Discipline is easier when the chart has less authority.
A vague instruction such as “be patient” leaves the hardest decisions until the moment urgency is highest. A usable process does the opposite. It names the setup, chart conditions, risk, order, exit and stopping conditions in advance. The live task becomes a check: did the written condition happen, or not?
- 01
Does the setup match the rules?
- 02
Are risk and invalidation written?
- 03
Did the required bar close?
- 04
Act once, then record.
- 01
Price moves quickly.
- 02
Urgency becomes the signal.
- 03
Entry arrives before the rule.
- 04
The story is written afterward.
The purpose of a plan is not to predict the next candle. It is to make the choice auditable before the outcome can rewrite the story.
If-then planning has been studied across many kinds of goal pursuit. The evidence is not a forex-profit study, but it supports a useful design principle: connect a specific situation to a specific action before it occurs. “If a required field is blank, then I do not place the demo order” is stronger than “I will try to be careful.” [8]
Turn a bias label into something you can actually record.
Knowing a bias name does not make anyone immune. The useful move is to translate it into observable behaviour: entered before the close, increased size after a win, widened a stop, added a trade after the session cap, or changed a filter mid-sample.
| What to watch | What it can look like | Process control |
|---|---|---|
| Outcome bias | A win makes a weak decision look clever; a loss makes a sound decision look foolish. | Score the setup, risk and execution before revealing net R. |
| Recency | The last few trades start to feel like the whole distribution. | Review the whole unchanged sample and its size, not just this week. |
| Loss aversion / disposition | A stop is widened or a loser is held because closing makes the loss final. | Place analytical invalidation and maximum risk in the plan; changes are deviations. |
| Overconfidence | A winning run increases size, frequency or rule flexibility without a sound reason. | Keep size and entry limits fixed for the version under test. |
| Revenge trading | A loss creates an urge to recover money through the next order. | Use a prewritten session stop and close the platform when it triggers. |
| FOMO | Speed, hype or a sharp candle becomes a reason to bypass the setup. | Require the same completed-candle check; record a missed move as a no-trade. |
| Rule drift | Filters, exits or definitions change while the sample is still being collected. | Label the rule card version. Test a change from the next trade forward as a new version. |
| Fatigue / noisy environment | Attention slips while alerts, extra charts or tiredness add more decisions. | Use a readiness gate, one workspace and an end time; switch to review if the gate fails. |
Prospect theory describes reference-dependent choices under risk. Odean later found a preference for realising winners rather than losers in a large sample of stock-brokerage accounts. Those were not spot-FX accounts, so the result is a warning about a documented investor pattern, not proof about every trader. [2][3][12]
Barber and Odean found that the most active households in their stock-account sample performed worse after costs, with overconfidence offered as an explanation for high turnover. The safe lesson for a forex process is modest: a winning run is not new permission to add frequency or size. [4]
IPO evidence found that personally experienced returns were strongly related to later subscriptions. A separate investor study linked positively distorted memories of past returns with overconfidence and trading frequency. Keep the full log visible so the latest run cannot become the whole history.[6][7]
FCA guidance flags urgency and hype as reasons to slow down. Its experiment with more than 9,000 consumers found that some app engagement features changed trading frequency and risk taking. That does not predict one person's next action, but it is a good reason to mute prompts that are not part of the trading plan. [10][11]
A good decision can lose. A poor decision can win.
In classic experiments, people rated decision quality more favourably when they knew the outcome was favourable—even when they had the same information that was available at decision time. That is outcome bias. In trading, it tempts a learner to copy a lucky breach and abandon a compliant setup after one loss.[1]
Keep it. Do not increase size because one outcome felt easy.
Keep it. It belongs in the strategy sample.
Tag it. Profit does not turn an early entry into a valid one.
Tag it. Fix the decision environment before adding risk.
- Setup matched the written rules before entry
- Trigger confirmed as written
- Risk and size inside the cap
- Order and exit rules followed
- Record complete, including costs
- Gross R
- Spread, commission and financing
- Slippage against the planned fill
- Net R
- Maximum adverse / favourable excursion
Do not combine these into one grade. “+1.3R, process 2/5” is much more informative than “great trade.” Likewise, “−1.0R, process 5/5” lets the loss remain a valid part of the strategy record.
Write rules that remove decisions, not rules that promise control.
Prewritten rules do not control price, fills or gaps. They control what you are willing to do. Use firm boundaries for exposure and process; leave uncertain outcomes uncertain.
Named pairs, timeframe, session, setup version and data source.
A measurable close, cross, rejection or other condition written in advance.
Invalidation, size, cash/R risk, open heat and cost allowance.
No new order, platform closed, record saved and review delayed.
A note for version two, never a silent change to the current trade.
Evidence, readiness, costs or platform conditions fail the gate.
Bounded risk matters because leveraged OTC forex can produce rapid losses, while orders may not execute at the intended price. The CFTC advises prospective customers to understand margin, account agreements and dealer disclosures. A self-imposed loss cap should sit inside—not replace—those operational facts.[13]
The routine should make the next action obvious.
- Normal concentration for this task
- No rush to recover or prove anything
- Session end time is visible
- Session and timezone checked
- Scheduled news checked
- Spread and feed look usable
- Cash or R cap written
- Maximum new entries written
- Open and correlated risk counted
- Promotional alerts muted
- Correct account and order size
- Checklist and journal already open
A “no” does not need to become a debate. It can simply make this a review session rather than a trading session.
- Open the rule card and journal before the chart.
- Check timezone, session, scheduled events and spread.
- Write session cash/R cap, entry cap and end time.
- Add existing and correlated exposure.
- Complete the readiness gate; a failed gate means review only.
- Wait for every eligibility field and the confirmed trigger.
- Calculate size from invalidation and the fixed loss budget.
- Read the ticket back: symbol, side, size, order and stop.
- Do not widen risk or add an unplanned position.
- If a pause trigger fires, cancel pending entries and finish.
- Save planned and actual prices, costs and screenshots.
- Score the five process fields before calculating net R.
- Tag observable deviations; leave personality out of it.
- Record no-trades and missed moves as valid decisions.
- Leave strategy changes for the scheduled review.
Short-term total sleep deprivation is associated with poorer performance across several cognitive domains in a meta-analysis; it does not provide a personal pass/fail threshold for trading. Use your own readiness gate conservatively: when concentration is unreliable, do review work rather than adding market risk.[9]
A losing run is information—not an invoice the next trade must pay.
“Revenge trading” is an informal name for taking new or larger risk in an attempt to recover a loss. In a study of professional Chicago Board of Trade traders, traders with morning losses took above-average afternoon risk while trying to recover. That market and population differ from retail FX, but the evidence is a strong reason to prewrite a stop rather than improvise one while behind.[5]
- BEFOREWrite the cap
Set a session cash or R limit, maximum entries and non-loss pause triggers.
- TRIGGERStop the session
The cap, one serious rule breach or unreliable conditions ends new orders.
- SEPARATELeave the chart
Close the platform. Do not use a smaller instant trade as a substitute for stopping.
- AUDITCheck the records
Eligibility, rule version, fills, costs and deviations come before P/L.
- RESTARTUse a written condition
Resume on demo only when the next planned session and setup criteria are both present.
- Session cash or R cap reached
- Maximum number of new entries reached
- One order creates exposure beyond the written book cap
- Platform, feed or order state cannot be verified
- An unplanned order is placed
- A stop is widened beyond planned invalidation
- Size is increased to recover a loss
- The rule card is being debated rather than followed
- Next planned session—not an immediate retry
- Journal and screenshots complete
- Risk cap reset according to the written plan
- Original strategy version still unchanged
The losing trade has the better process.
Both examples use invented prices on historical-style replay. They are not current market analysis or trade suggestions.
Compliant EUR/USD replay loss
- Plan
- Version 1.3 pullback setup; London demo window; H1 close required.
- Before reveal
- Entry 1.08420, invalidation 1.08270, target 1.08720, planned risk £20.
- Execution
- All fields complete; 15-pip stop; size rounded down; no changes after entry.
- Outcome
- Stop filled with spread and slippage. Net result −1.08R.
- Deviation tags
- None.
- Review
- Keep in the valid sample. Do not change the setup from one compliant loss.
Profitable GBP/JPY replay breach
- Plan
- Version 2.0 breakout; closed M30 candle above the range required.
- Before reveal
- No valid entry yet: the candle had not closed and size was blank.
- Execution
- Entered early during a fast candle, then widened the stop without recalculating risk.
- Outcome
- Exited later for +1.35R using the amount originally risked.
- Deviation tags
- Early entry; missing size check; stop changed.
- Review
- Exclude from the strategy sample, retain in the deviation log, and add an order-ticket gate.
Memory is not a neutral database. Research on investors found positively biased recall of past performance and tested showing participants their actual returns as a way to reduce overconfidence and trading frequency. A journal is useful when it preserves what was written at decision time, not when it becomes a diary written from memory at the weekend. [7]
OPEN THE PRIVATE, DEVICE-LOCAL JOURNALRecord the plan before the result can rewrite it→A weekly review needs enough trades to interpret—and one next action.
Wins, losses, no-trades, screenshots, costs and missing fields.
Setup matched / did not match, plan followed / broken, complete / incomplete.
Net R distribution, costs, time of day and repeated deviation tags.
Keep the rules unchanged, pause, improve the environment or start a clearly labelled new version.
How many setups matched, became no-trades or broke a rule? Which records or costs are missing?
Adherence rate by field. Which observable deviation repeated, and in which environment?
Net R distribution, average cost, drawdown and sequence—reported separately for valid and invalid records.
Keep the version unchanged, pause it, repair one process check, or define a separately tested version.
Practise the gates before they protect real money.
- 01Write one setup
Use the plan builder to define when it applies, the trigger, invalidation, size, exit and pause rules.
- 02Take the risk quiz
Check that margin, position size, drawdown and correlation do not get mixed together.
- 03Run 20 blind replays
Use the simulator with no strategy changes during the sample. Record valid no-trades too.
- 04Journal at decision time
Save the plan before reveal; score process before outcome; use neutral deviation tags.
- 05Review once
Choose one process fix or keep the version unchanged. A new strategy idea starts a new sample.