Wintermute is the leading crypto market maker providing liquidity across most vetted trading platforms.

Not so brief post on ETFs in TradFi

I will not go too much into the basics, assuming that anyone interested in this subject already knows what an ETF is. Let’s just be clear that the ETF market is huge (Source), it is growing consistently in AUM and as a percentage of AUM invested:

Published by Statista Research Department, Feb 19, 2021

Mutual funds on the way out

The “killer” features of ETFs — the driving forces behind ETF growth — are low cost and ease of purchase. In short, mutual funds are expensive and replicating ETFs yourself is hard.

Buying the S&P500 Index all by yourself is a pretty annoying ordeal, starting from execution, followed by quarterly rebalancing, and finally handling all kinds of corporate actions (dividends, M&A, divestments etc). Before the dawn of ETFs this problem has been “solved” by people allocating money to mutual funds, but those had (and still have) very hefty management fees and very little execution transparency. If you are still banking with an old school bank and have a meaningful balance, your bank would very likely approach you proposing to manage your money. Please don’t do that. 😊 Buy some ETFs instead — management fees on the most basic indices like S&P or Nasdaq will be very close to zero, or sometimes even negative, meaning that you get paid to hold these indices. Low and negative management fees are in turn coming from fund managers discovering that they can achieve an extra yield by lending out stocks in the basket to market makers and short sellers. This lending has been without doubt a very lucrative proposition for ETF fund managers, as they have full discretion over how much of this extra yield they share with the ETF holders and how much they keep to themselves.

Keep in mind, though, that the ETF issuers’ mandate is following the underlying index as closely as possible and they are not getting paid to outperform it. I have had a few bizarre conversations with ETF issuers, pointing out various ways that they could be making money for the ETF holders, and they turned them all down as it was way outside of their mandates — any deviation from the underlying index was bad, whether negative or positive.

Ways to structure ETFs

Most ETF issuers have a very clear mandate, covered by a certification, which in turn makes them investable by a wider range of institutions. The most common way for an issuer to set up a fund is to track a particular index (SPX, DAX, MSCI World etc). That way, any decision making with regard to building the underlying basket is delegated to an index provider (S&P, MSCI etc), and the issuer can focus on distribution and the operational side of the business.

There are number of ways to structure your ETF to track a certain index or commodity:

• Physical replication. Most of the stock index ETFs in the US are physically replicated, meaning that you can (most of the time) exchange a share of the ETF for a basket of the underlying securities (or a claim to Gold in the underground vault in case of physical Gold ETFs).
• Futures replication. Especially prominent with commodity ETFs, it is much easier to track oil using CME futures than it is creating an operationally complex system of physical creations/redemptions via storage facility. The downside of this is that these futures can occasionally do weird things (like oil futures going negative last year) and that there are position limits on exchange level, basically capping the maximum size of a given ETF. Futures replication can also be used to achieve leverage with double/triple short and leveraged ETFs on stock indices.
• Swap replication. Certain indices are often rather tough for the fund manager to replicate. Swaps work as follows:
  • Suppose the ETF issuer needs to track the MSCI World index but doesn’t want to buy into every single country in the index because it is complicated. However, this issuer has a friendly bank it can do a swap against.
  • ETF issuer seeds the fund with EUR 10 million from first investor or Market Maker.
  • It proceeds to buy something very vanilla and cheap with this cash, typically government bonds or boring European stocks that nobody is interested in.
  • Finally it calls the friendly bank and enters into the swap agreement: the bank writes a swap on the MSCI World index and the Issuer writes a swap on the basket of boring things that it just bought.
  • The bank is happy: it charges an annual swap fee and quite likely sold the boring basket, moving it off it’s balance sheet.
  • The ETF issuer is happy: it is now perfectly replicating the MSCI World index with just one swap agreement in place.

European ETFs have been especially big on swap-based ETFs, given that a lot of these issuers grew from big banks (Deutsche, BNP, SocGen etc). Swaps allow you to achieve any exposure that you want, and the bank will always write you a swap (for the right price, of course).

Ways to create ETFs

Like with most securities, ETFs have primary and secondary markets. While the latter is pretty clear and open to anyone with a broker account, primary markets are generally open to a select group of institutions: Authorized Participants (APs). These are generally market makers, banks, and large institutions. The key feature available to APs is the ability to create or redeem the given ETF, effectively minting or burning ETF units. This is done in two ways: cash or “in-specie”:

• With cash creations (redemptions), the AP submits cash (ETF units) to the Issuer and receives ETF units (cash) at a predetermined time of the day, for instance daily close of the underlying index.
• “In-specie” creations function slightly differently and involve an exchange of ETF units for the exact (or close enough) underlying basket or asset.

The costs for creating/redeeming vary a lot, depending on cost of execution for the underlying basket/asset, replication method used and any applicable taxes. For example, an ETF on FTSE100 index could have -1/+1 bps redemption/creation fees with one issuer and -1.5/+51.5 bps with another, simply because there would be stamp duty payable and fixed fee to settle 100 different stocks for the physical replication of the latter, compared to swap based replication of the former.

Governance trickery

Another topic we need to cover, as it will become important for further discussions, is governance. Due to the growth of ETFs and their widespread use, some ETFs are holding very significant ownership in some of the stocks! There are some interesting implications of this:

• ETF issuers don’t have a mandate to participate in corporate governance. Therefore, for the companies with a significant portion of holdings “stuck” in ETFs, governance becomes a rather funny concept.
• Some of the ETFs out there are sector-based. A case can be made that this kind of structure lessens the competition. If a significant portion of the sector is controlled by a number of passive funds, their holders are better off with less competition and oligopoly-type competition.

There has not been any solution to these issues in TradiFi to date, creating an opportunity for a different type of structures in the future.

What’s required for ETF to succeed

If you are an ETF issuer, what are your biggest business challenges? You need to structure the product, sure. You’d probably like to differentiate your product too, as the world may not need the 10th S&P ETF which is also quite a commoditized product. This drive for differentiation results in all kinds of long-tail ETFs like ETP tracking Corn or Ark Invest line of funds. Unless you are a S&P ETF, liquidity is likely your major concern. But liquidity in ETFs is hard to come by. This is because there are way more trading participants trading equities: directional players, quant firms, long-short hedge funds, etc. For most ETFs, well, you have a bunch of long-term investors on one side and occasional sellers on the other side (or vice versa!), unless you have a trusted liquidity provider.

I remember presenting to ETF Issuers as a market maker. Our presentation would be simple. We press the button (stopping quoting) and liquidity disappears. That is why ETF issuers are fighting to attract market makers. I would go so far as to say that until there are top class market makers in ETFs with specialized knowledge, the space cannot scale.

So far so good for market makers, right? Maybe, but ETF market making is not an easy nut to crack.

Challenges of providing liquidity in ETFs

• Cash heavy business
  • Just like for any cash product, the key ingredient for market making ETFs is to have inventory by either 1) creating a big number of ETF units, hedging it with futures on the same index or 2) approximating the hedge or borrowing the ETF from long holders. Both ways make the ETF MM business rather cash heavy, with the larger market makers operating billions of USD of inventory. Most of this inventory costs money to hold (management or borrowing fees), so you have to run a very tight shop in terms of making sure that you are not sitting on inventory that doesn’t trade much, bleeding on management fees and interest every day. On top of this, most creations/redemptions have a minimum number of units required to initiate the primary market trade. Often a market maker would end up selling 1 million USD of a given ETF and then having to create 5 million USD of inventory, sitting on the resulting 4 million of the ETF that nobody else wants to buy. I remember sending sad lists of ETFs we wanted to buy/sell at NAV (fair value at the given day close) to other MMs, only to be met with similar lists in return…
• Perfect hedging challenges
  • Perfect hedging has been a tough one to crack as well. Indices like the S&P have a super liquid future on CME, making hedging rather trivial. An automobile sector on Stoxx600 (a broad European index) would have a future as well, but its liquidity would not be significantly better than that of the ETF, and if you wanted to achieve a near perfect hedge, you would have to trade the underlying stocks or cover at least 75–85% of the basket. And then you would try to hedge a broad global index like MSCI Emerging Markets, and not having access to Brazil or China could introduce some significant variance in your model.
• Pricing challenges
  • Pricing faces similar issues. Asian markets would be closed when Europe opens, so you would have to reverse engineer pricing from any European ETFs referencing similar products and to keep an eye on US ETFs opening up for any surprises. On top of that, you have to monitor any respective news for any market that is closed for whatever reason, or fall prey to small and large arbitrageurs or other MMs who did their homework better than you did.
• Taxes
  • And then there is another major component — taxes. Dividend taxes in particular are a proper challenge to a well-run MM shop. A big part of ETFs would withhold dividend tax on constituent companies, meaning that if you are long the ETF and short the underlying stocks, you would be losing money on the dividend day (as you would have to pay the full dividend on your short stock positions). Add to this the wide creation/redemption ranges for physical ETFs and the large minimum creation/redemption sizes, and every dividend season becomes a marathon to dump the ETFs that you don’t want to hold with big downside and not a big upside from being right (as it is either very expensive or outright impossible to borrow these ETFs to short them during the dividend season).
• Latency
  • Finally, there is a question of how fast you are. ETF market makers faced the same challenges as any other HFT businesses. Getting the fastest signal from the leading derivatives exchange (CME or Eurex) was essential and so we started with getting the fastest fiber lines between all the data centres, following with a network of microwave towers assisted with short-range lasers and other sci-fi stuff. Needless to say, it is all super expensive infra that creates a pretty strong barrier to entry for any new Market Maker.

As you can see, the job is tough. You need a lot of cash lying around for infrastructure investments and buying into inventory. Even then, the amount of optimizing you need to do to make the best use of the inventory is non-trivial and small mistakes can cost you a lot if you don’t do your homework well enough. Even in traditional markets, few market makers have made it. Many desks, being unable to make money on ETF market making, have had to resolve to getting paid by the ETF issuers to quote or focus on very niche parts of the market.

Closing statement

I have barely touched the tip of the ETF Foundation iceberg, not going too much into more specific cases (actively managed ETFs, smart beta, short and leveraged ETPs, etc), but I hope that this gives you an initial indication of what can be copied and what can be improved when it comes to reinventing these products in DeFi. It is also important to understand how crucial and complex the market making part will be, as projects, exchanges, and the investors are evaluating how to create liquidity moving forward and how to engage, incentivize, and evaluate their partners on the trading side.

Evgeny Gaevoy is the founder and CEO of Wintermute, a leading global algorithmic trading firm in digital assets.

Follow us on LinkedIn and Twitter (CEO @EvgenyGaevoy, COO @emgurevich)