Your Broker Statement Is Not a Trading Journal

This is the second post in a short series on why traders log trades at all. The first one argued that an account balance is a sum, and a sum cannot be decomposed after the fact, so a balance can never name the strategy that is losing. A reasonable reply to that is: fine, but I do not only have a balance. I have a full statement, every trade, every fill, every fee. This post is about why that is a different objection than it sounds like, and why the answer is not that your broker's records are inadequate.

The Statement Is the More Reliable Document

Start by conceding the strong form, because it is correct. When your log and your broker's statement disagree, the statement is right and your log is wrong. It has the actual fill rather than the price you thought you got. It has the partial you forgot about. It has the swap charged overnight on the position you were sure you closed on Friday, the commission tier that changed mid-month, and the exact millisecond of an execution you remember as roughly lunchtime.

Hand-kept logs drift, and they drift in predictable ways. People round entry prices. They record the stop they meant to use rather than the one that was live on the server. They log a scaled exit as a single close. Anyone telling you that your journal is a more accurate record of what happened than your broker's is selling you something, and the honest position is the opposite: the statement is the arbiter, and reconciling against it is a good habit rather than a chore.

And For What It Is For, It Is Enough

The second concession is bigger. For a whole class of real questions, the statement is complete and a journal adds nothing at all. How much did I make this quarter. What did I actually pay in commission and financing. Does my platform's reported P&L match what the broker says it is. Which of these positions is still open. What do I hand my accountant. On every one of those the statement is not merely adequate, it is the answer, and there is no version of a trading journal that improves on it.

We have argued this before from the other direction. When a prop firm's dashboard and a trader's own records disagree about the money, the dashboard wins, and the useful work is reconciling the gap rather than disputing it. So the category error under discussion is not that traders over-trust their statements. It is subtler than that: it is expecting a record of transactions to answer questions about decisions.

Four Things a Statement Structurally Cannot Contain

The word doing the work in that heading is structurally. These are not fields your broker forgot to include, or that a richer export format would recover, or that a better parser could infer. They are facts that never existed on the broker's side of the connection in the first place. Your broker received an order. Everything upstream of that order happened in your head.

1. Why you entered

Your broker saw a market order for 0.80 lots. It did not see a London session reversal, or a failed breakout you were fading, or the fact that this was the third attempt at a level you had been watching since Tuesday. Without that label, per-strategy results are not merely hard to compute, they are undefined, and that is the entire argument of the previous post in this series rather than something to re-litigate here.

2. What you intended to risk

This is the expensive one, and it is the field most people do not realise they are missing until they try to do something with their data.

A statement has a fill and an exit. It does not have the stop you planned before you entered. Two consequences follow, and both of them are load-bearing:

• Results stay in dollars. A trade where you risked $50 and a trade where you risked $500 sit in the same column, so a good day and a reckless day are indistinguishable if they happen to end at the same number. R-multiples exist to fix exactly this, and R is a ratio whose denominator is planned risk. No planned risk, no denominator, no R.
• MAE loses its meaning. You can compute from price data how far a trade went against you before it recovered. What you want to know is how far it went against you as a fraction of what you were willing to lose, because that is the version that tells you whether your stops are too tight or whether you are surviving on luck. Adverse excursion measured in pips is trivia; measured in R it is a diagnosis.

One honest note here, because it cuts against our own product. A live broker connection does not solve this. SignalDeck's MT4/MT5 sync pulls entry, exit, size, realised P&L and account balance automatically, and it still asks you to type the stop-loss level in by hand, because the bridge does not reliably expose it. That is a real limitation and it is not specific to us. If the number was not in your head at entry, no integration invents it afterwards.

3. The trades you did not take

A statement is a record of what executed. It is silent, and necessarily silent, about the setup you passed on because you were already near your daily limit, the one you talked yourself out of at the open, and the two you missed because you were watching a losing position instead.

This absence has a particular shape. A trader who systematically skips their best setups and takes their marginal ones is describing a serious problem, and their statement looks unremarkable, because the good trades that would have made the difference are not in it. There is no amount of analysis on executed trades that recovers a distribution you never sampled.

4. Which rule you broke

Moving a stop further away, adding to a position that is already underwater, opening a trade inside the news window your own plan forbids, doubling size after two losses. Mechanically, every one of these produces a perfectly ordinary row. A widened stop is a modification, and on many statements it is not even a separate line. A revenge trade has the same shape as a planned one.

Some of these leave a forensic trace if you go looking, which is a real caveat worth stating: an unusually large position after a losing streak is visible in the size column if you sort for it. But a trace you have to reconstruct by inference is not the same as a field, and it only works for rule breaks that change the numbers. The ones that do not change the numbers leave nothing.

The Same Trade, in Two Records

One trade, recorded twice. Nothing in the left column is wrong or missing for its purpose, and nothing in the right column is exotic.

The left column is a complete and correct record of a transaction. The right column is a record of a decision that happens to include a transaction. Only the second one can be scored, and the difference between them is five fields that no counterparty was ever in a position to supply.

Reconstructing It Later Is Not a Recovery Path

The obvious response is that none of this needs to happen at entry. Export the statement at the end of the month, sit down with it, and fill in the missing columns from memory. This is the most common plan and it fails for two separate reasons.

The first is ordinary: you will not remember. Beyond a week or two, one EURUSD trade is not distinguishable from another, and the level you were watching, the reason you sized down, and the thing that made you exit early are gone. This is annoying but survivable, because at least you know the data is missing.

The second is worse, because you will remember, and the recollection is not random. Memory reconstructs with the outcome already known, and it reconstructs in a direction. A trade that lost becomes one that was not really your setup. A trade that won gets admitted to whichever strategy you currently believe in. Nobody does this deliberately, and the resulting dataset is not merely noisy, it is biased toward whatever conclusion you already hold. A backfilled label is a story about a trade rather than a record of one, which is why the previous post insisted the label goes on at entry, before the outcome exists to contaminate it.

There is a smaller practical point underneath both. Reconstruction is not less work than logging. It is the same work, done later, at lower quality, in a single unpleasant sitting rather than thirty seconds at a time.

What you genuinely can rebuild from a statement is the mechanical half: instrument, size, entry and exit with timestamps, fees, realised result. That is two of the four fields analysis needs, and as the previous post put it, they are the two that come free from an import and do none of the analytical work.

On a Funded Account, the Firm Dashboard Is Also a Statement

Prop traders usually have something better than a raw broker statement. The firm's dashboard adds the number a statement lacks, which is distance to your drawdown limit, and it is the authoritative record of whether you are in breach. Nothing here suggests otherwise.

It is still a record of outcomes. It reports that your equity fell 3.1% on Tuesday. It cannot tell you that 2.4% of that came from one trade taken eleven minutes after a loss, in a session you had already decided to sit out. Both accounts fail the same way on the dashboard and they are different problems with different fixes.

The field that matters most here is also the one no statement carries: the worst intraday excursion on the trades that survived. A firm's daily loss limit is tested against your equity path rather than your closing balance, which we went through in detail for Topstep and The5ers. A week where you never breached but came within a few hundred dollars four times is a warning, and it is invisible in every record that only stores where trades ended.

What SignalDeck Adds to an Import, and What It Does Not

The honest framing of the product is that it does not replace your statement. It imports it and then asks you for the two fields your broker could not supply.

• Import handles the mechanical half. CSV or a live connection brings in instrument, size, fills, timestamps, fees and realised P&L, so the part the statement is authoritative about is not retyped.
• Strategy label and planned risk at entry. The two fields that carry the analysis, captured before the outcome exists, which is the whole point of capturing them at all.
• R-multiples and expectancy computed from planned risk, so trades taken at different sizes become one comparable series. Basic R analytics are on the Free tier; SQN, Kelly and Monte Carlo drawdown are Pro ($30/mo).
• MAE and MFE against your planned stop, which is the version of adverse excursion that means something.

Three limits worth knowing before you plan around any of it. MT4/MT5 live sync is Elite ($50/mo) and, as noted above, still needs the stop-loss entered by hand. Whether your trades arrive automatically at all depends on your platform, and our platform support guide lists which are live-sync, which are CSV only, and which are neither, so check it against your own setup rather than assuming. And the third field in this article, the trades you passed on, has no automated capture anywhere in the product: it is a note you write or it does not exist. We would rather say that than imply the software solves it. The Free tier carries unlimited trades and basic R analytics but caps you at 3 strategies and 10 tags, which is a real constraint on the attribution described here; unlimited strategies and tags are Pro. Everything is unlocked at no cost during beta.