# ETF Market Making - Locking profits via hedging

I am interested in deeply understanding the way ETF market makers operate to profit. I already know that market makers profit from buying at the bid price and selling at the ask price, and I am also aware of the creation/redemption mechanism that allows them to make a profit if the ETF value deviates too much from the NAV. What puzzles me is how they use hedging to 'instantly lock the bid ask spread profit', as I have read in many different places. For example, the following paragraph can be read in a Virtu Financial's repot - emphasis mine:

Our strategies are also designed to lock in returns through precise and nearly instantaneous hedging, as we seek to eliminate the price risk in any positions held. Our revenue generation is driven primarily by transaction volume across a broad range of securities, asset classes and geographies. We avoid the risk of long or short positions in favor of earning small bid/ask spreads on large trading volumes across thousands of securities and other financial instruments.

By instantaneous hedging, I understand that they might be using strongly correlated products such as futures. However, what I do not understand is the fact that, if they aim at instantaneously hedging their position on ETF, they need to do a market order in the hedging product, hence paying the bid/ask spread there. Therefore, the previous paragraph only makes sense to me if the bid/ask spread of the future (for example) is smaller.

More precisely, continuing with the ETF shares and futures hedging example, the mentioned paragraph only makes sense to me if the following relation holds at least most of the time:

$$ETF_{Bid} \leq Future_{Bid} \leq Future_{Ask} \leq ETF_{Ask}$$

If this relation holds, and the market maker follows the following rule

1: If I buy the ETF at the bid price, I perform a market order to sell a future

2: If I sell the ETF at the ask price, I perform a market order to buy a future

then indeed the market maker is capable of locking returns. For example, if he buys an ETF share at time t=0 and sells it at time t=1, then \begin{align} P\&L(1) &= ETF_{Ask}(1) - ETF_{Bid}(0) + Future_{Bid}(0) - Future_{Ask}(1)\\ &= ETF_{Ask}(1) - ETF_{Bid}(1) - (Future_{Ask}(1) - Future_{Bid}(1)) + (ETF_{Bid}(1) - ETF_{Bid}(0)) - (Future_{Bid}(1)-Future_{Bid}(0)) \end{align} The first two terms represent the ETF Bid/Ask spread vs. the future Bid/Ask spread, whereas the second two terms gauge the hedging efficiency (HE)

Using the above-mentioned relation, it can be easily seen that

$$ETF_{Ask}(0) \geq + Future_{Bid}(0) \geq ETF_{Bid}(0)$$

$$-ETF_{Bid}(1) \geq -Future_{Bid}(1) \geq -ETF_{Ask}(1)$$

Hence,

$$ETF_{Ask}(0) - ETF_{Bid}(0) \geq HE \geq -(ETF_{Ask}(1)-ETF_{Bid}(1))$$

Thus, assuming that the hedging efficiency is on average 0 (and, when different from 0, very small and little risky), under this hypothesis the market maker can only make profit on average and with little risk by instantaneous hedging techniques if the mention realtion holds.

I know this is an oversimplified example, but is this roughly how this instantaneous hedging thing works? I.e., using the fact that the hedging instrument has a tighter bid/ask spread.

If this is not the case, I would very much appreaciate a precise explanation/bibliography to understand what they exactly refer to.

I've worked in this industry for a while and have run ETF market making for quite a few years. It's very difficult to perfectly lock in profit as you detailed above. With fast equipment it can be done sometimes.

But most of the time you really are just hedging to model - and there is risk in that case. For example, you might sell ETF X and then hedge buy buying what looks cheapest, in this case stocks A,B,E. The trick is the scale. As you trade more and more products you are able to hedge through a more holistic view of the risk. You might sell QQQ, buy SPY, and then sell a few higher beta tech names, for example.

• thanks for the answer! In any case, although maybe the perfect profits lock is not possible, I assume that the key of ETF hedging is that the bid/ask spread of the hedging contract is smaller than the ETF's one, right? I mean, the hedging works in such a way that, when your ETF limit order is hit, you perform a market order in the hedging instrument (whichever you find to be more appropiate at the moment). Did I get it right now? Jan 18, 2021 at 9:18
• Well, some part of the hedge is misaligned. The spread could be wider - that doesn't matter itself. For example. I can sell SPY and then buy some Nasdaq futures. They might have a wider spread, but if the offer is low enough that I think it's a good price based on hedging model risk, then I'll take it - even if the bid is \$10 below the offer. Jan 18, 2021 at 17:55
• but the problem I see is that the moment you unwind your position (say, because your SPY bid is hit) you need to perform a market order in the hedging contract. Hence, the money you make from the ETF bid/ask spread is compesated and actually made negative by the money you loose with the hedging contract bid/ask spread. If you make money in that particular transaction, it would only be because of the misalignment of the hedging contract. Hence, you would be making money from spread trading rather than from the market making activity itself. The only possible advantage against... Jan 18, 2021 at 20:18
• ... a pure spread trade would be that you are not paying the ETF bid/ask spread. But that can be easily avoided with a lead/lag scheme, right? Jan 18, 2021 at 20:19
• also implicit here is that the ETF market-maker has a ton of cash (ie stock), futures and options business crossing their desks, plural. So the bank doesn't need to hedge the ETFs specifically. It just needs to hedge the combined exposure to that, the underlying stock and any correlated derivatives thereof. An ETF trade might actually go completely unhedged, because it hedges an options trade made earlier on a different desk. All the market-making bank cares about is its aggregate exposure, across all channels (of which ETFs is but one small example). Jan 19, 2021 at 4:46

An ETF typically appoints one or multiple Authorized Participants (aka APs) which are allowed to buy and redeem ETF shares directly with the fund, by exchanging the fund shares against a basket of the underlying securities. These APs are often the market makers and their arbitrage involves managing an inventory of underlying securities and fund shares. I can think of one case where such hedging would be close to 'instantaneous' - closing auctions.

The non-transparent ETFs (instanceOf Precidian) introduce specific challenges into the market making process since the fund composition is supposed to be unknown and creation/redemption process is more complicated.