How we count

We trade only our own money. We have no clients and no investors, and we are not raising capital. We keep a full record anyway, because the people who decide whether to work with us are better served by figures they can check than by assurances they have to take on trust. We share it with them directly, on request, rather than publishing it here.

Who the record is for

The readers we have in mind are practical ones. Each of them is deciding whether we are a sound business to deal with, and each of their questions is easier to answer with numbers in front of them.

• A venue

  Wants to know, for example for one of its programs, that a firm quoting on its book has the capital to stand behind those quotes and keeps careful track of its positions.

  For venues

• A bank or payment provider

  Wants to understand where the money in our accounts came from and how it moves.

  The firm

• A counterparty

  Wants some evidence that we settle what we owe.

  Ask for the record

• A regulator

  Wants to see how the firm trades, and that any trade can be reconstructed from what we keep.

  Operating rules

How we count

The record shows three things: operating revenue, trading results, and the value of positions that are still open. The figures are unaudited.¹

Open positions take the most care. A contract that has not resolved has no settled value, only an estimate of what it is worth. For each open position we take an independent price estimate and use the conservative end of its range, the lower bound, rather than the midpoint.² To take a made-up case, if the estimate for a contract we hold runs from 40¢ to 46¢, we count it at 40¢.

Open position · made up

Estimate 40–46¢

30¢ 40¢ 43¢ 46¢ 56¢

• Counted: the lower bound
• The estimate’s range
• Midpoint, not used

Some positions have no current estimate at all, for example because nothing has traded in them recently. We count those at a value between their full cost and zero, and never above what we paid.

We measure the lifetime result two independent ways. The first works from the accounts: what they are worth today, minus the money we have put in, plus the money we have taken out. The second builds up from the activity itself by adding operating revenue to trading results. The two draw on different inputs, so a gap between them means something has gone wrong somewhere: a transfer recorded in one place and missed in the other, a fee booked twice, an open position valued two different ways. We check one against the other. The first has to land inside the range the second allows (it is a range because of the positions with no current estimate), and a result outside it is an error we track down.

Case A

From activity: a range

From the accounts

Case B

From activity: a range

From the accounts

Two further rules apply to every figure. Operating expenses and taxes are excluded, so the record shows what the trading produced before the costs of running the business.³ And every figure is rounded in the cautious direction: totals down, and the value of open positions toward the larger loss. Nothing is ever rounded up.

Words we do not use

We never describe these figures as a return, a yield or an ROI, and we never annualize them. Those terms answer an investor’s question: what did my money earn, and what might it earn over a year? We are not raising money, so nobody needs to read our record with that question in mind, and we would rather not invite it.

There is a practical problem too. A return needs a denominator, and a firm that adds and withdraws its own capital as it sees fit has no single honest one. Divide by the starting balance and you get one answer. Divide by the average balance or the peak and you get others. Annualizing a short or uneven history then stretches whichever answer you picked. Plain dollar amounts, counted conservatively and checked two ways, sidestep both problems.

Why the rules are public and the figures are not

The figures go to the people who have a reason to see them: a venue we quote on, a bank or reviewer doing its checks, a counterparty. Each copy carries its as-of dates and notes. Publishing the figures to everyone would mostly tell the people we trade against how large we are.

The rules are different. Anyone can read them here, which is the point: when a figure arrives, the method behind it is already on the record. When a figure helps explain the work itself, it will appear in a research note, dated and with its method beside it. Requests for the record, and questions about how we count, go to compliance@cipherg.io.

Footnotes

1. Unaudited means no outside accountant has examined the figures. This note sets out how they are counted so readers can judge the method for themselves. ↩

2. The lower bound is the bottom of the range the estimate treats as reasonable for the contract’s value. Using it makes an open position more likely to be understated than overstated. ↩

3. Because expenses and taxes are left out, the figures will be higher than the firm’s profit after expenses and taxes for the same period. Anyone comparing the record with a bank statement or a tax return should expect that difference. ↩

@techreport{cipherg-n001,
  author      = {Dodson, Grayson},
  title       = {How we count},
  institution = {CipherG Liquidity LLC},
  type        = {Research note},
  number      = {001},
  year        = {2026},
  month       = oct,
  url         = {https://cipherg.io/research/how-we-count/},
  note        = {Version 4, last changed 2026-10-24}
}