Statistics, journal and the improvement loop

Without a journal you do not have a system, you have a collection of memories. Day trading and scalping generate hundreds of decisions per week – only a journal with numbers lets you separate edge from noise and tilt. This lesson closes the course with a measurement framework.

Lesson goal

• Build a day trader’s journal that counts itself.
• Define the metrics that decide whether a system lives or dies.
• Introduce the weekly loop: review -> hypothesis -> test -> implementation.
• Write down stop trading rules based on statistics.

What the journal must measure

Minimum record of 1 trade:

Date | Time | Instrument | Setup (ORB/VWAP/Orderflow/KZ) | KZ | Long/Short
Gap_ATR | VWAP_slope | OR_N_ATR | Vol_ratio
Entry | Stop | Target | Risk_R | Size | Slippage_R
Exit | Result_R | Result_PLN | Hold_time_min
Plan compliance (0/1) | Emotion 1-5 | Error | Video_link

Key metrics – formulas

WinRate = wins / all
AvgWin_R = average R of the winners
AvgLoss_R = average R of the losers (positive)
Expectancy_R = WinRate*AvgWin_R - (1-WinRate)*AvgLoss_R
Profit Factor = sum(Profit_R) / sum(Loss_R)
Payoff = AvgWin_R / AvgLoss_R

MaxDD_R = largest equity drop in R from the peak
Month_profit_R = sum of R in the month
% of plan = trades_on_plan / all
Cost_R = sum((commissions+slippage)/Risk)
Net_edge_R = Expectancy_R - cost_avg_R

Survival rule: Net_edge_R ≤0 after 50 trades – the system has no edge. Stop, redesign.

The tables you must have every week

Table 1: By setup

Table 2: By KZ / hours

Conclusion: outside the KZ you do not trade.

Table 3: Errors

The improvement loop – weekly

Friday, 2h review:

1. Export the trades + video of the 3 worst and 3 best.
2. Calculate Tables 1-3.
3. Pick 1 hypothesis: “If I do not trade outside the NY KZ, edge will grow by 0.1R”.
4. Define the change: “Outside 14:30-17:00 size=0”.
5. Forward test 20 trades the next week with one change only.

Changing more than 1 element at a time is forbidden.

Hard rules – statistics say stop

Stop of the day: DailyLoss = -3R or -2% of the account – end of day
Stop of the week: WeeklyLoss = -6R – 2 days off, review only
Stop of the system: 20 trades with % of plan <70% – you go back to demo
Cost stop: Slippage_avg >0.25R – cut size by 50%
Concentration stop: 2 "entered without a signal" errors in 1 day – end of day

Emotion journal

After every KZ block rate 1-5: Focus, FOMO, Fatigue. If <3 – a 20 min break.

Journal template – ready for Excel / Notion

Columns: Date, Setup, KZ, Gap_ATR, VWAP_slope, Risk_R, Result_R, Slippage_R, Time, Compliance, Error, Video_link, Note.

Dashboard formulas:

=WinRate: =COUNTIF(Result_R,">0")/COUNTA(Result_R)
=Expectancy: =AVERAGEIF(Result_R,">0")*WinRate + AVERAGEIF(Result_R,"<0")*(1-WinRate)
=ProfitFactor: =ABS(SUMIF(Result_R,">0")/SUMIF(Result_R,"<0"))
=%of plan: =AVERAGE(Compliance)

Example calculation on a real session

30 trades:

• Win 14, Loss 16 -> WR 46.7%
• AvgWin 1.7R, AvgLoss 1.0R
• Expectancy = 0.4671.7 - 0.5331.0 = 0.794 - 0.533 = 0.261R
• PF = (141.7)/(161.0)=23.8/16=1.49
• Average cost 0.12R/trade -> Net_edge = 0.141R
• 30 trades per month -> 4.23R net

Is that good? Yes, if % of plan >85% and MaxDD <5R. If % of plan is 60% – fix execution first.

How to tell a bad system from bad execution

Bad system: % of plan 90%, negative expectancy after 50 trades -> change the setup/filters
Bad execution: % of plan 60%, negative expectancy, but it would be positive if you had held the plan -> do not change the system, fix hotkeys, KZ, limits
Mixed: % of plan 80%, slightly positive expectancy but costs >30% of profit -> cut costs, change the order type

Weekly dashboard – example of a decision formula

IF Net_edge_R <0 AND trades >30 -> stop the system, redesign
IF % of plan <75% -> 1 week on demo, work on error #1
IF MaxDD_R > 2 * average monthly profit in R -> cut size by 30%
IF trades outside the KZ have a negative edge -> size outside the KZ = 0 for the next week

This turns emotions into rules.

Archiving – why it pays off

Store every week: a CSV export of trades, KZ video, a dashboard screenshot, 1 page of conclusions. After 3 months you have 12 pages of a playbook based on your own data. That is more than most paid courses give you.

The last rule

The journal is not a punishment. It is proof that you respect your time and capital. If you do not want to fill it in – you should not trade that day.

When the plan says DO NOT TRADE – based on the journal

• You did not complete yesterday’s review.
• The last 20 trades % of plan <70%.
• Costs >30% of gross profit.
• MaxDD_R exceeded by 50%.
• No recordings from 3 sessions.

The mistakes that cost the most

1. A journal without numbers: a descriptive “I entered badly” with no R, slippage or time.
2. Measuring P&L instead of R: a 1000 PLN loss at 1000 PLN of risk is not the same as at 100 PLN of risk.
3. Changing 5 things at once: a new setup + a new KZ + a new size.
4. Reviewing losses only – the winners have mistakes too.
5. No data limit – a 200-column journal that you never fill in.

Measurable exercise (Weekend, 2h)

1. Take the last 20 trades and copy them into the template.
2. Calculate WinRate, AvgWin, AvgLoss, Expectancy_R, PF, % of plan, Cost_R.
3. Build Tables 1-3.
4. Pick the 1 error that cost the most R and write 1 blocking rule.
5. Plan 20 forward trades with that one rule.

Check yourself

1. Why do we measure the result in R instead of PLN?
2. You have 30 trades: 12 wins avg 1.6R, 18 losses avg 1.0R. Calculate the expectancy and the PF.
3. What does Net_edge_R ≤0 after 50 trades mean?
4. Why is % of plan execution more important than P&L in the first 100 trades?

Frequently asked questions

How many trades do you need to assess a system? Min 30-50 as a first look, 100 for stability.

Do I have to record video? Yes, at least the KZ. Memory lies.

Excel or a dedicated journal? Start with Excel so that you understand the formulas. Then TraderVue, Edgewonk.

What if I only have 10 trades / month? Combine months, keep the market conditions.

---

The journal is your knowledge P&L account. Without measurement there is no learning, only repeating the same mistakes.

---

End of course 18

You have completed 8 day trading and scalping lessons: from matching a style, through preparing the day, ORB, VWAP, orderflow, killzones and tools, to statistics. Every module needs its own forward test of 30-50 trades on a simulation before you combine them into a playbook.

Closing rule: you only trade what you have measured and recorded, and what has a positive Net_edge_R after costs. The rest is observation.

This is not investment advice. Every strategy needs testing on your markets, costs and broker.

---