Documenting Your Trading Data: What to Record All Year
A complete trading history is not created at year end but on every trading day before it. Recording as you go leaves you with a copying exercise later; not recording leaves you with a reconstruction. What it takes is five items per event, one firm rule about exports, and four moments in the year.
Why is year end the wrong time for this?
Because by then you can only find what still exists. Three things cannot be recreated after the fact: closed accounts, expired retrieval windows, and details that were never recorded anywhere.
An effort effect comes on top. Recording as you go costs minutes per event, reconstruction costs days, and it reliably falls into a period when you had planned to do something else.
The scope here is deliberately narrow: not how to analyse your trading, but which data has to accumulate across the year so that analysis becomes possible at all.
Which items accumulate per event?
Five, and none of them can be derived later if missing.
- Type of event. Buy, sell, internal transfer, deposit or withdrawal. A short fixed pick list rather than free text, otherwise grouping becomes impossible later.
- Date and cost basis. The two most important fields of all, because they underpin every later view of a position.
- All cost components. Not the commission alone but spread, financing costs and transfer costs, each in its own column.
- The provider. Tracked separately, because holdings and costs arise per provider and can only be recombined later with that field present.
- Inflows outside trading. Anything that arrives without you having bought or sold, with its date and its value at the moment it arrived.
The fifth category gets forgotten most often, because it does not feel like a trade. It appears in no trade list yet changes your holdings regardless, and the value at the moment of the inflow can only be reconstructed later with effort.
What does a connection supply, and what does it not?
The distinction is worth drawing, because it shows where your own effort actually sits. An exchange or broker connection typically supplies timestamps, quantity, price and the fees charged, meaning the mechanical part.
What it does not supply is anything requiring judgement: whether a transfer between your own accounts was an internal movement or an event with a counterparty, which idea a trade belonged to, and what context an inflow arrived in.
Check deliberately, once, which field each cost type lands in. Some interfaces deliver ancillary costs as their own event while others net them straight into the result, and in the second case your cost ratio disappears from the analysis without anyone noticing.
Why is the provider's report not enough?
Because it answers a different question. It shows totals for one provider and one period, not individual events across several providers and several years.
For a quick overview that suffices. But as soon as something has to be evidenced, split or combined across providers, you need the layer beneath it.
The practical test is simple: could you trace the life of a single position from three years ago out of your own records, including every cost? If not, that layer is what is missing.
What is lost when switching providers or closing an account?
Everything behind the login, and that is more than most people expect: execution details, cost statements, order history, and the ability to generate exports at all.
Order is the critical point. Once the account closes the access is gone, and with it the source. So: save first, transfer second, close last. What else needs attention in that sequence is covered under switching brokers and which records you lose.
Remember the technical access too. API keys stop working with the account, retroactively for anything not yet synced.
When should you save exports?
At three points, all of which can be set as fixed dates.
- Quarterly, a complete transaction export in a machine readable format. Four files per year and provider, no more.
- At the turn of the year, the full statements for the completed year saved as documents as soon as they are available.
- Before any change to an account, meaning before a transfer, a closure or a change of provider.
Choose a format you can process afterwards. A screenshot evidences nothing you could later sort, total or reconcile.
Where should you keep all this?
In one place rather than four. That is the one decision here that is hard to reverse, because scattered records cannot be merged later without the gaps in the same periods remaining visible.
A folder of documents is an archive rather than a record. It answers whether a document exists but not what a particular event cost or which positions were open during a period. You need both separately: the archive for evidence and a structured record for analysis and reconciliation.
One criterion decides more than the rest: the record has to survive a change of provider. Anything living only inside a provider's system ends with the account. What else deserves weight when choosing a provider is covered in comparing trading brokers.
What does an annual rhythm look like?
Four moments suffice, and none takes long when the previous one was kept.
Monthly you check completeness: is every event recorded, including those outside trading? That takes minutes and prevents most later gaps.
Quarterly you save the exports and check whether every cost component came through.
Before any account change you save again, regardless of the quarter. That moment cannot be scheduled, and it is the most important one.
At the turn of the year you reconcile the providers' statements against your own records, while the period is fresh enough for discrepancies to be resolved.
How do you know your records are complete?
Through a single control calculation, and it is the most important part of the whole routine. Derive the holdings per provider at a given date from your recorded events and compare them against the actual holdings there.
If the two agree, your records for that period are complete. If they diverge, an event is missing, and almost always one from the fifth category, meaning an inflow or an internal movement that did not feel like a trade.
Check the cash balance too, not just the positions. Missing costs show up there while staying invisible in quantities. The check takes a few minutes per provider and replaces any guesswork about whether the data holds.
What if a period is already undocumented?
Rebuild what is necessary rather than everything. Attempting to reconstruct several years in full usually fails on effort and availability, and it is not necessary either.
This order works. First save whatever is still retrievable, at every provider and as far back as possible. That is the only step under time pressure, because retrieval windows end and accounts get closed.
Then settle the point that genuinely counts: can you trace the date and cost basis of the positions you still hold today? That question is independent of how patchy everything else is.
Only then comes the rest, and one boundary applies: whatever cannot be evidenced with reasonable effort gets documented as a gap with its period and a reason rather than estimated. A named gap holds up better than an invented figure, because it can be accounted for during analysis.
Where does the real advantage sit?
In the fact that your own record is the only body of data that knows nothing of a provider change, an account closure or a system migration. It is also the only one producing a combined view across several providers.
That is more than convenience. Multi year analysis, meaning the development of your metrics per setup, your hit rate and your cost ratio, needs an unbroken time series. Every switch cuts that series exactly where the old provider's history stops being retrievable.
A continuously kept trading journal supplies most of the five items anyway, since timestamps and costs come from the exchange connection. What has to come from you is the type of event and its attribution to a provider, and that is two clicks.