Security Selection: Choosing Inside a Fixed Sleeve

Security selection chooses which holdings fill a sleeve whose size was already fixed by the allocation. The Anantara Multi-Asset Portfolio's equity sleeve is Rs 300 crore of a Rs 500 crore mandate, so a point earned inside that sleeve reaches the portfolio as 0.60 points. Selection decides what sits in the box; it never decides how big the box is.

Most writing about choosing holdings starts with the holding. The room the holding must fit into was measured, argued over and signed off before anybody looked at a single name. The dimensions of that room put a hard ceiling on what any choice inside it can be worth. The ceiling is arithmetic, and arithmetic does not care how good the choosing was.

The worked object throughout is the Anantara Multi-Asset Portfolio, an invented discretionary mandate of Rs 500 crore run by Faiz Ahmad Ansari for an invented charitable endowment whose investment committee is chaired by Rukmini Deshpande. Its policy shape is equity 60.0 per cent at Rs 300 crore, fixed income 30.0 per cent at Rs 150 crore and cash 10.0 per cent at Rs 50 crore. Every figure belongs to one stated twelve month period, and one period is far too short to be evidence about what any approach achieves.

What is security selection, and what has already been decided before it starts?

Security selectionChoosing which individual holdings go into a portfolio, once it has already been decided how much money each asset class gets. is the act of deciding which individual holdings occupy a given pot of money. Selection is not the act of deciding how large that pot is. In the Anantara Multi-Asset Portfolio the pot is called the equity sleeveThe part of a portfolio set aside for one asset class. Saying sleeve keeps a share of that part from being read as a share of the whole portfolio., it holds Rs 300 crore, and by the time Faiz Ahmad Ansari looked at his first candidate name that Rs 300 crore was already a settled number written into the papers of the mandate.

Two earlier decisions had closed before selection opened. The mandate fixed the constraints: equity must sit between 50 and 70 per cent of the portfolio, no single holding may exceed 5 per cent of the portfolio, nothing unlisted, and the fixed income sleeve carries a stated minimum credit standing. The allocation then chose 60.0 per cent equity inside that band. The choice turned a range written in the papers into a single rupee number. The ceiling on what selection can contribute to the whole portfolio was therefore set by other people, at an earlier meeting, and it is arithmetic rather than a matter of opinion.

Think of a wedding caterer given a hall, a date and a budget of Rs 8,00,000/-. The caterer chooses every dish, and those choices decide whether the evening is remembered. The caterer cannot make the hall bigger or move the date. A brilliant menu inside a small hall feeds the number of people the hall holds. A free menu inside a fixed hall is the shape of every selection decision below.

Why does the same holding have two different weights, and which base is right?

One rule holds up everything that follows: a weight is a share of something, and the something has to be named. The largest holding in the Anantara Multi-Asset Portfolio is Rs 23 crore. Against the Rs 500 crore portfolio that is 4.6 per cent. Against the Rs 300 crore equity sleeve the identical holding is 7.67 per cent. Neither number is a rounding of the other, and neither is wrong.

The reason is that a weight is a division, and a division needs a denominator. The baseThe total a share is measured against. Change the total and the same rupee amount becomes a different percentage, even though nothing about the holding has moved. is that denominator. Rs 23 crore over Rs 500 crore gives 4.6 per cent; Rs 23 crore over Rs 300 crore gives 7.67 per cent. Since the sleeve is smaller than the portfolio, a share of the sleeve is always the larger number, by exactly the factor 500 over 300. A weight quoted without its base is a fraction with the bottom half missing. Such a figure is not weak, it is unreadable.

Now the part that decides whether a rule has been broken. The single holding capA stated ceiling on how much of a portfolio any one holding may be. The ceiling is written against a named total, and the name matters as much as the number. in this mandate is written against the portfolio: no holding above 5 per cent of Rs 500 crore. Five per cent of Rs 500 crore is Rs 25 crore. The largest holding at Rs 23 crore is therefore inside it, with Rs 2 crore of room to spare. Read the 7.67 per cent figure against a 5 per cent cap and a breach seems to appear. The comparison sets a share of one thing beside a limit on another, and it answers nothing.

Later stages multiply by this share, so the rounding convention matters here too. The record rounds the sleeve figure to 7.7 per cent when it simply names it. Any arithmetic that multiplies by the sleeve share uses 7.67, not 7.7, and that rule is followed everywhere below.

How security selection fits into portfolio construction: where in the order does it sit?

Five stages run in order, and each one hands the next a smaller room than it was given. The mandate sets the constraints. The allocation sets the sleeve sizes inside those constraints. Selection fills the sleeves. Implementation gets the chosen holdings into the portfolio at some cost. Monitoring reports what happened afterwards. The Anantara Multi-Asset Portfolio moved through all five before the stated twelve month period closed.

The order is not a filing convention but a chain of nested permissions. The mandate permits equity between 50 and 70 per cent; the allocation picks 60.0 per cent inside that; selection then works inside the Rs 300 crore that 60.0 per cent produced. Because each stage inherits a smaller room, the same manager can be entirely right about a holding and still be immaterial to the portfolio, and that is a statement about the room rather than about the judgement.

The same nesting explains why any account of investing that begins with the holding deserves suspicion. Beginning there skips the two stages that set the size of every consequence, and skipping them is exactly why the choosing of holdings comes after the mandate and the allocation rather than before them.

Asset allocation vs security selection: which of the two can move the base?

Comparing the two on importance produces an argument nobody wins. Compare them on what each can move instead, and the difference stops being a matter of taste. Asset allocation changes how much money sits in each class. Security selection changes what sits inside one class. Only the first of those changes the base that every selection result is later multiplied by.

Put a number on it. The Anantara Multi-Asset Portfolio's equity sleeve is Rs 300 crore of Rs 500 crore, so the multiplier from the sleeve to the whole is 0.60. A gain of one percentage point inside the equity sleeve arrives at the portfolio as 0.60 points. Two points inside the sleeve arrive as 1.2 points. Allocation sets that multiplier and selection cannot touch it. The two decisions differ in kind, not in importance.

The direction of the arithmetic trips people up, so it is worth stating twice. Going from a sleeve figure to a portfolio figure, multiply by 0.60. Going the other way, from a portfolio figure back to a sleeve figure, divide by 0.60. Multiplying where the arithmetic calls for dividing understates a sleeve result by a factor of nearly three, and it happens in committee papers more often than anybody admits.

What is the allocation effect, and what is it measured against?

Once a stated period closes, the gap between what the portfolio returned and what its benchmark returned can be split into parts. The excess returnThe difference between what a portfolio returned and what its stated benchmark returned over the same period. An excess return says nothing on its own until it is split apart. for the Anantara Multi-Asset Portfolio over the stated twelve month period was plus 1.6 percentage points gross, being 14.2 per cent against the composite benchmark's 12.6 per cent, with the risk-free rate at 6.5 per cent for the same period. The word gross is load bearing: what costs did to that figure belongs to implementation and is covered separately.

The allocation effectThe part of a return gap that came from holding different amounts in each asset class than the benchmark held, rather than from holding different things inside a class. is the part of that gap that came from holding different class weights than the benchmark held. The composite benchmark is 60 per cent a broad equity index and 40 per cent a broad bond index, both deliberately unnamed here. The portfolio held 60.0 per cent equity, 30.0 per cent fixed income and 10.0 per cent cash. So the equity weights matched, the fixed income weight was ten points lower and the cash weight was ten points higher. An allocation effect measures exactly that difference in shape.

The record gives plus 0.35 points for the stated twelve month period. On Rs 500 crore that is Rs 1,75,00,000/-. An allocation effect is defined entirely by the benchmark's weights, so quoting one without naming the benchmark and its weights reports a difference from nothing in particular. The composite is therefore described in full every time it appears, even though it is never named.

What is the selection effect, and what does it complete?

The selection effectThe part of a return gap that came from holding different individual securities inside a class than the benchmark held, with the class weights held aside. is the other part: what came from holding different things inside a class rather than different amounts of each class. For the Anantara Multi-Asset Portfolio over the stated twelve month period the record gives plus 1.25 points. On Rs 500 crore that is Rs 6,25,00,000/-.

Run the check that every attribution needs. Plus 0.35 and plus 1.25 sum to plus 1.60, the whole gross excess return. In rupees, Rs 1,75,00,000/- plus Rs 6,25,00,000/- is Rs 8,00,00,000/-, and Rs 8,00,00,000/- is 1.6 per cent of Rs 500 crore. A split that does not add back to the number it split has not been checked, and a reader who never sees the sum has not been shown one either.

One warning matters more than either figure. The record splits the same gross 1.6 points a second way, into about 0.49 points that came from carrying more market exposure than the benchmark and about 1.11 points of everything else. The second split asks a different question on a different basis. Both pairs sum to 1.60. Neither pair is the true one. A term from the first pair may never be set beside a term from the second. The resulting number describes nothing at all.

What is fundamental investing when it is treated as a way of filling a sleeve?

Fundamental investingFilling a sleeve by forming a view on each business one at a time, so every holding is there because of a reason specific to it rather than because of a rule applied across many. is one way of answering the selection question. Holdings enter the sleeve because somebody formed a view on each business separately, one at a time, and judged that view strong enough to fund. The Anantara Multi-Asset Portfolio's equity sleeve was filled this way across 28 names.

How such a view is formed is a separate subject. How a business is researched, how a valuation is built and what makes one company more attractive than another are each covered under their own subjects. The portfolio question is not whether the view is right, it is how a right view is converted into a position of a particular size inside a room of a particular width.

Held to that line, fundamental investing is a filling rule with three properties worth noticing. Each holding needs its own justification, and a justification is expensive in time. The number of holdings is therefore bounded by how much research capacity exists. And because each holding is argued individually, nothing in the method itself forces anybody to check what the 28 names have in common. Checking for a shared condition is a separate job, belonging to the risk and monitoring stages rather than to selection.

How position sizing changes portfolio outcomes: what is the same 20 per cent move worth?

Position sizingDeciding how many rupees go into a chosen holding. Sizing is a separate decision from choosing the holding, and it decides how much the choice is worth. is the second half of every selection decision and it is routinely treated as an afterthought. Choosing the holding decides whether the move is favourable. Sizing the holding decides how much the move is worth. The Anantara Multi-Asset Portfolio makes the point cleanly because two of its position sizes are locked in the record.

An equal weightThe size every holding would have if the money in a sleeve were divided evenly between all of them. An equal weight is a measuring stick for reading actual sizes rather than a target. holding in this sleeve is Rs 300 crore divided by 28, or Rs 10.71 crore. Rs 10.71 crore is 2.14 per cent of the Rs 500 crore portfolio and 3.57 per cent of the Rs 300 crore equity sleeve. The largest holding is Rs 23 crore, about 2.15 times the equal weight.

Now run the same 20 per cent gain through both. On Rs 23 crore a 20 per cent gain is Rs 4,60,00,000/-, or 0.92 points of the portfolio and 1.53 points of the sleeve. On Rs 10.71 crore the identical 20 per cent gain is about Rs 2.14 crore, or 0.43 points of the portfolio. Being right about the holding and being right about its size are two separate decisions, and here the second one scales the first by more than two times from an identical move.

What does a cap on any single holding actually prevent?

Less than most readers assume, and the arithmetic settles it in one division. The cap in the Anantara mandate is 5 per cent of the portfolio, or Rs 25 crore. Divide the Rs 300 crore equity sleeve by Rs 25 crore and the answer is twelve. Twelve holdings, each sitting exactly at the cap, would fill the entire equity sleeve and breach nothing at all: not the cap, not the 50 to 70 per cent equity band, not the listing requirement.

The sleeve of the Anantara Multi-Asset Portfolio holds 28 names instead. The spread in this sleeve is therefore something the manager chose, not something the cap produced, and a reader who treats a cap as evidence of spread has read a ceiling as though it were a floor. A ceiling says how far one thing may go. A floor would say how many things there must be, and this mandate states no such thing.

The cap does do something, and being precise about it is worth a moment. The cap stops any single holding from moving the portfolio by more than a bounded amount, and the bound is computable: at Rs 25 crore, a holding falling by half costs 2.5 points of the portfolio. The cap leaves entirely alone how many names there are and whether those names rest on the same underlying condition. No per name limit can see that condition.

Why does a top ten share need a floor before it can be read?

Concentration is usually reported as the share of a sleeve held by its largest handful of names, and that share is almost always read as though any number were possible. It is not. If a sleeve holds 28 names, the ten largest of them cannot possibly hold less than ten twenty-eighths of it, and ten twenty-eighths is 35.71 per cent. The ten largest can never be smaller than ten average ones. The floor is made of arithmetic, and no amount of careful spreading gets below it.

So a reported top ten share only becomes informative once it is set beside that floor. A figure of 40 per cent against a floor of 35.71 per cent describes a sleeve that is close to evenly spread. A figure of 70 per cent against the same floor describes something quite different. A top ten share below its own arithmetic floor is not a low concentration reading, it is an impossible one, and the usual cause is that somebody changed the base without saying so.

The hazard is not hypothetical. The base error and the floor error are the same error wearing two hats: a share computed against the portfolio and then reported as a share of the sleeve will read far too low, and the floor is the cheapest test that catches it. One division, done in a second, before any discussion of whether the concentration is comfortable.

What does the record refuse to say about the selection effect?

Here is where the worked instance ends honestly rather than tidily. The selection effect for the Anantara Multi-Asset Portfolio is plus 1.25 points of the whole Rs 500 crore portfolio, or Rs 6,25,00,000/-. A natural next question is how much of that arose inside the equity sleeve. The multiplier is sitting right there, so the temptation is to divide 1.25 by 0.60 and announce 2.08 points inside the sleeve.

The division is valid only if the entire selection effect arose in the equity sleeve, and nothing in this record says that it did. The fixed income sleeve holds securities too, and a selection effect can arise there. The word if is doing all the work in that sentence, and where a record is silent the honest report states the conditional and stops, rather than converting silence into a number.

So the finished worked instance reads like this. The sleeve is Rs 300 crore of Rs 500 crore, or 60.0 per cent. An equal weight holding is Rs 10.71 crore, being 2.14 per cent of the portfolio and 3.57 per cent of the sleeve. The largest is Rs 23 crore, being 4.6 per cent of the portfolio and 7.67 per cent of the sleeve, about 2.15 times an equal weight, with Rs 2 crore of room under the Rs 25 crore cap. The gross excess return was 1.6 points, split by the record into 0.35 of allocation and 1.25 of selection. Where inside the portfolio the 1.25 arose is not supplied.

How does anybody use this in a committee room, on a Tuesday?

An investment committee like Rukmini Deshpande's uses it as a reading habit rather than a calculation. Every weight that reaches the papers is read with the question what is this a share of, and a weight arriving without an answer goes back. The habit stops the two most expensive mistakes in this whole area. The first is comparing a limit with a measurement on a different base. The second is reporting a concentration figure that could not exist.

An analyst reviewing a manager's record uses the multiplier the same way. A sleeve number and a portfolio number differ by a factor that has nothing to do with the manager. Told that the equity holdings did well, the analyst converts the sleeve figure to what the whole portfolio received before treating it as a result. A lender assessing a borrower whose assets sit in a managed portfolio does the reverse conversion to work out what a fall in one sleeve does to the total.

A household does exactly the same work with a pen, and the arithmetic is identical at every scale. Suppose savings of Rs 20,00,000/-, of which Rs 8,00,000/- sits in the equity part, and one holding is Rs 1,00,000/-. The Rs 1,00,000/- holding is 5.0 per cent of all savings and 12.5 per cent of the equity part. Both numbers describe the same rupees. A household that reads only the second will think it is more than twice as exposed as it is. A household that reads only the first will underestimate what a bad year in that holding does to the equity part.

What is security selection not?

Security selection is not the whole of investing, and treating it as the whole is the single most common distortion in popular writing about portfolios. Selection is one stage of five, working inside a room whose width was set at the previous stage. Selection is not the allocation, which chose that width. Selection is not implementation either, and what the chosen holdings cost to acquire is set out under implementation. And it is not research: forming a view on a business is a different craft, covered separately.

Selection is exactly this: the decision about which holdings and at what size, taken against a fixed total. Both halves of that decision matter, and the second half, the size, is the one that converts a correct view into a number the portfolio can actually feel. Everything else in this guide is arithmetic that makes those two halves visible.