Trading Journal Performance Analysis and Evaluation

Good to see you again. In the previous lesson, you used Bar Replay to execute a deliberately fixed practice rule and built a journal containing entries, exits, planned risk, actual fills, costs, and adherence notes. That journal is now your dataset.

This lesson completes the risk-controlled paper-trading module. You will calculate the core performance measures from net trade results, construct a closed-trade equity curve and maximum drawdown, and learn why a promising-looking set of trades can still be weak evidence. The purpose is not to “prove” a strategy after a handful of replays; it is to make your evaluation reproducible and resistant to self-deception.

This is educational analysis, not investment advice.

---

Start with a clean definition of a trade result

Before calculating any statistic, choose one source of truth: net P/L in one account currency, after all known costs. For each closed trade:

$$\text{Net P/L}=\text{gross trading P/L}-\text{commissions}-\text{fees}-\text{slippage allowance}-\text{conversion costs, if applicable}$$

A few practical rules keep the data meaningful:

• If your simulated fill already reflects buying at the ask and selling at the bid, do not subtract a separate spread estimate a second time.
• If TradingView Replay uses idealized fills, record a consistent slippage assumption yourself. A liquid US large-cap share, a UK share, and a major cryptocurrency should not automatically receive the same assumption.
• Do not combine GBP and USD P/L without converting one into the other using a documented convention.
• Keep each strategy version separate. “Daily Breakout v1” should not be mixed with a later version that adds trailing stops or RSI filters.

For comparison across trades with different position sizes, retain the realized $R$-multiple from the prior lesson:

$$R_{\text{realized}}=\frac{\text{net P/L}}{\text{planned account risk}}$$

A net result of USD 180 on a planned USD 100 risk is $+1.8R$. A net result of negative USD 105 is $-1.05R$. This makes a GBP-priced stock trade and a USD-priced stock trade comparable once each has been expressed in $R$, provided they belong to the same rule set.

The short opening and expectancy sections of this video reinforce an important point: winning frequently is not the same as having an edge.

---

Calculate win rate, average win, average loss, and expectancy

Suppose your replay journal contains these ten net results. The fixed planned risk is USD 100 per trade, solely to make the example easy to inspect.

Trade	Net P/L	Realized $R$	Classification
1	USD 180	$+1.80R$	Win
2	negative USD 105	$-1.05R$	Loss
3	USD 190	$+1.90R$	Win
4	negative USD 102	$-1.02R$	Loss
5	USD 0	$0R$	Flat
6	USD 170	$+1.70R$	Win
7	negative USD 104	$-1.04R$	Loss
8	negative USD 103	$-1.03R$	Loss
9	USD 210	$+2.10R$	Win
10	negative USD 106	$-1.06R$	Loss

This sample has four winners, five losers, and one flat trade.

Win rate

Define the win rate as profitable closed trades divided by all closed trades:

$$p_{\text{win}}=\frac{N_{\text{win}}}{N}$$

For the example:

$$p_{\text{win}}=\frac{4}{10}=40\%$$

The loss rate is:

$$p_{\text{loss}}=\frac{N_{\text{loss}}}{N}=\frac{5}{10}=50\%$$

The remaining $10\%$ is the flat trade. Some platforms exclude break-even trades from their denominator; that is acceptable only if you define and apply the convention consistently. Counting flats in the total gives the clearest “what happened per executed trade” view.

A $40\%$ win rate is neither good nor bad in isolation. A trend-following system may win relatively rarely but capture larger moves. A short-term mean-reversion method may win often while suffering occasional larger losses. The next two values tell you which situation you have.

Average win and average loss

Calculate average win from positive net P/L only:

$$\bar{W}=\frac{\text{sum of winning net P/L}}{N_{\text{win}}}$$ $$\bar{W}=\frac{180+190+170+210}{4}=\text{USD }187.50$$

Calculate average loss as a positive magnitude, again using net results:

$$\bar{L}=\frac{\left|\text{sum of losing net P/L}\right|}{N_{\text{loss}}}$$ $$\bar{L}=\frac{105+102+104+103+106}{5}=\text{USD }104$$

In $R$-terms, the average win is $+1.875R$, while the average loss is $-1.04R$. The loss is slightly worse than the intended $-1R$, which is plausible once costs or imperfect fills are included.

Expectancy: combine frequency and payoff

Expectancy estimates the average net outcome per trade over a long sequence of trades:

$$E=p_{\text{win}}\bar{W}-p_{\text{loss}}\bar{L}$$

For the example in $R$-terms:

$$E_R=(0.40)(1.875R)-(0.50)(1.04R)=+0.23R$$

Equivalently, the ten trades produced a total net gain of USD 230:

$$E_{\text{money}}=\frac{\text{total net P/L}}{N}=\frac{\text{USD }230}{10}=\text{USD }23$$

With USD 100 planned risk per trade, USD 23 is $+0.23R$. Both calculations must agree when the same trade set and accounting convention are used.

There is a useful diagnostic relationship here. With the same average win and loss, the approximate break-even win rate is:

$$p_{\text{break even}}=\frac{\bar{L}}{\bar{W}+\bar{L}}$$

For average win $1.875R$ and average loss $1.04R$, break-even is about $35.7\%$. The observed $40\%$ is above that threshold, producing positive expectancy in this small sample.

But treat that as a description of the sample, not proof of future profitability. Ten trades can easily produce a flattering or discouraging result by chance.

---

Build the calculations in a spreadsheet

A lightweight spreadsheet is sufficient. With your experience in front-end development, it may be tempting to automate immediately; resist that for the first few replay batches. Manually checking the calculations once makes later automation safer.

Use columns such as:

Column	Field	Notes
A	Trade ID	Sequential number
B	Net P/L	One account currency; costs included
C	Planned account risk	The original risk, including allowance convention
D	Realized $R$	Net P/L divided by planned risk
E	Equity after trade	Starting balance plus cumulative net P/L
F	Running equity peak	Highest equity observed so far
G	Drawdown	Current equity relative to running peak
H	Rule adherence	Followed, Minor issue, or Violation
I	Exit reason	Stop, target, time exit, or another prewritten reason

If net P/L is in column B and rows 2 through 101 contain trades, these formulas provide a basic summary:

Total trades:       =COUNT(B2:B101)
Winning trades:     =COUNTIF(B2:B101,">0")
Losing trades:      =COUNTIF(B2:B101,"<0")
Win rate:           =COUNTIF(B2:B101,">0")/COUNT(B2:B101)
Loss rate:          =COUNTIF(B2:B101,"<0")/COUNT(B2:B101)
Average win:        =AVERAGEIF(B2:B101,">0",B2:B101)
Average loss:       =-AVERAGEIF(B2:B101,"<0",B2:B101)
Expectancy:         =SUM(B2:B101)/COUNT(B2:B101)
Average R:          =AVERAGE(D2:D101)

The direct expectancy formula is particularly robust:

=SUM(B2:B101)/COUNT(B2:B101)

It automatically includes flat trades and does not require you to reconstruct expectancy from rounded win rates and averages. Use the probability formula as a cross-check and as a way to understand why the result is positive or negative.

---

Maximum drawdown measures the painful part of the path

Net profit answers, “Where did the account end?” Maximum drawdown answers a different question: “How far did the account fall from a prior high before recovering or the sample ended?”

For a closed-trade equity curve, first compute the running peak:

$$P_t=\underset{i\le t}{\max }(E_i)$$

Then calculate each trade’s drawdown:

$$D_t=\frac{E_t-P_t}{P_t}$$

Maximum drawdown is the most negative value in that drawdown series:

$$\mathrm{MDD}=\underset{t}{\min }(D_t)$$

Using the ten-trade example and an initial account value of USD 10,000:

Point	Equity after trade	Running peak	Drawdown
Start	USD 10,000	USD 10,000	$0.00\%$
Trade 1	USD 10,180	USD 10,180	$0.00\%$
Trade 2	USD 10,075	USD 10,180	$-1.03\%$
Trade 3	USD 10,265	USD 10,265	$0.00\%$
Trade 4	USD 10,163	USD 10,265	$-0.99\%$
Trade 5	USD 10,163	USD 10,265	$-0.99\%$
Trade 6	USD 10,333	USD 10,333	$0.00\%$
Trade 7	USD 10,229	USD 10,333	$-1.01\%$
Trade 8	USD 10,126	USD 10,333	$-2.00\%$
Trade 9	USD 10,336	USD 10,336	$0.00\%$
Trade 10	USD 10,230	USD 10,336	$-1.03\%$

The maximum drawdown is:

$$\frac{10,126-10,333}{10,333}=-2.00\%$$

You may describe this as “a maximum drawdown of $-2.00\%$” or “a drawdown magnitude of $2.00\%$.” Just do not alternate between the signed and unsigned conventions in the same journal.

Maximum drawdown is not:

• the largest individual losing trade;
• the number of losses in a row;
• the strategy’s future worst possible loss;
• a full description of risk.

It is path-dependent. The same ten trade results in a different order can have the same expectancy but a very different drawdown. This matters psychologically and financially: a method with positive expectancy may still be impossible to execute if its normal losing sequences exceed your risk tolerance.

For this course, use closed-trade equity: the curve changes only when a trade is exited. A mark-to-market curve that includes unrealized P/L can show a deeper intratrade drawdown. Both conventions can be valid, but label the convention and do not compare them as if they were identical.

To implement this in the spreadsheet, place starting account equity in cell K1. Assuming net P/L is in column B, equity is in column E, running peak is in column F, and drawdown is in column G:

E2: = $K$1+B2
E3: = E2+B3

F2: = MAX($K$1:E2)
F3: = MAX(F2:E3)

G2: = (E2-F2)/F2
G3: = (E3-F3)/F3

Maximum drawdown: =MIN(G2:G101)

Enter the first formula in each appropriate row and fill downward. Format the drawdown column as a percentage.

---

Costs, samples, and violations: three ways metrics become misleading

A metric is only as credible as the data and rules that produced it. The following three checks should accompany every review.

1. Measure performance after friction

Imagine a system whose gross average result is $+0.20R$ per trade. If commissions, bid-ask spread, slippage, and conversion costs average $0.08R$ per completed trade, net expectancy falls to:

$$+0.20R-0.08R=+0.12R$$

That still may be viable. But a gross expectancy of $+0.05R$ becomes negative after the same costs. Small-target, high-turnover approaches are especially vulnerable because costs consume a greater proportion of their intended reward.

Your simulation should be conservative rather than theatrically precise:

• Use commission settings that match the broker or product you may eventually use.
• Record a realistic, fixed slippage convention appropriate to the liquidity and timeframe.
• Revisit the assumption if you later test smaller-cap shares, exchange-traded commodity products, or crypto pairs with different liquidity.
• Evaluate net results as the headline result; retain gross P/L only as a diagnostic.

2. Respect the sample size

Three to five replay trades were enough to practise execution. They are not enough to estimate a strategy’s real win rate, expectancy, or drawdown.

A sample can look strong because a few unusually large winners occurred early. It can also look poor because a normal losing streak occurred early. The uncertainty is larger when:

• the setup occurs rarely;
• trades are concentrated in one market regime, such as a strong bull trend;
• several trades are highly correlated, such as breakouts in multiple technology shares during the same week;
• you changed the strategy after observing early outcomes;
• the instrument has unusual volatility or liquidity conditions.

As a practical benchmark, collect at least 100 fully defined historical or replay instances before treating the metrics as a rough estimate, and seek a larger sample across distinct market conditions before drawing stronger conclusions. A set of 200 trades from one narrow condition is still not the same as robust evidence across ranging, trending, calm, and volatile periods.

3. Separate strategy results from execution errors

The journal column for rule adherence is essential. At review time, calculate two views:

1. All executed trades: this is the real performance of your current process, including mistakes.
2. Fully rule-followed trades: this estimates how the documented rule set performed when actually implemented as intended.

Typical violations include entering before a bar closes, moving a stop without a rule, taking profit early because of discomfort, skipping a valid setup, or changing an indicator setting during a test.

Do not delete violating trades because they are embarrassing or unprofitable. Tag them. A profitable rule violation is still a violation, and a rule-followed loss is still valid strategy data.

There is one nuance: a missed valid signal has no executed P/L, so it cannot be silently added to your trade-return average. Record it separately as a missed opportunity. If you later reconstruct it as a hypothetical trade, label it explicitly as reconstructed rather than mixing it with the real replay execution record.

A disciplined review comment is therefore specific and conditional:

“Practice Rule v1 produced $+0.23R$ average net expectancy across ten daily replay trades, with $-2.00\%$ closed-trade maximum drawdown. The sample is too small to validate the method. One trade had a minor operational issue, and further tests must use the same cost assumptions and unchanged rules.”

That is much more informative than “the strategy works” or “the strategy failed.”

---

Key takeaways

A useful trading journal becomes evidence only after you evaluate net, consistently recorded results:

• Win rate measures how often completed trades were profitable.
• Average win and average loss measure the size of each outcome category.
• Expectancy combines frequency and payoff into the average net result per trade.
• Maximum drawdown measures the deepest peak-to-trough fall in the equity curve, so the order of results matters.
• Costs, small samples, and rule violations can materially distort every metric.

For your next review session, run another short Bar Replay batch using the unchanged Practice Rule v1, then append it to the same journal rather than replacing the earlier results. Keep the strategy version, instrument category, timeframe, cost assumptions, and adherence tags visible.

The next module shifts from manual chart evaluation to Pine Script foundations. You will begin by learning TradingView’s bar-by-bar execution model, including why series values and repainting matter when a chart rule is translated into code.