the streak do not follow the rules: an entry ahead of the signal, a stop that has been moved, a position enlarged "to win it back". Formally this is a losing streak of the system; in fact it is a streak of violations.

A protocol for the duration of the streak

Step 1 — fix the amount of risk per trade and do not change it upwards until the review is finished. If you reduce it, reduce it one time, not every day.

Step 2 — label every losing trade in the streak as "by the rules / not by the rules". The split is made from the note written before the entry, not from memory. How to keep such a record is covered in detail in the piece on the investor's journal; the definition of the tool itself is in the glossary.

Step 3 — if there are no violations, look at external factors. Did the streak fall in a period of earnings reports and corporate events, when your signals are noisy by definition? Check the events calendar and the news feed. A strategy that was not designed for event-driven moves should not be traded in such weeks.

Step 4 — check whether the sample is sufficient at all. Conclusions about a "breakdown" drawn from a short history are the beginner's most expensive mistake, because they lead to endless reworking of a system that works. What counts as a sufficient number of observations is covered in the entry number of trades — number of trades.

Step 5 — a pause, if the streak has exceeded the historical maximum or if the share of violations in it is large. A pause means stopping trading for a period named in advance, not "I'll trade a bit less".

What the protocol does not give you

An honest caveat: not one of the steps answers the question "will profitability come back". It answers only the question "do I have grounds to continue". Separating execution from the idea, knowing your own worst run and keeping risk fixed preserve the account long enough for the statistics to assert themselves — that is all that lies within your power.

It is also worth remembering that some trades take place outside the regular order book — in the negotiated trades mode — and their result cannot be compared directly with exchange execution. Mixing them into one set of statistics distorts the picture of the streak.