What the Book Clears

Eleven bars, one book

A bar is not the same thing as a measurement. Lesson 71 measured a correlation of 0.9464 and lesson 76 counted 113 orders in a year; those are inputs, and a book cannot pass or fail them. A bar is a number with a side: clear it and you may proceed, miss it and you may not. Eleven lessons state one. Six others hand the book something it needs in order to be judged at all.

The book under test is the one module 9 licensed and lessons 79 and 80 priced: two moving-average rules, a 2-bar fast average against a 5 and a 3 against a 10, at one and a half per cent a trade under a three per cent daily-loss limit, carrying 1.06 independent bets. The faster of the two is lesson 63’s winning cell, so this is not a hypothetical book. It is the one the course actually produced.

Read the verdict column first and then read what earned it. Every one of the four passes is a pass by omission or by exactness. The filter condition is cleared by not running a filter, which is free. The chain is cleared by a resting stop, which lesson 77 measured as free. The position cap is cleared at exactly two out of two, which is a pass with no room in it. Only lesson 66 is cleared on evidence, and the evidence is seven trades.

Then read the failures, because they are not near misses. Lesson 67 asks for 589 trades against 7. Lesson 65 asks for a live record and there is none. Lesson 79’s figure is over a hundred per cent of the account before the overhead is added and 170.3 after. These are not numbers a better month improves.

The single quantity underneath three of the five failures is the same one, and it is worth putting in its own table. Every bar that names a count of trades is asking for evidence, and the book has one record of evidence.

The middle column is the one to sit with. Running four rules rather than one saves twenty-four trades, which at the measured pace is exactly ten years rather than 10.42, and every argument module 9 had about correlation was an argument about those five months. The last column says what the record is against what each bar asks: the book has a twenty-second of what lesson 65 wants and a hundred-and-fiftieth of what lesson 80 wants.

Almost nobody has laid their own thresholds out in one column and put their own record beside them. It is a page of a notebook and an afternoon, and it produces the only number that ranks what to do next, which is the largest gap.

Module 10’s list of what the day contains that is not the decision is now complete, at five items: the twelve operations, the chain, the rate, the wage and the overhead. Every one of them is a subtraction, and the list has no other kind of entry on it.

Where it stops

Take the eleven in the order the course produced them and stop at the first failure, because that is what a threshold is for.

Lesson 64 is first, and it fails. The winner’s t of 3.65 clears the raised bar of 3.54, and then the eligibility count that no backtest prints moves the same cell from one in 147 to a coin flip.

Everything after lesson 64 was computed on an edge of a tenth of an R, which lesson 65 introduced as a hypothesis in order to have something to compute with.

So ten of the eleven bars were being cleared or missed by a number that was assumed, not measured.

That is the finding, and it is not a complaint about the course. It is what the course was built to make visible. The tenth of an R was put in at lesson 65 in the open, named as a hypothesis on the page where it entered, and carried in the open through sixteen lessons. Any reader who wants a different verdict supplies a different edge, and every figure on this page moves with it.

This course never had an edge to give you. What it has is the eleven places where you can find out whether you do.

Which is why the honest use of the table is as an order of work rather than as a scoreboard. Four of the eleven you already clear and three of those cost nothing, so they are not where your afternoon goes. Two cannot be answered until you name an instrument and a jurisdiction, and naming them is an hour. The five failures rank themselves by the last column of the second table: the live record at 22.3 times what you have is the one that is reachable, and the 589 trades at 84.1 times is the one that is not, and knowing which is which is the difference between a plan and a wish.

So write the eleven bars down the left of a page, put your own number beside each, and start on the smallest gap that is not already closed.

What this does not settle

That eleven is the right count. It is the number of lessons in this course that state a bar a book can be held to, and the boundary is arguable: lessons 69, 71, 72, 73, 74 and 76 produce measurements the book uses rather than bars it must clear, and a reader who promotes any of them gets a different denominator. What that reader does not get is a different numerator. The five failures are the same five however the set is drawn, because none of the six borderline lessons is a failure for this book.

That the verdicts are close calls. Four of the five failures are not close. Seven trades against 589 is a factor of 84, and 101.6 per cent of an account is not a near miss on a hundred. The one that is genuinely arguable is lesson 64, where the published t clears the published bar and the failure arrives only once the eligibility count is set, which is a rule this course states and no backtesting package enforces.

That the two unanswered bars might rescue it. They cannot, because both can only move in one direction. Naming an instrument tells you the size at which lesson 68’s capacity binds, and it either binds or it does not; it never adds trades. Naming a jurisdiction tells you how much of lesson 78’s relief you get, and full relief leaves the verdict where it was while partial relief moves it further away. Two unanswered questions are two ways to be worse off and no way to be better.

That this settles whether an individual can trade profitably. It settles what one book, built from sixty closes and a 253-cell search, does against eleven bars this course wrote. That book was chosen so that a reader with a spreadsheet could reproduce every figure in it, and that choice is exactly what guaranteed a seven-trade record. A programme running thousands of bars across many instruments starts at a different place on the first two bars and at the same place on the last four, because lessons 79 and 80 are arithmetic on a pace and a wage and do not care how the rule was found.

And the concession that costs most: this page has graded a book and not taught you to build a better one, and there is no lesson after it that does. Eighty-one lessons have produced eleven ways to find out that something does not work and not one procedure that produces something that does. That asymmetry is deliberate and it is also a real limit: a course made only of falsification hands you a very good filter and no candidates to put through it. Module 11 is where the candidates come from, and it starts where this book is weakest. This one carries 1.06 independent bets on a single instrument, which lesson 74 showed is the number a family of long-or-flat rules collapses to, and lesson 82 asks what a second instrument does to it.