What Is a Trading Decision Log, and Why Every Vote Should Be Dated

A trading decision log is the dated record of what a strategy’s decision-makers actually decided, when they decided it, and how each decision was made – including the votes that lost. It is not a journal of feelings and it is not a highlight reel. It is a ledger. Its purpose is to make a process checkable by someone who was not in the room.

You probably already know how to read a filing. Read a decision log the same way: for what it states, what it omits, and the date stamped on every line.

What a trading decision log records

Ask ten people what a decision log contains and you will get ten answers. Here is the working definition this brand uses, and it is narrow on purpose.

A decision log records, per entry:

• The instrument and the action considered.
• The vote – how many in favor, how many against.
• The dissent, attributed and argued, not summarized away.
• The size and the fill, when a decision turns into an order.
• The as-of date: the moment the decision was made, not the moment it was published.

That last line is the one most people miss. A log without dates is a story. A log with dates is evidence.

Why dissent belongs in the log

Most published track records are written by the winner. A position that worked gets an entry. A position that failed quietly disappears, or gets folded into a later success with a sentence like “we adjusted.”

That is not a log. That is a narrative with a good editor.

A decision log keeps the losing votes because they are the part you can learn from and the part that proves the process is real. If a committee never records disagreement, you are not reading a process. You are reading a press release. The dissent is where the method shows itself – what was argued, what the downside case was, and why the majority went the other way anyway.

Dissent also gives you something to measure. A strategy that records its minority views and later checks whether they were right is doing something a tip sheet cannot: it is building a file on its own judgment.

Why every vote should be dated

Dates do three jobs, and none of them is decoration.

First, an as-of date fixes a claim in time. “We were long before the move” is worthless. “We entered on the ninth, filled at the open on the tenth, and here is the log entry dated that morning” is a claim you can verify against the tape.

Second, dates expose the tell. A track record published all at once, with no dates attached to the individual decisions, is almost always assembled after the fact. When you cannot see when each call was made, you cannot tell foresight from hindsight. The dates are the only thing standing between a track record and a rewrite.

Third, dates make losses schedulable. If wins are dated and losses are not, the record is not a record – it is selection. The point of dating every vote is that losses get published on the same schedule as wins, in the same format, with the same discipline. If the cadence breaks when the news is bad, you have your answer about the operator.

What a decision log is not

It is worth stating the negative, because the term is used loosely.

A decision log is not a watchlist. A watchlist is a set of ideas; a log is a set of commitments with dates and votes attached.

A decision log is not a P&L statement. The P&L tells you what happened to the money. The log tells you what was decided and why, which is how you judge whether the next decision will be any good.

A decision log is not a marketing document. The moment a log is written for an audience rather than for the record, it stops being evidence. You can tell the difference by looking for the entries that make the author look bad. If there are none, you are not reading a log.

How to read a decision log you did not write

Here is a short list you can apply to any published record, ours included.

1. Look for a date on every entry. Undated decisions are unverifiable decisions.
2. Look for a losing vote. A record with no dissent is a record with no process.
3. Look for losses on the same cadence as wins. Same format, same schedule, no quiet months.
4. Look for the fill, not just the call. A decision that never became an order is an opinion.
5. Look for what is missing. The gaps in a record are as informative as the entries.

None of this requires special access. It requires only that the operator published enough to be checked – and that you take the time to check it.

The public log at Diers Capital

Diers Capital keeps a public log of every vote, dissent and fill, dated as of the day the decision was made. It stays free to read whether or not you are a member. The membership adds the logged-in view on top: current positions, the full decisions history, and the cross-account picture.

We publish the losses on the same schedule as the wins, because a log with the losses removed is not a log. You can read the process and decide for yourself whether it holds up.

Nothing here is investment advice, and nothing here is a recommendation to buy or sell any security. A decision log is a record of what was decided, not a promise about what happens next. Past decisions, published or otherwise, do not tell you what future returns will be.

The useful question is not whether a strategy made money last quarter. It is whether you can see how it decided, when it decided, and what it admitted when it was wrong. A dated log is the only way to answer that.

A revenue multiple is a sounding line. Drop it over the side and it returns a number: so many times forward revenue, so many turns of EV to ARR. The number looks like a measurement of depth. Often it is a measurement of how the line was weighted.

False depth is the failure mode. The chart reads calm, the multiple reads reasonable, and the hull is still closing on a reef that the same numbers, read correctly, would have shown months earlier. This is an audit of that failure – not a forecast of any company’s fate, but a sequence of checks you can run against a filing before you trust the depth it reports.

Step 1: Establish what the revenue actually is

Start with the top line and refuse the label. “Revenue” on a SaaS income statement is frequently a blend of recurring subscription and one-time, low-quality line items: professional services, implementation, training, hardware. Those dollars bill once and leave. They carry the same weight in the multiple as a subscription dollar and none of the same durability.

Pull the revenue disaggregation note. If subscription and services are not separated, that absence is itself a finding. A company that reports a single blended revenue figure is asking you to price its recurring base at the rate its most repeatable dollars deserve. You cannot audit a multiple you cannot decompose.

Step 2: Test whether it recurs

Recurring revenue is a claim about the future, and the only honest test of it is retention. Look for two numbers: gross revenue retention and net revenue retention.

Gross retention tells you how much of last year’s base simply stayed. If it sits in the low 90s or below, a meaningful slice of the base churns every year, and growth has to run hard just to stand still. Net retention, which includes expansion, hides that churn work behind upsells. A headline net retention of 115% built on 88% gross retention is a different business from the same net figure built on 96% gross. Same sounding. Different seabed.

If the company does not disclose retention, or discloses only net, treat the depth as unverified.

Step 3: Check what survives to cash

Recurring revenue that never converts to cash is a claim, not a position. Move down the statement and follow the dollars.

Two checks matter most. First, gross margin adjusted for stock-based compensation: if the software margin looks like 80% but only after excluding a large non-cash expense, restate it and see what is left. Second, the burn multiple – net cash burned per dollar of net new recurring revenue added. A burn multiple above 3 means the company is spending more than three dollars of cash to buy one dollar of new recurring revenue. That is not scaling; that is buying soundings with the equipment fund.

A high multiple on revenue that consumes cash to exist is false depth read as growth.

Step 4: Check concentration and duration

Depth is not uniform across a reef. Neither is revenue.

Read the concentration disclosure. If a single customer or a small handful of contracts make up a material share of the base, the multiple is priced on a base that can leave in one renewal cycle. Then read contract duration: multi-year, annual, monthly. A base of monthly contracts priced at the same multiple as a base of three-year commitments is mispriced, whatever the headline growth rate says.

Step 5: Date the sounding

Every number you have gathered is as-of a date, and the date matters as much as the number. A multiple on ARR as of the last quarter-end is stale the moment the quarter that follows disagrees.

Look for the as-of stamp on each figure: revenue, retention, cash. Where a company presents a trailing-twelve-month figure beside a current-quarter narrative, reconcile them. The gap between the two is where false depth usually hides – the narrative is current, the metric is old, and the multiple is priced on the fresher-sounding of the two.

Step 6: Compare the rate of change, not just the level

Two companies can carry the same multiple and be nothing alike, because the multiple is a level and the risk lives in the slope. A base growing 40% at 90% gross retention is converting soundings into chart. The same base at the same multiple, decelerating from 60% to 20% with retention drifting down a point a quarter, is a hull that has already touched and is waiting for the keel to announce it.

Pull three or four periods of the same metric and read the direction and the second derivative. Deceleration under a flat multiple is the single most reliable warning that the depth reading was wrong.

Why false depth reads as depth

None of these checks is exotic. The reason a false sounding passes is structural. Multiples are quoted as a single number, and a single number is easy to price, easy to compare, and easy to repeat. The audit is a sequence of smaller numbers that disagree with each other, and disagreement is uncomfortable to hold in one glance. So the market holds the glance and drops the sequence.

The reef does not care which was easier.

What an auditable process does about it

A process you can trust does not escape false depth by being smarter than the market on any given name. It escapes it by refusing to price a multiple it has not decomposed, and by publishing the soundings that came back shallow on the same schedule as the ones that came back deep.

That second part is the one most strategies leave out. It is easy to publish a method that has only ever been applied to winners. A log that carries the bad soundings next to the good ones, each stamped with the date it was taken, is verifiable in a way a narrative is not. You can check it. You can disagree with it. You cannot be quietly told a different story later.

The short version

Before you trust a SaaS revenue multiple, decompose the revenue, test the retention, follow the dollars to cash, size the concentration, check the contract duration, date every figure, and read the slope rather than the level. Where a company – or a strategy – will not show you those soundings, the depth is not deep. It is unmeasured, which is the most dangerous reading of all.

Nothing here is a recommendation to buy or sell any security, and no figure in this piece should be taken as a current quote for any company. It is a method for reading numbers that are already public, applied on the same schedule to the ones that flatter and the ones that do not.