The three cases that built the engine: profitable, and getting worse.

Case one

Profitable while drawdown expands.

The account is up over the period and the depth of its excursions is increasing. Both facts are true simultaneously and only one of them is visible on the equity curve at a glance. What this pattern usually describes is a system whose wins have grown or clustered while its losing sequences have lengthened — the net result holds, and the path to it has become materially harder to sit through. The structural risk is that the profitability is what keeps the operator in a position whose drawdown profile has changed underneath them, and a drawdown profile that has been quietly expanding is the one most likely to eventually exceed what the operator can hold. Profit is not evidence against this. It is what makes it survivable long enough to become serious.

Case two

Strong profit factor with inefficient risk deployment.

This is the case the contradiction engine is most directly built for, and it is the one that maps onto profit-without-efficiency. Profit factor is a ratio of gross profit to gross loss and it says nothing whatsoever about how much risk was deployed to produce either. A branch can post a genuinely strong profit factor while consuming far more risk per unit of result than the plan intends — the wins are large relative to the losses, and both are large relative to what should have been committed. Raw profit factor has no way of seeing this, which is precisely why the engine reads risk-adjusted profit factor against risk-adjusted efficiency rather than either raw figure. Two ratios that both look at profit and disagree about whether it was earned are more informative than either alone.

• Profit factor is blind to deployment. A metric with no risk denominator cannot report over-deployment.
• Variance-driven gains produce the same reading as edge-driven gains, and only one of them repeats.
• The pattern often coincides with aggression rather than skill, which is why it tends to precede a bad month rather than follow one.

Case three

Equity growing while acceleration slows.

The ACCEL pipeline exists for this one. Compounding velocity is the second derivative of the equity curve, and it can be falling while the first derivative remains firmly positive — the account is still growing and growing less quickly each period. On a chart of equity this is close to invisible, because the line continues rising and the human eye reads direction far more readily than curvature. It matters because compounding is the mechanism the whole system is built to exploit: a plan that assumes geometric growth and receives decelerating linear growth will hit its milestones late in a way that is not obvious until the shortfall accumulates. Deceleration is also the earliest of the three cases to appear, which makes it the most valuable and the easiest to dismiss.

What unites them

In all three, the reassuring number is the one with no risk in its denominator.

Net profit, profit factor, and equity level share a property: none of them carries any information about what was committed to produce the result. That is not a flaw in those measures — they answer the question they were built for — but it does mean that a system monitored exclusively through them is monitored through instruments that are structurally incapable of reporting deterioration in risk quality. Every metric the SDE adds puts risk into the denominator, and every contradiction it detects is ultimately a disagreement between a reading that accounts for risk and one that does not. The three founding cases are not three separate problems. They are one blind spot observed from three angles.

The key idea

The dangerous period is the one where the surface still reads fine.

A system that is losing money announces itself and gets attention. A system that is still profitable while its structure erodes gets the benefit of the doubt for as long as the profit lasts, which is exactly the window in which the erosion is cheapest to correct and least likely to be looked at. The whole structural layer exists to make that window visible, and the contradiction engine is the part of it that watches two instruments argue about whether the profit is real.