Siyabulela Nomoyi, Quantitative Portfolio Manager at Satrix
An index fund promises to deliver the same return as its benchmark, such as the FTSE/JSE All Share or Capped All Share index. It sounds simple: buy every share in the index in the same proportion and hold on. In practice, it takes skill and care. The goal is not to copy the index at any cost. It is to match the index’s return as closely and as cheaply as possible, without exposing investors to avoidable losses.
Tracker funds never match their indices perfectly, and there’s a good reason for that. An index is a mathematical model. It pays no trading costs, holds no cash, reinvests dividends immediately, and rebalances at closing prices without any liquidity constraints. A real fund can’t do any of that, so the fund and the index often end up with slightly different returns. That gap is called tracking error.
Tracking error comes from several places. Including client flows (handling money flowing in and out of the fund), index rebalances and corporate actions, the timing of dividends, transaction costs (which always work against the fund), and the fund’s own structure – which includes optimisation and cash drag (the small amount of cash a fund must hold).
Trading links all of these, because every decision the manager makes most likely turns into a trade, and every trade costs money. Good index management aims to shrink the total gap between the fund and the index, rather than fixing one part at the expense of another.
Large indices like the FTSE/JSE All Share have a long “tail” of small companies. The index holds 121 stocks, but the 50 largest make up about 90% of it, while the 50 smallest make up just 4.5%. Many of those small stocks are rarely traded, have wide gaps between buying and selling prices (bid-offer spreads), and can jump sharply on modest trading.
Optimisation is the solution. Using a risk model, a manager can hold a more liquid (easier to trade) group of shares, weighted so the fund behaves very much like the index. The first few holdings cut tracking error sharply. Deep in the tail, however, each extra stock adds almost nothing while costing a great deal to trade, and takes longer to buy or sell.
This is not stock picking. The optimiser doesn’t judge whether a small company is good or bad. It answers a risk question: which mix of holdings gives the index’s risk profile at the lowest cost? The fund stays passive in intent and precise in outcome. It also busts a common myth, that an index tracker must buy every new company added to the index.
At a quarterly index review, every tracker fund, derivative and structured product linked to the index may need to buy a newly added share at the same moment, in the closing auction, in a stock that normally trades only a fraction of that volume. Fair price discovery tends to break down at these moments. When a thinly traded stock is suddenly in demand because of its index inclusion, the auction price reflects forced buying rather than real value. Traders who see this demand coming buy in advance and sell into the auction.
Two recent examples show what this looks like. In December 2025, ASP Isotopes (ISO) entered the All Share and rose 163% in the closing auction on the day of rebalancing. It then fell 56% next trading day. In the recent September index review, Stadio (SDO) was added. It rose 23% in the auction and gave back 9% the next trading day.
Neither move had anything to do with how the businesses were doing. A manager who buys at any price in that auction isn’t tracking the index efficiently. They pay a premium to whoever sold into the demand, and their own buying pushes the price up even further.
Satrix reviews every company that enters or leaves the index before each rebalance and uses its own judgement to manage risk. This is vital, given the scale of the assets it manages. Part of that risk management is looking at how easily a stock can be traded, and whether Satrix’s own buying would end up being the market for that stock.
Satrix also weighs the actual tracking-error cost of holding less than the index weight, holding none at all, or building the position gradually. For a small tail stock, the tracking-error cost of being patient is usually a fraction of the cost of buying into a distorted auction. This approach applies to every mandate, from institutional portfolios to the CIS range, and follows all board notice requirements.
The result: Satrix has not held ISO since its inclusion in December 2025, and SDO is still under review. Over the 12 months to the end of August 2026, the Satrix Capped All Share ETF has been managed with a very low tracking error.
Good index tracking is not blind copying. A manager who chases exact replication at any price may look accurate on rebalance day but deliver a worse outcome afterwards. A manager who understands liquidity, price discovery and the true size of tracking-error risk stays very close to the index and keeps more of the return for investors.
Ed’s note: Earn a verified CPD hour by learning more about active and passive investing in CPD Bundle 19 on EBnet.Edu here.
ENDS
