August 30, 2026

What to record, and the fields that mislead

Every guide says keep a log. Almost none says which fields, and the obvious ones produce a document that looks like a record of success while the account disagrees.

Why this strategy needs one more than most

Premium selling produces a long series of small wins interrupted rarely by a large loss. That pattern is uniquely hostile to memory. The wins are frequent, similar and forgettable; the losses are few enough to feel like exceptions rather than data.

So the impression formed by living through it is systematically more favourable than the outcome, and the gap can persist for years. A log is the only thing that closes it.

The field that flatters you

Cumulative premium collected. It only goes up, it's the easiest number to track, and it's the headline on most logs.

It's also close to meaningless on its own. Premium collected doesn't net out buybacks, doesn't include the mark on stock acquired through assignment, and doesn't account for the capital that was committed to earn it. A log reading $2,400 collected can sit alongside a $1,900 unrealized loss on shares that can't be called away, and nothing in the number hints at it.

If one number has to be the headline, make it total return on capital committed, measured against what that capital would have done in the obvious alternative — usually just holding the index.

Entry fields

• Date, underlying, strike, expiration, contracts
• Credit received, per contract and total
• Days to expiration at entry
• Delta at entry — the comparable number across different stocks and strikes
• IV rank at entry — whether this was expensive or cheap for this underlying
• Capital committed — strike × 100 × contracts for a cash-secured put, not the buying power figure
• One line on why this strike. Written before the outcome is known.

That last field is the one people skip and the one that eventually earns the log its keep. Six months later it's the only way to distinguish a thesis that was sound and unlucky from one that was never sound, and hindsight cannot reconstruct it.

Exit fields

• Date and how it ended: expired, closed, assigned, rolled
• Debit paid, if closed
• Realized result on the option
• Days held, so return can be annualized honestly

And for assignments, the field that matters most: the position doesn't end at assignment. Carry it forward as an open stock position with its adjusted basis, and the trade stays open until those shares are gone. A log that closes the row at assignment records the premium as a win and drops the consequence.

Recording a roll

Two entries, never one. Close the original trade at its actual result including the loss, then open the new one at the credit received.

As the rolling piece covers, the single net-credit line is how four consecutive losses become four recorded wins. The two-entry version drops your win rate to the true one, which is the entire point of keeping the record.

Reviewing it

The log's value is in what it reveals across many rows, and the questions worth asking are usually the uncomfortable ones.

• What's the largest single loss, and what was the size at entry?
• Did contract count rise after wins or after losses?
• What fraction of positions were on correlated underlyings at the same time?
• Do the assigned positions, tracked to their eventual exit, show a different picture than the option results alone?
• Against holding the index over the same period with the same capital, where does this actually land?

That last one is the one most logs are structured to avoid, and it's the only one that answers whether the strategy is doing anything.

The format doesn't matter

A spreadsheet is fine. A text file is fine. Dated entries in a repository are fine. The constraints that matter are that it's written at entry rather than reconstructed, that rolls appear twice, and that assignments stay open until the stock is sold.

Anything satisfying those three will outperform any tool that doesn't.