Some MORE comments about index construction. 1) Generically...
Baufinanciaphaster 👹@bauhiniacapital
15 views
Jan 07, 2025
~10 min read
Advertisement
1
Some MORE comments about index construction.
1) Generically speaking, there are VERY few large US stocks which issue large amounts of new equity in order to buy something. It happens from time to time, but it is reasonably rare.
When a company like MSTR issues new shares to
1) Generically speaking, there are VERY few large US stocks which issue large amounts of new equity in order to buy something. It happens from time to time, but it is reasonably rare.
When a company like MSTR issues new shares to
2
buy BTC, it gets a little complicated. The MSTR issuance was not a large block. It was an at-the-money offering which means the brokers doing the offering sell shares bit by bit in the market. At a certain point, the share count of newly issued shares will become 10% of TSO.
3
At that point, NDX would create the immediate index upweight. Until the 10% mark, the change in share count will be pushed back until the quarterly review. The same thing goes for S&P500. For MSCI, it is 5% and more than US$1bn for an ad hoc upweight. For FTSE the numbers are
4
slightly different. Each index has their own treatment for equity offerings which deliver newly-issued shares to investors.
2) Lots of companies buy back shares. When that happens, the share count falls. Company-specific Free Float coefficients (FFW, FIF, IW, FAF... every index
2) Lots of companies buy back shares. When that happens, the share count falls. Company-specific Free Float coefficients (FFW, FIF, IW, FAF... every index
5
has its own acronym) are adjusted at quarterly or annual reviews, or not, depending on index rules. TOPIX in Japan adjusts share count on a monthly basis, with newly-issued shares added at the time of delivery, convertible bond conversion share count added on the second to
6
last trading day of the month after the change (ditto for options exercise, and cancellation of treasury shares), but FFW is *currently* adjusted once a year, except in *exceptionally* rare ad hoc changes. There are other oddities to do with large cancellations.
For MSCI, FFW
For MSCI, FFW
7
changes can happen immediately if there is a large secondary share sale by a non-float holder (a sale of say 7% of the company by the founder who owned 35% before the sale), but would not happen at all at the time, or even at the next quarter if the same seller sold 2% and the
8
raw FFW did not move past a 5% threshold (MSCI rebalances FFW based on 5% tranches for most developed market stocks so unless it is a big enough change, it won't be reflected at all; FTSE on the other hand will trigger a FFW change on ad hoc events where the precision gets to a
9
ridiculously precise level (one recent sample (I've changed a few digits (but not digit count/precision) to anonymise it) case; "increased investability weighting from 51.38184518227% to 63.2706539711248%."
But generically, if a company buys back its shares, that means the stock
But generically, if a company buys back its shares, that means the stock
10
price might rise as net income/share count (i.e. EPS) goes up, but passive trackers will sell that stock when the Index Provider tells its index data subscribers Index Share count drops.
3) Generally speaking, indices have a selection universe, which is then limited by
3) Generally speaking, indices have a selection universe, which is then limited by
11
factors such as liquidity, profitability (if not profitable, it gets removed from the universe), float weight minima, trading suspensions, significant volatility events, etc to remove it from the eligible constituents list. One reconstitutions/rebals, the index provider will
12
decide if something ELIGIBLE but not an index member will replace a current index member which has fallen too far from grace.
There is an index in Taiwan which has 50 members. If an existing member gets ranked 66 or lower, it gets the boot. If a non-index member ranks 35 or
There is an index in Taiwan which has 50 members. If an existing member gets ranked 66 or lower, it gets the boot. If a non-index member ranks 35 or
13
higher, after meeting all the various eligibility requirements, it gets added automatically. And if there is no member below 65, they take the lowest-ranking member and delete it.
Another index does similar but rather than cardinal ranking, it uses a 90%/110% representation
Another index does similar but rather than cardinal ranking, it uses a 90%/110% representation
14
threshold for each sector (because for that index, sector representation is a thing). If a sector suddenly becomes under-represented because eligible but not current-member stocks in a given sector increase in value, skewing that sector vs others, then the stock closest to the
15
representation border might get added.
MOST indices with the highest AUM tracking as a percentage of index market cap are straight non-capped indices (TOPIX, S&P500, S&P/ASX 200, etc). That means non-members close to the border which subsequently significantly outperform the
MOST indices with the highest AUM tracking as a percentage of index market cap are straight non-capped indices (TOPIX, S&P500, S&P/ASX 200, etc). That means non-members close to the border which subsequently significantly outperform the
16
index and/or other stocks which are in the index but close to the border will generally get added if they go up enough. Stocks which disappoint, and fall below the border far enough, will generally get deleted.
4) Some index families are 'tranched', some cumulative.
The S&P500
4) Some index families are 'tranched', some cumulative.
The S&P500
17
is big and famous and there are a couple of changes at the end of the week. TPL was added late last month when Marathon Oil was given the boot because it was taken over. The % of float shares of APO and WDAY to be bought this Friday is higher than the TPL event in Nov24 because
18
the S&P family is a bit weird and fun. When TPL was added to S&P500, it was deleted from S&P400. And it was deleted from S&P Completion Index. So there were a bunch of sellers and a bunch of buyers and there was net buying.
APO and WDAY will see the same gross buying (as % of
APO and WDAY will see the same gross buying (as % of
19
Index Shares) but not the same gross selling. They are in fewer S&P indices.
AMTM will get deleted from S&P500 to be demoted not to S&P MidCap 400 but it will jump a level lower to S&P SmallCap 600. But that means it will also be added to S&P Completion Index.
AMTM will get deleted from S&P500 to be demoted not to S&P MidCap 400 but it will jump a level lower to S&P SmallCap 600. But that means it will also be added to S&P Completion Index.
20
The FTSE 100 and 250 are stacked. A stock cannot be a member of both. So if it goes from 250 to 100, there could actually be net selling, depending on tracking AUM as a % of Index Market Cap. If it goes from Small to 350 (which is the total of 100 and 250), it will see buying.
21
S&P/ASX is stacked cumulatively. There is money which tracks S&PASX300. Other money which tracks S&PASX200. Some money which tracks S&PASX100, then some which tracks 50 and 20.
The rules here are a bit weird because 300 is reconstituted once a year and theoretically, a stock
The rules here are a bit weird because 300 is reconstituted once a year and theoretically, a stock
22
going into 200 has to be a member of 300 as it goes in. But if it is in 200, it is in both 200 and 300. If in 20, it is in 20, 50, 100, 200, 300, and All-Share. So there is incremental promotion/deletion rather than complete promotion/deletion as with FTSE UK 100/250.
23
5) Some indices are equal-weighted, cap-weighted, fundamentally-weighted, dividend-weighted, or other and these are MUCH more fun from a professional's perspective.
On equal-weighted indices like the S&P500 Equal-Weighted Index, with 500 members, each member will be rebalanced
On equal-weighted indices like the S&P500 Equal-Weighted Index, with 500 members, each member will be rebalanced
24
to be 0.2% of the portfolio once a quarter. If a stock outperforms the index, it will get trimmed down. If it underperforms so the pre-rebal weight is 0.15%, then index trackers will buy 33% of the share count already owned.
Obviously, that 0.2% of total tracking AUM is
Obviously, that 0.2% of total tracking AUM is
25
larger for the smallcaps than for the large caps. If tracking is $100bn. each stock gets US$200mm. For Apple, that means nothing. But for the smallest cap in the S&P500 which is staying in this time ($FMC), $200mm is 3% of float. In that case, the holding is 3.2x ADV and a big
26
move might mean a significant flow as a percentage of ADV.
If it falls, there's more to buy. If it rises, there is some to sell.
HOWEVER, it is at the border. It is currently probably near the top of the list of potential S&P500 deletes for March 2025. IF it were to get
If it falls, there's more to buy. If it rises, there is some to sell.
HOWEVER, it is at the border. It is currently probably near the top of the list of potential S&P500 deletes for March 2025. IF it were to get
27
deleted from S&P500, it would get deleted from S&P500 Equal-Weight, which would be a greater percentage of shares than just S&P.
Stocks at the borders are fun. Equal-weighted indices are fun.
For dividend-weighted indices, there are different ways to construct an index but
Stocks at the borders are fun. Equal-weighted indices are fun.
For dividend-weighted indices, there are different ways to construct an index but
28
if an index is TRULY dividend weighted, it will generally favor small caps MUCH more than large caps. Fundamentally-weighted indices can get even weirder. If one takes a large index such as S&P500 and then were to create an index which weighted each company by revenue, operating
29
profit, and net profit over the past three years, and then rebalance quarterly, the rebalance would favour low multiple stocks which underperform high multiple stocks which outperform. So in late 2021, that would have been like capping tech and buying value. In Q1-Q3 2022, as
30
high multiple stocks fell sharply, they would have seen incremental upweights from super low to slightly less low weights, and as they fell more, that would have continued.
Structurally, these types of indices are noise-harvesting indices (at rebalances, one sells a bit of
Structurally, these types of indices are noise-harvesting indices (at rebalances, one sells a bit of
31
those stocks which have outperformed and buys those which have underperformed; and when stocks are added or deleted, it can depend). For div yield-weighted indices, if a stock goes up too much, the yield will go too low and the stock will get kicked out. If a stock falls, it will
32
get added to an index. Unless it gets kicked out of the eligibility universe by falling TOO far. These are fun too. For professional traders in index rebals, these are fun because they offset momentum bias seen in regular large-cap indices which promote based on a stock going up.
33
6) Price-weighted indices create their own problems. And they are fun. There are relatively few of these. The two most famous ones are the Dow Jones Industrials and the Nikkei 225.
The Nikkei 225 is interesting because it was famous, there are well-known futures and options on
The Nikkei 225 is interesting because it was famous, there are well-known futures and options on
34
the index, it is generally quite correlated to the large domestic index which tracks large caps (TOPIX) and also to MSCI Japan (which is the Japan index many other managers might benchmark to) so while it is never used as a hedge by LOs, it is an index used by structured product
35
issuers/sellers, by macro traders, etc.
The concept of a price-weighted index is that one owns exactly the same number of shares of every stock in the index.
If index trackers own 25,000,000 shares of everything, if a company has 100 million shares out, the index trackers
The concept of a price-weighted index is that one owns exactly the same number of shares of every stock in the index.
If index trackers own 25,000,000 shares of everything, if a company has 100 million shares out, the index trackers
36
will own 25% of the shares out. That "works" except for a stock like Fast Retailing where the family owns 40+% of the company, it means Nikkei 225 trackers would own 25% of the remaining 60% or 40% of the "Float". But remember that other passive trackers own probably 20-25% of
37
that 60% which means in the end, actual float not held by the family or passive trackers is only 10-15% of shares out. And that makes something which trades a lot "because large cap" squeezy, which makes it look more liquid than it actually is.
And because the Nikkei made a
And because the Nikkei made a
38
rule change for the Nikkei 225 two years ago, there is capping. Inclusions may or may not be capped on the way in, either for price/weight or liquidity. Exclusions are not mitigated on the way out. And stocks over 10% are capped by a 0.1x reduction in their base coefficient
39
(which may be different based on historical stock splits, etc, but not issuance/buybacks/etc) until it gets below 10%. For Fast Retailing, right now that means an expected $6bn selldown at end-March.
Fun all around.
Fun all around.
40
Lastly, I’d like to thank the young, delightfully fetching, and obviously genetically-gifted Mary, Yolanda, Kinutta, Worthy, Angela, Mary, Mary, Angela, Abbey, and Mary - all of zero followers/following for zeroing in on index construction content for a first foray on Twitter.
41
@AB32432219 They were famous for a minute.
