Post-Mortem and Performance Marketing, Side by Side

A post-mortem reconstructs what was known, what was assumed and what was decided at the moment a position was taken. Performance marketing puts a result on display so as to attract. The two need opposite inputs. A post-mortem needs the reasoning and can be finished with the result line empty. Performance marketing needs the result and can be finished with no reasoning at all.

No figure below was read off a market. The reference asset, its price, the two-legged arrangement and the one margin percentage all come from teaching rather than from a screen. No position described below carries a result at all, and that absence is exactly the condition under which reconstructing a decision without its result can be shown.

Here is the situation this is built for. Somebody hands an analyst a review of a position that was taken a while back. The review is careful. The sums tie. The writing reads like an honest account. And it is still impossible to tell, from reading it, whether the document is a reconstruction of a decision or an advertisement for the person who took it. Both kinds of writing look backwards at the same position, so neither can be recognised by its subject, by its tense or by its tone. Every instinct a reader reaches for first is the wrong one.

The separation turns out to rest on what each document is made of, not on how it sounds. Ask what the writer could not have finished without. A post-mortemA reconstruction of a decision as it stood at the moment it was taken, written from the position of somebody who does not yet know how it went. could not have been finished without the reasoning, and it can be handed over with the result line still blank. Performance marketingA display of a result already achieved, put up in order to attract the person reading it. could not have been finished without the result, and it can be handed over with no reasoning in it whatsoever. The two needs are the whole comparison, and what follows is the two definitions in full, the test run on a real position, and what happens when one line is struck out of each.

What is a post-mortem on a position, and what does one actually produce?

A post-mortem reconstructs a decision as it stood at the moment it was taken. Five things go into it: what was known, what was computed from what was known, what was assumed, what was checked, and what was not checked. The list is short on purpose. A reconstruction that grows extra rows starts collecting justifications rather than inputs.

Everything else hangs off the awkward phrase in that definition, as it stood, so the phrase is worth defending rather than skating over. A post-mortem is written from the position of somebody who does not yet know how things went, even when the writer knows perfectly well. Writing from that position is an act of deliberate forgetting, and it is difficult in a way that nothing else about the exercise is difficult. The information to be left out is sitting in the writer's head throughout, and it is the most vivid thing in there.

Think of a cook who has just taken a loaf out of the oven. There are two things she could write in her notebook. She could write that the bread was good. Or she could write that the oven was set to a particular heat, that the dough had rested for a particular length of time, that she had assumed the flour was the same batch as last week, and that she had not checked whether the kitchen was colder than usual. Only the second note is worth anything the next time she bakes, and notice that she could have written every word of it before the loaf came out. The second note is a post-mortem, and it did not need the loaf.

A post-mortem produces, in the reader's own terms, a list of the assumptions that turned out to be carrying weight. The list of assumptions is what transfers. Everything else in a reconstruction is a record of one occasion, but an assumption nobody knew they were making is a hole to be walked into again, in a completely different position, with completely different figures. The output of a post-mortem is not a verdict; it is a list of load-bearing assumptions, and that list is the only part that travels to the next decision.

One thing a post-mortem is not. A reconstruction is not an assessment of whether the decision was clever, and it is not a scorecard. A reconstruction that ends by grading somebody has stopped reconstructing and started judging. The grade will almost always be borrowed from how things turned out, and that borrowing is the failure named below as resulting.

What is performance marketing, and how is one recognised that does not look like an advertisement?

Performance marketing displays a result in order to attract, and its real subject is the person or the service that produced the result rather than the position itself. The last clause is the definition doing its work. A piece of performance marketing looks as though it is about a contract, a strategy or a set of positions. The subject is none of those. The subject is who was standing behind them.

Most readers can spot the loud version. A number in a large typeface, a name under it, and something to click. The hard version is the quiet one, and it is quiet in a specific and recognisable way. The recognition cue is structural rather than a matter of tone: any reasoning that appears is present to explain why the result should be believed, not to be examined. Reasoning in a post-mortem is there to be picked apart. Picking it apart is how the load-bearing assumptions get found, and the writer has an interest in the reader doing it. Reasoning in performance marketing is there as support for a number that has already been stated, and it has finished doing its job the moment the number is accepted.

Now the case that catches most people, and it catches them because they are sorting on the wrong axis. Accuracy is not the axis this comparison runs on, so a document can be entirely accurate, carefully arithmetical, free of every exaggeration and still be performance marketing. Every figure can be correct. Every base can be named. Every period can be stated. None of that moves a document across the line. The line runs between documents by what each one is built out of, not by whether the building is sound.

A wedding caterer shows photographs of eleven weddings. Nothing in the photographs is false. The food was that colour, the tables were laid that way, and the guests were that many. The photographs do not contain how many staff were rostered, what the kitchen assumed about the delivery arriving on time, or which of the eleven nearly went wrong and why. The invitation is to trust the caterer, and the invitation is made by showing outputs. The photographs are accurate and they are also not a review of anything.

What single test separates the two, and how is it run in ten seconds?

The test is to ask what each piece of writing could not have been finished without. The test is the whole of it, and it takes one line to state and about ten seconds to run on any document.

The test runs in two passes. The two passes ask opposite questions, and both answers are wanted. First, the result is covered and what remains is read. If something still reads, and that something states what was known and what was assumed, the document is a reconstruction. Second, the reasoning is covered and what remains is read. If something still reads, and that something is a figure with a name under it, the document is a display. Covering the result and seeing whether anything remains, then covering the reasoning and asking the same question, gives two answers that together settle which of the two documents it is.

Notice what the test never asks. The test does not ask whether the writing is honest, whether the arithmetic is right, whether the author is qualified or whether the tone is measured. A careful, honest, well-sourced piece of performance marketing is still performance marketing, and a scrappy, badly typed reconstruction is still a reconstruction. The two needs are opposite rather than merely different. One is therefore not a degraded version of the other, and neither can be improved into the other by being made more careful.

There is an everyday version of this that runs without thinking. When a friend describes a journey they took, it is usually clear within a sentence whether they are explaining how to get there or reporting that they got there. The route and the arrival are different subjects. One of them survives being told by somebody who has not yet arrived.

What does a reconstruction record, line by line, on a real position?

Abstract definitions are cheap. Here is the test run on the position this record actually holds, with every figure multiplied out rather than merely quoted.

The position is a long forward on the invented reference asset struck at Rs 2,130.00/-. Before the five lines, one piece of housekeeping. Two figures below both read Rs 2,000.00/-, and a reader is entitled to wonder whether one was copied into the other. The two were not copied. Rs 2,000.00/- is the spot price of the invented reference asset, and Rs 2,000.00/- is also the strike of the call that appears later. The two agree because that call is struck at the money. Struck at the money means the strike and the spot price are the same number. Two different quantities, one value, and the agreement is the definition rather than a coincidence.

1. What was knownThe spot price of the invented reference asset was Rs 2,000.00/-. Financing cost 6.50 per cent a year. The asset pays nothing at all while it is held. The absence of any payout matters more than it looks, and it gets its own line below.
2. What was computedRs 2,000.00/- carried for one year at 6.50 per cent a year: Rs 2,000.00/- times 1.065 is Rs 2,130.00/-. The carryThe cost of holding the underlying from today until the agreed date, which here is the financing on the money borrowed to buy it. is Rs 130.00/-, and it is the entire distance between the spot price and the forward price.
3. What was assumedThat nothing would be paid out on the asset for the whole of the year. Read that again as a sentence about the future rather than as a fact. A claim about the year ahead is exactly what it is. The no-payout line is a load-bearing assumptionOne the decision would have come out differently without. Remove it and the arithmetic underneath changes.: a payout would subtract from the carry, and a large enough one would put the forward price below the spot price rather than above it.
4. What was checkedThe carry check was run before the figure was written down anywhere. Rs 130.00/- accounted for the whole of the gap, so the forward price of Rs 2,130.00/- was recorded as a cost of holding rather than as a view about where the price was heading. All four checks were run and all four findings are printed in the next section.
5. What was not checkedNo likelihood was attached to any price at any point. None was available to attach. The not-checked line is where a reconstruction earns its keep, and it is the line a reader should look at first in somebody else's, including in their own from last year.

Putting the result first quietly selects which reasoning gets written down, so the order of those five lines is not decoration. Starting with what was known puts the rows in the order they existed. Starting with what happened means every row afterwards is chosen for how well it explains the opening. A document built that way carries the same headings and a different subject.

What do the four checks find when they are run as the content of a reconstruction?

A reconstruction with no list of things to look at drifts into being a story about how somebody felt on the day. The four checks handed over earlier in this sequence stop that drift. Each one is a question with an answer rather than an impression. Here they are run against the record, with what each one found printed rather than described.

The carry check asks whether a forward price was read as arithmetic or as a view. Finding: Rs 2,000.00/- carried for one year at 6.50 per cent a year is Rs 2,130.00/-, and the carry of Rs 130.00/- is the entire gap, so there is no room left over for anybody's opinion about direction. Nothing in the figure reports what anybody thinks, so the identical forward price would arrive if every party involved were certain the price was about to fall. Recorded as a cost.

The what-moves check asks of every headline quantity what sum would actually change hands in the period, and then divides. Finding: on the invented two-legged arrangement the notionalThe amount the payments under an arrangement are multiplied by. It is a multiplier, and it does not change hands. is Rs 1,000 crore, the fixed payer owes 7.20 per cent a year and the floating payer owes a benchmark reading 6.00 per cent a year for the first period, so the gross legs are Rs 72.00 crore and Rs 60.00 crore and only the net of Rs 12.00 crore moves. Rs 1,000 crore divided by Rs 12.00 crore is 83.33 times. Rs 12.00 crore divided by Rs 1,000 crore is 1.2 per cent. The other 98.8 per cent of the headline stays exactly where it was. Recorded with the net named beside the notional.

The paid-for check asks of every number on a payoff drawing whether it knows what was paid. Finding: the call is struck at Rs 2,000.00/- and cost a premium of Rs 180.00/-. At a price of Rs 2,130.00/- the payoffWhat a contract delivers at expiry, ignoring what was paid for it. is Rs 130.00/-, and the profitThe payoff net of what was paid for the position, which is a different number from the payoff and often a different sign. at that same price is Rs 130.00/- less Rs 180.00/-, or minus Rs 50.00/-. The breakeven priceThe price at which the profit on a position reaches nil, here the strike plus the premium paid. is Rs 2,000.00/- plus Rs 180.00/-, or Rs 2,180.00/-. Recorded as a payoff and a profit, separately, with the label on each.

The posted check asks that every move be restated as a share of what was posted, with the base said out loud. Finding: with an initial margin of 8.0 per cent of the exposure, a teaching figure that no authority set, Rs 2,000.00/- of exposure stands on Rs 160.00/- posted. A 4.0 per cent adverse move on that exposure is Rs 80.00/-, and Rs 80.00/- is 50.0 per cent of the Rs 160.00/- posted. Rs 2,000.00/- of exposure against Rs 160.00/- posted makes the exposure 12.50 times what was posted. Either base alone is a half-sentence, so both are recorded in one sentence.

Four checks, four questions, and not one of them needed to know how anything turned out. The rest of this comparison is built on that observation. Every finding above was computable on the afternoon the position was taken.

One more thing belongs beside every check rather than after all of them, so it goes here rather than in a note at the foot. Running a check catches a misreading and changes nothing else. It does not alter what the position obliges, it does not reduce what is owed at any price, and it is not a reason to hold anything. The Rs 80.00/- still moves against the Rs 160.00/- posted. The profit at a price of Rs 2,130.00/- is still minus Rs 50.00/-. A reader who runs all four is better at reading and no better protected, and those are genuinely different things.

The paid-for finding is the one readers most often write down wrongly, and the wrongness is invisible, so it deserves a drawing of its own. Two numbers describe the same position at the same price. One of them knows what was paid and the other does not, and a reconstruction that records only the first has recorded the wrong thing.

Why does the reconstruction still stand when the result line is struck out?

Now the part this comparison exists for, and it starts with an absence rather than with an argument. No position in this record holds a result, none at all, so the line where a result would sit has to be printed as a labelled blank. No outcome is attached to it and no history stands behind it. No probability and no distribution are attached either, so not one line states even what was likely at the time. No volatility figure is attached either, so the call premium of Rs 180.00/- is given rather than rebuilt from one.

So the row is printed empty, and then a line is drawn through it. A visible removal teaches more than an absence merely reported. The next step is counter-intuitive enough that it is better looked at than described. Every one of the five reconstruction lines and every one of the four check findings is still complete with that line struck out. The spot price is still Rs 2,000.00/-. The multiplication still gives Rs 2,130.00/-. The carry of Rs 130.00/- is still the whole of the gap. The assumption about nothing being paid out is still load-bearing. The four findings still read exactly as they read before. Nothing above the struck line was reading from it.

The mirror of that is the second half of the demonstration, and it has to be run rather than asserted, so the figure below prints both documents with the same line struck through in each. Remove the result from a piece of performance marketing and there is nothing left to display. The result was the subject rather than the evidence. The strike leaves a description of a service and a claim on the reader's trust, and neither of those is about the position at all.

Two consequences follow, and both are worth carrying away whatever anybody does with derivatives. The first is a rule that can be pointed at any review, in any subject: if it falls apart when the result is removed, it was never a review of the decision. Not a poor one, not a partial one. Not one at all. The review was organised around something that was never the decision.

The second is the practical one, and it is the reason this is worth the trouble: a post-mortem can be written the same afternoon a position is taken, before anything whatsoever has happened. That sounds like a trick and it is not. Everything a reconstruction records was in the room at the moment of the decision. Writing it then, rather than later, is also the only way to be certain that no result got into it. At that point there is no result to get in.

There is one more thing this demonstration says, and it is exact rather than rhetorical. Since no position in this record has a result, no real example of performance marketing can be built from it, and the inability to produce an example of a thing states something precise about what that thing is made of. The specimen on the right of the figure above is a shape with nothing in it. The empty shape is the finding.

What is resulting, and why does it survive inside careful writing?

Judging a decision by how it turned out rather than by what was known when it was taken is called resultingJudging a decision by how it turned out rather than by what was known at the moment it was taken., and the term is Duke's, in Thinking in Bets, 2018. Naming it matters more than it usually does with a piece of vocabulary. The error is almost impossible to see from inside, and a name gives the reader something to look for.

Here is why it survives so comfortably inside work that is careful in every other respect. The result is the only part of the whole story that is unambiguous, so it becomes the organising principle of a review by default, and nobody ever has to decide to let it. Everything else in a reconstruction is contested or approximate. Memory supplies what was known. Interpretation supplies what was assumed. Record keeping supplies what was checked. The result is a fact with an edge on it, and writing gravitates to whatever is solid.

What does a post-mortem not do, and who is it unfair to?

A reader who has just watched a reconstruction run without a result will reasonably overestimate what it delivers, so this belongs here. A post-mortem improves the account that can be given of a decision. The improved account is the whole of the claim. A post-mortem does not establish whether the decision suited the person who took it, it does not change what any position obliges, and it is not a form of protection against anything at all.

The last of those is worth being blunt about. Nothing in a written reconstruction reduces what is owed, moves a price, or alters what a contract requires on any day. The habit is a reading habit. The habit makes a person harder to mislead, including by themselves, and that is a real thing to be, and it is not the same thing as being safer.

Then the finding that matters most and gets written about least. Somebody whose position went against them did not necessarily decide badly, and somebody whose position went their way did not necessarily decide well. The pairing is the same asymmetry from the other side, and it is the whole reason resulting is worth the effort of avoiding. A person who lost money on something they did not fully understand has not failed at anything. Such a person was sold a document whose result line was the subject. The document was designed so that the reasoning could not be examined, and there was none in it to examine anyway. Reading is a skill and the absence of a skill is not a character fault.

How does somebody who never writes a review use any of this?

Most people who need this comparison will never publish a line about a contract. Most people are on the receiving end, and the receiving end is where the sorting actually pays.

The household comes first, and it matters most while getting the least attention. A message arrives from somebody plausible about an arrangement or a contract, and it carries a figure and a name and an invitation. Running one covering needs no knowledge of derivatives at all. The figure is covered and what is left is read. If what is left is the name and the invitation, the kind of document is settled, and it was settled in about eight seconds without evaluating a single claim in it. The covering is not scepticism and it is not cynicism. The covering is a sorting step, and it leaves the reader free to take the thing seriously afterwards if it survives.

A lender reads a borrower's own account of a position the same way, and for a specific reason. The position itself is already written down somewhere. The lender is pricing the quality of the process that produced it. A borrower who hands over a reconstruction with a load-bearing assumption named in it has shown the lender something the figures cannot show. A borrower who hands over a set of results has handed over something the lender already has access to.

An analyst uses the comparison as a filter on incoming writing rather than on positions. Anything whose reasoning is present only to support a figure goes in a different pile from anything whose reasoning is present to be argued with, and the two piles get read with different levels of attention because they are answering different questions.

And somebody running a small operation of any kind, a shop, a workshop, a two-person consultancy, gets the most transferable version of all of it. The assumptions are written down before the outcome is known. A list made afterwards is the list that explains what happened rather than the list that produced it. That habit costs one sheet of a notebook and it works on a supplier decision, a hiring decision and a lease exactly as it works on a forward.

Who sets the things not printed here?

Three requirements are touched by everything above, and the authority, the Securities and Exchange Board of India (SEBI), is named for each of them and the value for none of them. A fourth row underneath holds the one invented percentage used here so that it can never be mistaken for something an authority set.

The value cell in each of the first three rows is empty. Each of those requirements is set by the authority printed inside its row, each of them moves, and a figure written out here would be wrong rather than merely old. Wrong is worse than absent. Absent sends the reader to sebi.gov.in and wrong does not. The fourth row is the mirror: 8.0 per cent of the exposure is printed there because it came from nowhere, is used here for teaching, and gives the Rs 160.00/- that the posted check divided by.

Should a reader who can run a reconstruction enter into one of these contracts?

The question arrives at this point, so it gets answered at this point rather than dodged into a line at the foot. A reconstruction of a decision is not a verdict on whether that decision suited the person who took it. A general account of how these instruments work cannot answer whether anybody should hold one.

The useful part of that refusal is naming what would have to be known first. A refusal with nothing after it is just a shrug.

None of the first four is held here. This record contains no likelihood attached to any price, no spread drawn around any figure, and no series or trend for anything. The fifth belongs to SEBI at sebi.gov.in, and it is named rather than stated because it moves.

A post-mortem improves the next decision, and it does not make one for anybody. Those two clauses are the whole honest position, and the second one is not a hedge attached to the first. The second clause is the shape of the thing.

The reference asset, the option pair, the forward, the two-legged arrangement, both of its sides and both specimen documents are invented.
Educational material. Not advice on any investment, tax, budget or market position.