Ravi had been trading for three years when he finally admitted something to himself: he knew his profit and loss to the rupee, but he could not describe his own operation. Ask him his win rate by setup, and he would guess. Ask whether he actually followed his stop-loss rules last month, and he would say “mostly,” which is a feeling, not a fact. What turned his trading around was not a new indicator. It was learning to audit your trading operation the way a business audits itself — scoring his process against evidence, fixing the weakest parts on a schedule, and scoring again months later to see if anything truly changed. This is the step-by-step version of what he did, and what any serious retail trader can do in a single weekend plus six disciplined months.

Step 1: Accept that you are a business, then audit like one

A trading desk — even a desk that is one laptop and one chai — has operations: research routines, entry rules, risk controls, record-keeping, review habits. Businesses in other fields stopped trusting gut feel about operations long ago. Software companies, to take the most developed example, score themselves with structured instruments like the delivery maturity assessment from SCMGalaxy, whose questionnaire runs a hundred questions deep and covers defined territories of engineering practice — from how security is governed to how AI-assisted development is supervised. Two design choices make it work, and Ravi stole both. The answers must come from records, not memory. And the arithmetic is fixed in advance, so the score cannot wobble with the scorer’s mood. Hold those two rules and everything below follows.

Step 2: Write your question set — twenty is enough

Ravi’s first draft had five vague questions. His second had twenty sharp ones, four in each of five areas:

• Preparation. Did a written plan exist before the market opened? Did it name levels, position sizes, and invalidation points — or just hopes?
• Risk control. Was every position sized by a rule? Was any stop-loss moved after entry, and how many times?
• Execution. How many trades matched a setup named in the plan? How many were improvised chases?
• Record-keeping. Does each trade have an entry in the journal with reason, size, exit, and outcome? Written the same day?
• Review. When did you last read your own journal? Did any rule actually change because of it?

Notice every question can be answered from artefacts: the journal, the broker statement, the plan file’s timestamps. If a question cannot be evidenced, rewrite it until it can. And if the artefact does not exist — no journal, no written plans — that is not a scoring problem; that is finding number one.

Step 3: Score one honest month

Pick the most recent full month. For each question: 0 if the practice is absent, 1 if it happened irregularly, 2 if the records show it happened consistently. Forty points possible. Ravi scored 17, and the shape of the 17 taught him more than the total. Preparation and execution were respectable; risk control was a horror show — his broker statement showed four positions where the stop had been widened mid-trade, something he would have sworn he did “almost never.” The statement does not negotiate. That is the entire point of auditing from records.

If the exercise stings, good; it is supposed to sting exactly once. Every re-audit after this one is a comparison, and comparisons can only flatter you if the work got done.

A warning about the two honesty traps. The first is grading on intention — “I meant to journal that trade” quietly becoming a 1. Intentions score zero; only artefacts score. The second is sample-shopping: picking your best recent month because the worst one “wasn’t typical.” Ravi nearly did this, then realised the untypical month was precisely the one his process was built to survive. If you have a trusted trading friend, swap audits — each of you scores the other’s records against the rubric. Nothing sharpens a question set faster than watching someone else apply it to you without mercy, and the mild embarrassment is far cheaper than the self-deception it replaces.

Step 4: Turn the three worst answers into a dated plan

Resist fixing everything. Take the three lowest items and assign each a horizon — fixes dated for the coming month, the coming quarter, and the coming half-year, matched to how structural the change is. Ravi’s plan looked like this. Month one: a hard rule that stop-losses are entered with the order and never modified, enforced by writing the stop in the plan before entry. Quarter: rebuild the journal into a format that takes ninety seconds per trade, because the elaborate template he had abandoned was designed for a version of himself that did not exist. Half-year: a full review cycle — monthly journal reads with one written rule change each, feeding a quarterly look at performance by setup.

Dated items with owners — even when every owner is you — behave differently from resolutions. They either close or they stare at you. For the discipline mechanics themselves, our staged guide to taking command of your own trades pairs naturally with this step, and if you suspect your weak spots are the common ones, the list of daily errors that keep tripping up intraday traders is a ready-made question bank for your next audit round.

Step 5: Re-audit on a date you set now

Before closing the spreadsheet, Ravi put a calendar entry six months out: same twenty questions, same scoring rubric, next full month of records. This step is what distinguishes an audit from a diary entry. The second score passes judgement on the plan itself — Ravi’s came back at 29 of 40, with risk control transformed and review still lagging, which seeded the next cycle’s plan. Professional tooling institutionalises exactly this loop; the SCMGalaxy OS platform that engineering teams use treats re-assessment and progress tracking as core features, not afterthoughts, because a maturity score’s value lives almost entirely in its comparison with the previous one. A trader’s version needs no platform. It needs the same calendar honesty.

Step 6: Read your scores the way professionals read theirs

Two reading rules, borrowed again from the software world. First, never let a decent total hide a dreadful area — the sample scorecard shows how an organisation can post a comfortable overall number while one domain trails embarrassingly, and trader scorecards have identical blind spots: beautiful journals, reckless sizing. Fix domains, not totals. Second, expect the trend, not the level, to carry the meaning. A 22 that was 17 last cycle is a better desk than a 28 that was 31. What you are tracking is whether your operation improves under deliberate pressure, which is a different question from whether this quarter’s market was kind.

There is a psychological dividend too. Traders who audit their process stop outsourcing self-worth to the P&L, because they finally own a measure of skill that market noise cannot fake or destroy. That shift — judging yourself on executable process rather than outcome — is the same territory mapped in our piece on the hidden mental habits of consistently successful traders. The audit is simply that mindset with a scoresheet attached.

The whole method on one page

Write twenty evidence-answerable questions across preparation, risk, execution, records, and review. Score one month from artefacts, 0-1-2. Take the three worst findings and stagger their fixes: one due in thirty days, one inside the quarter, one by month six. Re-score on a pre-committed date. Compare domains, not totals, and let each cycle seed the next. Ravi’s desk is not a different desk because he found a better strategy. It is a different desk because, twice a year, it gets inspected by its least forgiving client — the evidence — and the evidence keeps signing off.