Consider a hypothetical trader who has eight wins, then opens a position at five times the usual size without a stop. A subsequent loss may erase earlier gains. The lesson is the unplanned exposure change; this is not an author’s observed case or proof that the strategy was sound.

Confidence and cockiness are two different things

Confidence in preparation is different from certainty about the next price. State the evidence behind the decision, its limitations and the condition for declining or cancelling it. Several hundred observations are not a universal certificate of an edge: data quality, costs, rule selection and changing conditions matter.

“I know what I will do in this scenario” concerns a procedure. “I know the market will move my way” concerns a forecast. Keep them separate. A considered pause or reduced exposure may be consistent with the plan rather than evidence of weak confidence.

Where earned confidence comes from

Build confidence from a written method, a complete record and consistent risk checks. Several hundred trades provide sample estimates of win rate, payoff and net expectancy; they do not reveal immutable future probabilities. Review costs and changes in conditions as well as adherence to rules.

A loss need not be a verdict on personal ability, but its financial consequences still belong in the review. Check whether it was consistent with the method and whether current limits permit further exposure. The article on a trader's acceptance of loss separates acknowledging the result from ignoring its consequences.

How cockiness shows up in practice

Review volume changes, skipped conditions and stop modifications against the actual rules. These observations can identify a deviation without diagnosing cockiness from a winning run.

Actions to compare with the written rules
Sizing upOne per cent of risk per trade quietly becomes two, three, eventually five — with no justification in the plan
Skipping the checklist"I already know this is a good setup" — the ritual that used to filter weak entries disappears
Dismissing the stop lossYou move it further out or remove it, because "this trade cannot fail"
Several positions at onceSeveral positions can share exposure; calculate joint amounts rather than assuming an exact doubling
Brushing off losing runsThree losing trades become "bad luck" rather than a signal worth checking

Publicly discussing a result does not establish cockiness, and a calendar quarter is not a diagnostic boundary. Keep a complete record rather than judging skill from praise or a recent streak. The article on managing the trader's ego considers that distinction without treating publicity as a universal cause.

A streak and a probability model are different

Assume a hypothetical independent model with constant win probability 55%. Five specified consecutive trades all win with probability 0.55⁵ = 0.0503284375, about 5.03%. In 100 trades there are 96 overlapping five-trade windows; their expected number of all-win windows is 96 × 0.55⁵ ≈ 4.83. That is an expectation over repetitions, not a promise of several distinct winning runs in any one year. Adjacent windows overlap. A streak alone does not prove a skill change—one of the psychological traps to review.

A recent winning run does not make a loss due under an independent unchanged model. An unplanned size increase can nevertheless make a later loss much larger. Barber and Odean found poorer net returns among the most active US common-stock households in their 1991–1996 sample; this observational result does not establish the causal confidence mechanism of an individual trader. See also the overconfidence bias.

Decide whether you count parts or whole ideas

A high percentage of green rows can be correct arithmetic for the wrong unit. Check the grouping before calling it the success rate of your decisions.

Scroll sideways to see the full diagram

Two units for assessing win rateThree parts of idea A each earn ten zlotys; idea B loses forty. Three of four parts are positive, but only one of two whole ideas is positive.PARTS VERSUS WHOLE IDEASIdeaA: three profitable parts+PLN 10+PLN 10+PLN 10IdeaB: one losing part−PLN 40Parts: 3 of 4 positive → 75%.Whole ideas: 1 of 2 positive → 50%.Net result: 30 − 40 = −PLN 10.

Two hypothetical completed ideas. A has three parts at+10 PLN net each; B has one part at−40 PLN net. There are no other outcomes and costs are included. GroupingA/B was fixed before results and retained afterwards. The MQL5 Deal Properties documentation identifies a separate position ID linking deals associated with that position's lifetime.

Grouping was fixed before outcomes. No row was removed.

Both percentages are calculated correctly but answer different questions. Three positive parts out of four give75%; one positive idea out of two gives50%. The sum remains−10 PLN. Report the unit and net outcome together: splitting one idea into more exits does not by itself create independent decisions or an edge.

How to build confidence that survives a losing run

Compare hit rate with payoff and costs. In two hypothetical sets of 100 completed trades, A has 55 gains of PLN 100 and 45 losses of PLN 150: 5,500 − 6,750 = PLN −1,250. B has 40 gains of PLN 200 and 60 losses of PLN 100: 8,000 − 6,000 = PLN +2,000. These are invented pre-cost calculations, not market results. A positive sample total does not establish the next trade's probability.

Use the journal to estimate win rate, average payoff and net expectancy, and separately count decisions that met the plan. A losing week can show good adherence while still requiring a review of costs and strategy performance. Keep both records, including the exposure and account checks described in process and outcome review.

What to do tonight

A one-to-ten confidence score can be a personal note alongside the decision record. An 8/10 feeling is not automatically an 80% probability forecast. To assess a forecast, specify the event, horizon and price source before seeing the result.

One hypothetical path, three forecasts
Continuous linear segments within a fixed window1.1000 → 1.1080 → 1.0980
Touches 1.1050 at least onceYes: the continuous rise passes that level
Ends above 1.1000No: it ends at 1.0980
Stays above 1.0990 throughoutNo: the minimum is 1.0980

Do not replace the ending-price forecast with a touch forecast after seeing the path. A model touch is not confirmation of a real order fill. This exercise assesses the stated event, not future profitability.

After a few weeks, compare confidence ratings with rule adherence and net results. Keep all observations and check alternative explanations. A small personal sample can suggest a question to investigate; it cannot establish a general relationship between confidence and trading performance.