Passing a prop firm evaluation with one model

Most evaluation failures are rule failures, not edge failures

Traders fail prop firm evaluations for three structural reasons more often than they fail for a lack of edge. The account hits a drawdown limit. The account does not meet a minimum number of trading days. Or the account trips a consistency rule that caps how large any single winning day can be as a share of total profit. None of those three outcomes requires you to be unprofitable. They require you to trade in a way the rules do not allow.

I run one model on NQ and MNQ. Daily bias locked before the New York open. Entries only inside 10:00–14:00 ET. A valid setup needs an hourly manipulation and a 5-minute confirmation. Stop sits one tick beyond the 5-minute manipulation extreme. Target is a static 2R. The day stops after +2R or after 2 losses. If the manipulation never prints, there is no trade. That structure is not designed around evaluation rules. It happens to fit them. This post explains why.

The three rules that stop already-profitable traders

Drawdown rules

A drawdown rule sets a maximum loss from a peak or from the starting balance. Cross it and the evaluation ends. The number varies by firm. The shape of the problem does not.

Discretionary trading produces uneven daily risk. One day you risk 0.5R. The next day you press because the open looks clean and you risk 2R across three attempts. A third day you revenge-trade after a scratch. The equity curve has sharp down days even when the week finishes green. Drawdown rules punish the shape of the path, not only the final P&L.

A hard daily stop changes the path. My day ends at +2R or at 2 losses. Two full losses is a defined hit. It cannot expand into a third and fourth attempt because the session "still looks good." The maximum damage in a single session is capped before the open. That does not make the model profitable. It makes the downside per day measurable, which is what a drawdown rule is measuring.

Minimum trading days

Many evaluations require a minimum number of days with at least one trade. The intent is to stop a trader from passing on one outsized session and sitting flat for the rest of the period. The side effect is pressure to trade on days that do not meet the setup.

That pressure is where process breaks. You force an entry without the hourly manipulation. You widen the stop. You take a second attempt after the daily stop should have already ended the session. Minimum trading days do not require bad trades. They create an incentive for them if your model has no clear definition of a no-trade day.

In this model a no-trade day is a normal outcome. If price never sweeps a prior reference low or high by at least one tick and closes back through it inside the window, there is no entry. Tuesday 4 August 2026 was one of those days. Bias was locked LONG at 10:02 ET off the daily swing low at 27201.50. The London low at 29007.50 and the 25% retracement near 29067 were the deeper references. On MNQ the 09:00 candle swept the 08:00 low by 1.5 points and closed back through it. On NQ the same candle missed the 08:00 low by 3.25 points. No valid entry was taken on the session. The day finished without a trade. Under a minimum-days rule that session does not advance the day count. It also does not damage the account. Protecting the account matters more than padding the day count on a day the condition is absent.

Consistency rules

A consistency rule limits how large your best winning day can be as a share of total profit over the evaluation. The firm is checking that the result was not produced by one session.

Here is the arithmetic in plain form. Suppose the evaluation needs $3,000 in profit to pass. Suppose the consistency rule says no single day may exceed 40% of total profit. Then no day may contribute more than $1,200 of that $3,000. If you make $1,800 in one session and $1,200 across the rest, you are at $3,000 total but your largest day is 60% of profit. You have not passed. You must keep trading until total profit rises enough that $1,800 becomes 40% or less of the new total. That means total profit must reach at least $4,500 before that $1,800 day fits under a 40% cap.

Work it the other direction. You want to know the maximum day size that still fits a 40% rule at a $3,000 target. Maximum day = 0.40 × $3,000 = $1,200. Any day above $1,200 forces you to grow the denominator. The rule is not vague. It is a ratio. Largest winning day ÷ total profit must stay at or under the stated share.

Discretionary targets make that ratio hard to control. A runner that you manage by feel can turn into 4R or 5R on a trend day. That feels efficient in a personal account. Inside a consistency rule it can force extra days of grinding just to dilute one large winner. A static 2R target caps the size of a winning day before the trade is open. With a hard daily stop at +2R, the best day the model can post is +2R from the plan. You can still have a green day smaller than +2R if you only take one partial path to target, but you cannot blow past the daily cap by holding for more. That keeps the numerator of the consistency ratio bounded.

Why one model fits the rules better than discretion

Discretion has three problems inside an evaluation.

First, risk per trade changes with confidence. Confidence is not a number you can audit at the end of the day. A defined stop at the 5-minute manipulation extreme, one tick beyond, makes risk a function of structure. R is the distance from entry to that stop. Target is 2 × that distance. Every trade uses the same R logic.

Second, the exit is a feeling. Feelings do not respect consistency caps. A fixed 2R target removes the decision to hold. You will leave money on the table on trend days. That is a real cost. It is also the cost of keeping daily profit inside a band the rules can accept. I accept that trade for consistency. The week of 28 July is a small illustration, not a proof: 3 trades taken, 3 sessions with no entry, results +2.0R, +1.67R, and +1.14R, net +4.81R. Three trades is a small sample and proves nothing on its own. What it shows in structure is a set of capped outcomes rather than one large day and two scratches.

Third, the daily stop is optional when you are discretionary. Optional stops get ignored after a loss. A rule that the day ends at +2R or after 2 losses is not optional inside the model. That is the mechanism that keeps a bad morning from becoming a drawdown event.

One model also means one instrument family and one session window. I profile NQ and MNQ. Session boundary is 18:00 ET. Bias is built top down — daily, then 7-hour, then 4-hour — and locked at 10:00 ET. Entries only between 10:00 and 14:00 ET. That removes overnight noise and removes the urge to "make back" a London move that was never part of the plan. Evaluation rules do not grade your overnight opinions. They grade the path of the account during the period.

Why no-trade days help you on an evaluation

A no-trade day feels like lost progress when a calendar is counting down. It is not lost progress if the alternative is a forced loss.

Every forced trade has a full unit of risk attached. Two forced trades in a week can be 2R of damage that the good days then have to recover before any net progress counts. A no-trade day risks 0R. Under a drawdown rule, 0R is the best available outcome on a day without a setup. Under a consistency rule, a no-trade day does not inflate the largest-day ratio because it adds nothing to the numerator. Under a minimum-days rule, a no-trade day does not advance the counter — that is the only cost. The cost of a bad forced day is larger.

The model already defines when to stand down. No hourly manipulation means no trade. Manipulation that prints on MNQ but misses on NQ, as it did on 4 August, is a reminder to define the reference level on the instrument you are actually trading and to write it down before the window. If the condition is not met on that instrument, you do not enter. That decision is available every session. On an evaluation it is a protective decision, not a passive one.

Repetition matters more than prediction here. You are not trying to forecast how many trading days the month will give you. You are running the same checklist and accepting the count that the market produces. Process over feelings. The evaluation is a constrained version of the same process, not a different sport.

Fixed targets, hard stops, and the path of the equity curve

Put the pieces next to the three rules.

Drawdown rule: daily loss capped at 2 losses. Single-trade risk defined by the 5-minute extreme. Path volatility is reduced because the left tail of each day is bounded.
Minimum trading days: you cannot manufacture valid days. You can only avoid invalid ones. The way to finish the day count is to take the setups that meet the definition inside 10:00–14:00 ET, not to lower the definition.
Consistency rule: daily profit capped at +2R. Largest day is bounded by the model rather than by how far a trend runs. The ratio of largest day to total profit stays controllable as total profit builds.

None of that guarantees a pass. Accounts fail for many reasons, including a cold streak that hits drawdown inside the rules. The point is narrower. A discretionary approach creates extra ways to fail that have nothing to do with whether the underlying read is sound. One model with a fixed target and a hard daily stop removes those extra ways.

What the week of 28 July does and does not show

During the week of 28 July I took 3 trades and had 3 sessions with no entry. The three results were +2.0R, +1.67R, and +1.14R. Net +4.81R. Three trades is a small sample and proves nothing on its own. I am not using it as evidence that the model passes evaluations. I am using it to show what a week looks like when the daily cap and the no-trade rule are both active: multiple sessions at zero, no day above +2R, and a net that is the sum of bounded outcomes.

If a consistency rule were applied to a week with that shape, no single day would dominate the total. If a drawdown rule were applied, the losing days would have been cut at the second loss rather than at an open-ended bottom. That is the fit. The fit is structural.

What to do tomorrow

Write down the evaluation rules on one page before the open: drawdown limit, minimum trading days, consistency percentage, and profit target. Compute the maximum allowed winning day in dollars from the consistency percentage and the profit target. Keep that number next to the screen.
Run the bias the same way you would outside the evaluation. Daily, then 7-hour, then 4-hour. Lock it at 10:00 ET. Do not revise it.
Pre-define the reference levels for the instrument you will trade — previous day high and low, session highs and lows, prior swing points. Write the prices down. Do not rely on memory when the 10:00–14:00 ET window opens.
Take a trade only if the hourly manipulation prints and the 5-minute confirmation triggers. Stop one tick beyond the 5-minute extreme. Target 2R. Stop the day at +2R or at 2 losses.
If the manipulation does not print, log a no-trade day and shut the platform. Do not add a second model to satisfy a day count.
At the close, update three numbers only: days traded, largest winning day as a share of total profit, and distance to the drawdown limit. Those three numbers are the evaluation. Everything else is noise.

Consistency comes from repetition and a defined process — not prediction. The evaluation is just the same process with tighter constraints on the path.