Most of my early journals were P&L screenshots with a sentence of story underneath. They felt productive and changed almost nothing. What started changing behavior was scoring the process, then reviewing those scores weekly as patterns — not as a mood board of green and red days.
I still record outcomes. I just refuse to let outcome be the main grade. A clean loss that followed the plan scores high. A green trade that broke session rules, news filters, or size limits scores low. The journal’s job is to reinforce the behavior I want to repeat under stress, not to decorate a win rate.
What I score per trade (simple 1–5 or yes/no — keep it fast so I actually do it):
1. Setup quality — Did H1 bias, M15 location, and M1/M5 trigger align? Was this A / B / C, and did I only take A/B as planned?
2. Rule adherence — Session window, max spread, news blackout, risk %, daily stop. Any breach is a hard mark down even if the trade paid.
3. Emotional state — Calm / rushed / revenge / euphoric, noted in one word before or right after the fill. I am not writing a novel; I am tagging state so patterns show up later.
4. Execution notes — Intended entry vs fill, stop placement vs plan, whether I moved risk mid-trade without a rule.
End of day I add two lines only: total process score average, and whether I hit a circuit breaker. That is enough.
Weekly review (fixed slot, not “when I feel like it”):
- Sort or filter for low setup-quality scores that I still took. Those are FOMO or boredom trades.
- Count rule breaches separately from losing trades. Breaches are the priority fix; losses inside the rules are tuition.
- Look at emotional tags clustered around certain hours or after a first loss. If “revenge” appears after loss #1 in London, the fix is a hard pause rule — not a new indicator.
- Compare cold weeks: did I cut size as planned, or did I size up to “get it back”? Process scores expose that faster than equity curves.
What I stopped doing:
- Pasting equity screenshots as proof of progress.
- Writing long narratives that excuse a breach.
- Reviewing only winning trades. Winners that broke rules are more dangerous than honest losers.
A concrete rule that came from journaling, not from a course: after two full planned losses, the third idea that day requires an explicit A+ checklist tick or I flatten the platform. That rule exists because the journal kept showing the same emotional tag on trade three.
If your journal does not change next week’s behavior, it is a scrapbook. Score setup quality, rule adherence, and state. Review weekly for patterns. Let P&L be a consequence you record — not the grade you optimize in the moment. Process scores are slower to brag about. They are what keep a scalper solvent when gold is loud and the session feels urgent.
Advertisement
Journaling that actually changes behavior: process scores over P&L screenshots
Recommended broker for automated trading & scalping
-
LondonScalper
- Posts: 203
- Joined: Sat Sep 05, 2026 7:54 am
Re: Journaling that actually changes behavior: process scores over P&L screenshots
Process scores > P&L screenshots. Same evolution here.
I score each session 0–2 on a few rails only:
• Did I respect the pre-session risk sheet?
• Did I take only checklist setups (or note deviations)?
• News filter followed?
• Size rule followed after winners/losers?
Weekly review is just: which rail failed most often? Fix that one. Not “be more disciplined” as a vibe.
One trick: I hide P&L on the platform during the session when I’m rebuilding habits. The journal still gets the number after — but the scorecard gets filled first.
What’s your scoring scale — binary pass/fail per rule, or a 1–5 overall?
I score each session 0–2 on a few rails only:
• Did I respect the pre-session risk sheet?
• Did I take only checklist setups (or note deviations)?
• News filter followed?
• Size rule followed after winners/losers?
Weekly review is just: which rail failed most often? Fix that one. Not “be more disciplined” as a vibe.
One trick: I hide P&L on the platform during the session when I’m rebuilding habits. The journal still gets the number after — but the scorecard gets filled first.
What’s your scoring scale — binary pass/fail per rule, or a 1–5 overall?
Re: Journaling that actually changes behavior: process scores over P&L screenshots
How to Backtest a Scalping Strategy Properly
Most backtests scalpers run are, unintentionally, testing whether they can recognize their own strategy with hindsight — not whether the strategy actually works. Here's how to close that gap.
The Core Problem: Hindsight Bias
Scrolling back through a chart and marking "yes, this order block worked" is fundamentally different from identifying that order block in real time, before you know what happens next. Your eyes are drawn to the clean, obvious setups that worked and skip right past the ambiguous ones that would've been genuinely hard calls live. This alone can make a mediocre strategy look excellent on a casual backtest.
Bar-by-Bar Replay Is Non-Negotiable
Use a replay tool (TradingView's bar replay or equivalent) that reveals price one candle at a time, exactly as it would have unfolded live. Force yourself to mark your entries, stops, and targets based only on what's visible at that point — no peeking ahead. This is slower and more tedious than scanning a historical chart, and that tedium is the entire point; it's the only way to get an honest read on how the strategy performs under real uncertainty.
Track More Than Win Rate
Win rate alone tells you almost nothing about whether a strategy is worth trading. Track average risk-to-reward per trade, maximum consecutive losses (so you know what a realistic losing streak looks like and can size accordingly), and — critically — realistic spread/slippage costs applied to every single trade, not just the theoretical winners.
Sample Size Matters More Than Traders Want It To
Twenty backtested trades tells you almost nothing statistically. A strategy needs at least 100+ instances, ideally across different market conditions (trending, ranging, high and low volatility periods) before you can draw any real conclusion about its edge. If you're not willing to put in the reps to get a real sample, you're not backtesting — you're just confirming a hunch.
Forward Test Before Going Live With Size
Once a backtest looks genuinely promising, run it forward on a demo account (or very small live size) for a meaningful stretch before trusting it with real risk. This catches the gap between "I can identify this pattern with unlimited time to think" and "I can execute this pattern under live time pressure with my own money on the line" — which is a different skill entirely, and one backtesting alone can't fully prepare you for.
Most backtests scalpers run are, unintentionally, testing whether they can recognize their own strategy with hindsight — not whether the strategy actually works. Here's how to close that gap.
The Core Problem: Hindsight Bias
Scrolling back through a chart and marking "yes, this order block worked" is fundamentally different from identifying that order block in real time, before you know what happens next. Your eyes are drawn to the clean, obvious setups that worked and skip right past the ambiguous ones that would've been genuinely hard calls live. This alone can make a mediocre strategy look excellent on a casual backtest.
Bar-by-Bar Replay Is Non-Negotiable
Use a replay tool (TradingView's bar replay or equivalent) that reveals price one candle at a time, exactly as it would have unfolded live. Force yourself to mark your entries, stops, and targets based only on what's visible at that point — no peeking ahead. This is slower and more tedious than scanning a historical chart, and that tedium is the entire point; it's the only way to get an honest read on how the strategy performs under real uncertainty.
Track More Than Win Rate
Win rate alone tells you almost nothing about whether a strategy is worth trading. Track average risk-to-reward per trade, maximum consecutive losses (so you know what a realistic losing streak looks like and can size accordingly), and — critically — realistic spread/slippage costs applied to every single trade, not just the theoretical winners.
Sample Size Matters More Than Traders Want It To
Twenty backtested trades tells you almost nothing statistically. A strategy needs at least 100+ instances, ideally across different market conditions (trending, ranging, high and low volatility periods) before you can draw any real conclusion about its edge. If you're not willing to put in the reps to get a real sample, you're not backtesting — you're just confirming a hunch.
Forward Test Before Going Live With Size
Once a backtest looks genuinely promising, run it forward on a demo account (or very small live size) for a meaningful stretch before trusting it with real risk. This catches the gap between "I can identify this pattern with unlimited time to think" and "I can execute this pattern under live time pressure with my own money on the line" — which is a different skill entirely, and one backtesting alone can't fully prepare you for.
It’s Fairman 
Re: Journaling that actually changes behavior: process scores over P&L screenshots
Journaling Your Scalps: What Data Actually Matters
Most trading journals die within two weeks because they're either too shallow to be useful (just win/loss and pips) or so exhaustive they become a chore nobody sustains. Here's a middle ground built specifically for scalpers, where volume of trades makes an overly detailed journal impractical.
The Fields That Actually Move the Needle
Setup type. Which specific pattern was this — London sweep, NY news reversal, Asian range fade? Without this tag, you can't ever answer "which of my setups actually makes money," which is arguably the single most useful question a journal can answer.
Confluence score. Using your own checklist (see the earlier confluence post), how many of your criteria did this trade actually meet? This lets you eventually correlate confluence count with win rate — often revealing that your 5/5 setups dramatically outperform your 3/5 "I'll allow it" entries, which is exactly the kind of evidence that makes discipline easier to maintain.
Planned vs. realized R:R. Not just whether you won or lost, but whether you actually executed the trade as planned. This is where you catch the gap between your strategy's real edge and your own execution discipline.
Session and time. Which window did this occur in? Over enough trades, this reveals which specific hours are actually producing your results versus which ones you're trading out of habit with no real edge.
One-line emotional note. Were you calm, rushed, tilted from a previous loss, bored? This single subjective line, tracked honestly over time, tends to reveal patterns traders are otherwise blind to — a cluster of losses that all happened to occur after you noted "frustrated" or "chasing" is a pattern no chart-based metric alone will show you.
What to Leave Out
Don't screenshot and annotate every single trade in detail — at scalping frequency, this becomes unsustainable fast and is a big reason journals get abandoned. Save detailed chart annotation for your biggest wins, biggest losses, and anything that genuinely surprised you; log the rest with the quick structured fields above.
Reviewing It
A journal nobody reviews is just data entry. Set a weekly cadence — even fifteen minutes — to actually scan the week's entries for patterns: which setup type is carrying your results, which session is quietly costing you money, whether your emotional notes cluster around your worst trades. The value isn't in the logging itself, it's in what the aggregated data eventually tells you that gut feeling alone never would.
Most trading journals die within two weeks because they're either too shallow to be useful (just win/loss and pips) or so exhaustive they become a chore nobody sustains. Here's a middle ground built specifically for scalpers, where volume of trades makes an overly detailed journal impractical.
The Fields That Actually Move the Needle
Setup type. Which specific pattern was this — London sweep, NY news reversal, Asian range fade? Without this tag, you can't ever answer "which of my setups actually makes money," which is arguably the single most useful question a journal can answer.
Confluence score. Using your own checklist (see the earlier confluence post), how many of your criteria did this trade actually meet? This lets you eventually correlate confluence count with win rate — often revealing that your 5/5 setups dramatically outperform your 3/5 "I'll allow it" entries, which is exactly the kind of evidence that makes discipline easier to maintain.
Planned vs. realized R:R. Not just whether you won or lost, but whether you actually executed the trade as planned. This is where you catch the gap between your strategy's real edge and your own execution discipline.
Session and time. Which window did this occur in? Over enough trades, this reveals which specific hours are actually producing your results versus which ones you're trading out of habit with no real edge.
One-line emotional note. Were you calm, rushed, tilted from a previous loss, bored? This single subjective line, tracked honestly over time, tends to reveal patterns traders are otherwise blind to — a cluster of losses that all happened to occur after you noted "frustrated" or "chasing" is a pattern no chart-based metric alone will show you.
What to Leave Out
Don't screenshot and annotate every single trade in detail — at scalping frequency, this becomes unsustainable fast and is a big reason journals get abandoned. Save detailed chart annotation for your biggest wins, biggest losses, and anything that genuinely surprised you; log the rest with the quick structured fields above.
Reviewing It
A journal nobody reviews is just data entry. Set a weekly cadence — even fifteen minutes — to actually scan the week's entries for patterns: which setup type is carrying your results, which session is quietly costing you money, whether your emotional notes cluster around your worst trades. The value isn't in the logging itself, it's in what the aggregated data eventually tells you that gut feeling alone never would.
It’s Fairman
Re: Journaling that actually changes behavior: process scores over P&L screenshots
The Danger of Over-Optimizing Indicators
Every scalper eventually goes through a phase of tweaking indicator settings, trying to find the specific parameters that would have made their last several trades work perfectly in hindsight. This is one of the most seductive traps in retail trading, and it's worth understanding exactly why it doesn't actually produce what it promises.
What Over-Optimization Actually Is
Adjusting indicator parameters, backtest date ranges, or entry criteria specifically until they produce the best-looking results on a particular historical dataset — without any principled reason for the specific values chosen beyond "this made the backtest look better." The resulting settings are fit tightly to the noise and specific quirks of that particular historical sample, not to any genuine, repeatable market behavior.
Why This Is So Easy to Do Without Realizing It
Modern charting platforms make parameter adjustment trivially easy — a moving average period, an ATR multiplier, a Fibonacci level can all be tweaked in seconds, and each adjustment can be immediately checked against historical performance. This tight, fast feedback loop is exactly what makes over-optimization so tempting: it feels like genuine improvement, each small tweak producing a marginally better-looking backtest, when in reality it's often just increasingly precise fitting to noise that has no predictive value going forward.
The Telltale Signs You've Crossed the Line
Parameters that don't correspond to any principled reasoning ("I use a 23-period moving average" with no explanation beyond backtest performance) rather than reasonably round, defensible numbers. A strategy that performs dramatically worse on a slightly different date range or a different but similar pair than the one it was originally tuned on. Settings that required extensive, iterative fine-tuning to find, rather than working reasonably well across a range of nearby parameter values.
A Useful Test: The Robustness Check
A genuinely sound set of parameters should perform reasonably — not necessarily optimally, but reasonably — across a range of nearby values, not just the single exact combination you landed on. If a 20-period moving average works well but a 19- or 21-period version performs dramatically worse, that's a warning sign you've found a fit to noise rather than a genuine, stable edge. Similarly, testing the same settings out-of-sample, on a different date range or different pair than the one used to originally tune them, is one of the more reliable ways to catch over-optimization before it costs real money live.
The Underlying Point
More parameter tweaking isn't the same as more genuine edge. A simpler, less perfectly-fitted strategy that performs consistently across varied conditions is generally more trustworthy than an elaborately tuned one that only shines on the exact historical window it was built against.
Every scalper eventually goes through a phase of tweaking indicator settings, trying to find the specific parameters that would have made their last several trades work perfectly in hindsight. This is one of the most seductive traps in retail trading, and it's worth understanding exactly why it doesn't actually produce what it promises.
What Over-Optimization Actually Is
Adjusting indicator parameters, backtest date ranges, or entry criteria specifically until they produce the best-looking results on a particular historical dataset — without any principled reason for the specific values chosen beyond "this made the backtest look better." The resulting settings are fit tightly to the noise and specific quirks of that particular historical sample, not to any genuine, repeatable market behavior.
Why This Is So Easy to Do Without Realizing It
Modern charting platforms make parameter adjustment trivially easy — a moving average period, an ATR multiplier, a Fibonacci level can all be tweaked in seconds, and each adjustment can be immediately checked against historical performance. This tight, fast feedback loop is exactly what makes over-optimization so tempting: it feels like genuine improvement, each small tweak producing a marginally better-looking backtest, when in reality it's often just increasingly precise fitting to noise that has no predictive value going forward.
The Telltale Signs You've Crossed the Line
Parameters that don't correspond to any principled reasoning ("I use a 23-period moving average" with no explanation beyond backtest performance) rather than reasonably round, defensible numbers. A strategy that performs dramatically worse on a slightly different date range or a different but similar pair than the one it was originally tuned on. Settings that required extensive, iterative fine-tuning to find, rather than working reasonably well across a range of nearby parameter values.
A Useful Test: The Robustness Check
A genuinely sound set of parameters should perform reasonably — not necessarily optimally, but reasonably — across a range of nearby values, not just the single exact combination you landed on. If a 20-period moving average works well but a 19- or 21-period version performs dramatically worse, that's a warning sign you've found a fit to noise rather than a genuine, stable edge. Similarly, testing the same settings out-of-sample, on a different date range or different pair than the one used to originally tune them, is one of the more reliable ways to catch over-optimization before it costs real money live.
The Underlying Point
More parameter tweaking isn't the same as more genuine edge. A simpler, less perfectly-fitted strategy that performs consistently across varied conditions is generally more trustworthy than an elaborately tuned one that only shines on the exact historical window it was built against.
It’s Fairman