SAVINGCAPITAL
Sign in Explore the demo
CAPITAL JOURNAL

Understand your trades.
Improve your decisions.

Connect an account and your history imports itself. Every metric is computed from your own fills and carries the sample size behind it. Repeated behaviour is surfaced with the trades that evidence it — and nothing is claimed before the data can support it.

Explore the demo See how it works
Read-only access. The journal never trades, moves funds, or changes a setting on your account.
◱Trade review
43W · 25L
Every trade carries the plan it was opened against, so adherence is a fact rather than a memory.
◈Process score
8.0/10
3 of 6 dimensions scored. The rest say why they could not be.
◇Observed pattern
×16
After a loss you size above your median 64% of the time (16 of 25). After anything else …
Performance

Figures that say how much they know

Every metric carries its sample size. A profit factor from six trades and one from six hundred look identical on a dashboard and are not the same claim — so this one tells you which it is, and says nothing at all when the data cannot support an answer.

By day of week realized P&L
Mon: +$526.01 · 16 trade(s)MonTue: +$309.94 · 7 trade(s)TueWed: +$526.29 · 10 trade(s)WedThu: +$822.64 · 13 trade(s)ThuFri: +$508.40 · 9 trade(s)FriSat: +$345.45 · 10 trade(s)SatSun: +$69.54 · 3 trade(s)Sun
A strong weekday may reflect what was traded that day rather than the day itself. This shows the pattern; it does not explain it.
What is measured and what is withheld
Realized P&L✓net of fees and funding
Win rate✓breakeven excluded from the denominator
Profit factor✓unavailable, not infinite, with no losses
Average win / loss✓shown as both sides, not just a ratio
Account equity—needs balance history — not inferred
R-multiple—only where the original risk was recorded
The last two are the point. Inventing them is easy and would make every number above it untrustworthy.
Trading psychology

Behaviour you can check, not a diagnosis

Each observation names what happened, how many times, out of how many opportunities, and what it does not show. Every one links to the trades underneath it, so you can disagree with it.

Observed patterns with their evidence
After a loss you size above your median 64% of the time (16 of 25). After anything else it is 40% (17 of 42).
Size varies for many reasons, including conviction and volatility. This compares two rates; it does not read intent.
×16 of 25
1 of 25 active days were in profit at some point and finished negative.
Measured on closed trades in UTC order. A day that ends negative is not evidence of a decision; it is a prompt to look.
×1 of 25
3 of 24 active days went above 5 trades. Your typical day is 2; this compares against 2× that, because you have not set a limit of your own.
A busy day is not automatically a worse one; the P&L is shown so you can judge.
×3 of 24
A pattern seen three times is an observation. The same pattern seen forty times is a tendency. The count is always shown because the difference matters.
Three things kept separate deliberately
Measured behaviour
What the execution record shows. You re-entered within 90 seconds of a loss, 14 times.
What you reported
What you wrote in your own journal at the time. Yours, private by default, never used as an instruction to anything.
Tentative reading
A suggested interpretation, labelled as one, phrased as a question you answer.
Collapsing these three is how journaling software starts telling people they were panicking. A transaction log cannot contain an emotion, and this one does not pretend otherwise.
How it works

Three steps, and the third one waits

Most journals will happily score you on four trades. This one reports insufficient data until it is not, which is slower and the only version that means anything.

01
Connect or import
Link a read-only exchange key, upload a statement, or type a trade in by hand. A key that can place orders is refused rather than stored — the journal imports history and has no ability to trade.
02
Review the trade
Each trade carries the plan it was opened against: thesis, stop, target, tags. Afterwards you add what happened and what you would repeat. The plan and the hindsight stay separate.
03
See the pattern
Once there are enough trades to say anything, repeated behaviour surfaces with its evidence attached — and stays silent until the sample supports it.
What it does

Built around what the data can actually support

◈
Read-only, always
The journal cannot place an order, change leverage, move funds, or alter any automation. A credential that turns out to allow trading is refused, not stored.
◱
Every venue, eventually
Crypto exchanges by API key, CFD and FX brokers, MetaTrader through a bridge, and a file import that works for anything else. Bybit is live; the rest are listed with their real status.
◇
Positions, derived
Fills are the record and positions are replayed from them. A late fill re-derives the history correctly instead of patching a number whose origin nobody can reconstruct.
▤
Deposits are not profit
Cash movements live in a separate table from trades, so a transfer cannot be summed into a result by accident.
▥
Your notes stay yours
Private by default, and treated as data — never as instructions to a model, no matter what they contain.
◆
Written summaries, on request
A summary is generated when you ask for one, never on page load, and every figure it quotes is computed before it is written. It phrases findings; it does not produce them.

Look at it with data in it

The demo runs on a fixed fictional dataset. Every chart, total, calendar cell and trade row is computed from that one set, so they agree with each other — and you can click any figure down to the trades underneath it.

Explore the demo Create an account
No card required to look around. The demo connects to nothing.