A futures trading journal is not a list of trades won and lost: it is the only way to know whether your edge exists or whether you have spent three months living off a streak. Here is what to record on every trade, which metrics to watch (and which to ignore), a review routine that fits in five minutes a day, and the mistakes that turn a journal into a dead file.
What to record on every trade
The rule is simple: write down everything you need to reconstruct the trade without opening the chart, and nothing you will not use in a decision. In futures the list is fairly short:
- Instrument and contract. ES, MES, NQ, MNQ, CL, GC… and the expiry if you trade near the rollover. A micro and a mini are the same market at a different size: store instrument and size separately so you can group them later.
- Side and size. Long or short and number of contracts. If you scale in or out, record the full position, not every fill: you analyze decisions, not executions.
- Entry and exit price and time. With the exact time you can slice by session and by hour, which is where the surprises show up.
- Session. Globex, European session, New York open, midday, close. The same setup can have positive expectancy at 9:30 and negative at 12:00.
- Setup. The name of the idea that justified the entry, chosen from a closed list. Without a setup, the journal only measures your aggregate luck.
- Risk in dollars and in R. How much you would lose if the initial stop was hit. It is the number that normalizes everything else.
- MAE and MFE when the platform provides them: the maximum adverse excursion (how far the trade went against you) and the favorable one (how far it went your way). They tell you whether stops and targets sit where they should.
- A short note. Why you entered, why you exited, what you would repeat. Two lines written that day are worth more than a paragraph written on Sunday.
- Chart screenshot. At entry and at exit. The numbers tell you what happened; the screenshot tells you what you saw.
If position sizing forces you to do arithmetic every time, the position size calculator does it without signing up, and futures position sizing has the full reasoning.
What to measure: metrics that change decisions
The month's P&L is the result, not the diagnosis. These are the metrics that do point at what to change, with their formula and what they mean:
| Metric | Formula | What it tells you |
|---|---|---|
| Win rate | winners ÷ total | How often you are right. Alone it says nothing: 70% with large losses loses money. |
| Average win / average loss | mean of winners ÷ mean of losers | How much you make when right versus what you lose when wrong. It is the counterweight to the win rate: always read them together. |
| Expectancy | (win rate × average win) − (loss rate × average loss) | What you expect to make per trade, in dollars or in R. It is the number that decides whether a setup deserves capital. |
| Profit factor | gross profit ÷ gross loss | Robustness of the whole. Below 1 you lose; hugging 1 you depend on the streak and commissions decide. |
| R-multiple | trade P&L ÷ initial risk | Normalizes trades of different size and instrument: +2R in MES and +2R in ES are comparable. |
| Max drawdown | largest drop from a peak of the equity curve | How deep the worst stretch has been. In a prop account, compare it with the trailing drawdown you have left. |
| Average MAE / MFE | mean adverse and favorable excursion | Where stops should be and how much profit you leave on the table. |
| Net vs gross P&L | gross P&L − commissions and fees | What commissions take. In micros they can be a meaningful share of the result. |
Two reading notes. Win rate and average win/average loss are the two halves of the same equation: either one alone misleads. And profit factor and expectancy tell the same story two ways; once you have them, win rate stops being the headline and becomes context: it helps anticipate streaks, not judge the edge.
A worked expectancy example
Suppose 40 MES trades over a month, always with 2 contracts and a 10-point stop, so 1R = $100. The journal says:
- 22 winners with an average win of $180 (1.8R).
- 18 losers with an average loss of $110 (1.1R): some stops slipped and in a couple you moved the stop.
Win rate = 22 ÷ 40 = 55%, loss rate = 45%. Expectancy = (0.55 × 180) − (0.45 × 110) = 99 − 49.5 = $49.50 per trade, that is, 0.495R. Profit factor = (22 × 180) ÷ (18 × 110) = 3,960 ÷ 1,980 = 2.0. If you assume $4 of round-turn commissions per trade with those 2 contracts, net expectancy drops to $45.50 and the 40 trades leave about $1,820 net.
Now move a single variable. If the average loss rises to $150 because you let two or three losers run, expectancy falls to (0.55 × 180) − (0.45 × 150) = 99 − 67.5 = $31.50: 36% less with the same win rate. If instead the win rate drops to 45% with the original losses, it lands at (0.45 × 180) − (0.55 × 110) = 81 − 60.5 = $20.50. The journal exists to watch those two variables move before the equity curve shows them to you.
Slice by setup, session, instrument and account
A positive overall expectancy can hide one setup that loses and another that carries the account. The four slices that move the most decisions:
- By setup. The main slice: it turns "I make money" into "I win on the opening breakout and lose on midday fades". How many trades you need per setup before trusting the number is in the FAQ below; how to tag in bulk, in the setups features.
- By session hour. Group by 30- or 60-minute windows. It is common for most of the net result to come from two windows while another erodes it trade by trade.
- By instrument. ES and NQ move together almost always, but they do not pay or punish the same: the same 10 points are $200 per contract in NQ and $500 in ES. Measure in R to compare, and check the correlations between futures before adding positions that are really the same bet.
- By account. In prop firms the same trade is replicated across several accounts with different sizes, and each account has its own daily limit and trailing drawdown. The per-account slice tells you how much room is left in each, which is what decides whether you can trade tomorrow.
Review: 5 minutes a day, 30 a week
Cadence matters more than depth. A journal that demands an hour a day is abandoned in two weeks; one that demands five minutes survives.
At the close of every session (5 minutes). Tag the day's trades with their setup, write one line on each and check two numbers: the day's net P&L against your daily limit and the distance you have left to the trailing drawdown. Nothing else: today you do not analyze, you record.
Once a week (30 minutes). A table per setup with number of trades, win rate, average win and loss and expectancy; a slice by hour; the P&L of trades flagged as mistakes. Out of that comes one decision, written with a date and the sample size that supports it. One, not five: if you change three things at once, next week you will not know which one worked.
Once a month. Max drawdown, result per account and a review of the risk plan. If the cushion has grown or you have strung losses together, that is the moment to move risk per trade according to your risk ladder, not in the middle of a session.
Automatic sync, manual entry or spreadsheet
There are three ways to get trades into the journal, and the choice determines which data you will actually have.
- Spreadsheet. Free and flexible. The cost is that you type every trade: time and price errors creep in, commissions get forgotten and, without meaning to, the trades that do not fit get left out. It works with one account and a few trades a day; with two accounts or scalping, it stops working.
- Manual entry in a journal app. Better structure (fixed fields, setups from a list, screenshots next to the trade), but the same underlying problem: the data is whatever you decide to enter.
- Sync from the platform. Trades come from the NinjaTrader database or from the CSV that Tradovate, Rithmic, Sierra Chart or ATAS export: all of them, in order, with the exact time and the real commissions. Your work shrinks to what the machine cannot do: tagging and annotating.
The important nuance: syncing guarantees the data is complete, not that the journal has value. The value still lives in the tags and the notes, which are manual in any system. If you use NinjaTrader 8, exporting NinjaTrader 8 trades walks through both paths, database and CSV, step by step.
Mistakes that empty a journal of value
- Recording only P&L. You know how much, not why. Without initial risk there is no R; without time there is no session slice; without setup there is nothing to improve.
- Trading without tags. A journal without setups is an account statement with notes. The tag is what turns 200 trades into four questions that can be answered.
- Ignoring commissions. A micro scalper can have positive gross expectancy and negative net. Always measure net.
- Choosing what goes in. "That one doesn't count, it was a fat finger" is the sentence that has ruined the most journals. Everything goes in, and the fat finger gets its mistake tag.
- Documenting only the losers. Winners carry information too: how many were won outside the plan, how many left half the move on the table (MFE), which were closed out of fear and not by rule.
- Mixing sizes without normalizing. A month with 1 ES and another with 3 MES do not compare in dollars. They compare in R.
- Reconstructing the risk after the fact. If the stop "you had in mind" is decided after seeing the result, the R-multiples are inflated and the expectancy lies.
Frequently asked questions
How many trades do I need before drawing conclusions from the journal?
Under 20 trades per setup there is no verdict: with 10 trades, one standard deviation of the observed win rate is about 16 points, so 70% is consistent with a coin flip. Between 30 and 50 you start to see the sign of the expectancy and the big differences; from 100 you can fine-tune details like stops and time of day.
Is a spreadsheet good enough for a futures journal?
Yes to start, as long as every row records initial risk, time and setup and not just P&L. The limit comes with several prop accounts or many trades a day: the typing time and the transcription errors kill the discipline, and that is when syncing from the platform pays off.
Should I log trades in dollars or in R?
Both. Dollars are for managing the account: daily limit, trailing drawdown, payout. R is for comparing setups, instruments and accounts of different sizes on one scale. If your risk per trade is always the same, both measures rank trades identically.
Next step
Start with what pays off most: fix a closed list of setups and tag the trades of the last two weeks; the per-setup slice is the one that changes the most decisions. If you trade with NinjaTrader 8, connect the database with the NinjaTrader journal and spend the five daily minutes tagging, not copying numbers. And if your accounts are prop firm accounts, the prop firm tracker shows the distance to every limit without computing it by hand.