By Phil Mackintosh, Chief Economist, Nasdaq
In accordance with the new NMS-II rules that expanded core data to include depth, the committee recently filed a proposed rate sheet. Some industry participants have already complained because it’s not cheap enough.
But what new price is right?
This is where SIP accounting, and what creates the most efficient marketplace, gets complicated.
The current SIP tries to allocate costs of data based on usage and value
Before we start, let’s recall how the SIP works now. The SIP takes data from all exchanges and consolidates it for customers who want to know market-wide prices.
Even back when the SIP was created, they realized it wasn’t fair for data to be free. Instead, they set up a system to:
- Charge customers based (roughly) on their usage.
- Reward all the venues for their contribution to price discovery.
The current SIP only provides the top-of-book for each venue, but it still charges professionals more than retail investors. It also charges algorithms, dark pools and electronic traders even more for the exact same data. Most people would (mostly) agree that’s fair and economically efficient, based on the fact that professionals trade more than retail, and computers can trade even more again. It’s also true that professionals and computerized trading also earn significant revenues from the data.
Currently, the SIPs only distribute the data required to comply with trading rules in Reg NMS. In contrast, the new NMS-II plan expanded core data to include data traditionally only bought by more professional traders, those with enough trading to benefit from the information about quotes away from NBBO (depth) or auction imbalances. However, as this data isn’t required for small trades, the new NMS-II rules also made it voluntary to buy this new data from the SIP.
The point of a competing consolidator is to consolidate competitively
In short, the SEC expected the competing consolidators would bring two key benefits to investors:
- Streamline the provision of data to all participants, creating a single source that would provide a “validated NBBO” and additional data if each investor desired to purchase more, like depth or auction data for their customers, based on a single rates sheet.
- Reduce geographic latency.
The industry has voiced plenty of other concerns. As we forecasted back in 2020, it’s possible a faster consolidator will be able to extract more revenues for data than a slow one, adding to costs of fast data, creating a multi-tiered market depending on what participants can afford. Plus, investors might need a backup feed, each paying for more than one consolidator, also duplicating infrastructure. They will also create many NBBOs, which may complicate markets and best-ex, among other unintended consequences.
The SEC rule mandates data revenues for all exchanges
In mandating the creation of more core data for investors, the new rule acknowledges the value of data to traders and investors. Economically, it also forces an unbundling of depth data from exchanges – even the exchanges that offer data for free now.
However, our research shows that rewarding all venues for any data is inefficient. In the real world, fixed costs to the industry to connect to each additional venue matter so much that many smaller exchanges can’t give their free data away – even to the most sophisticated traders.
There is over 10-times as much data
But back to the original question: What new price is right?
Should it be based on the amount of data itself? As we mentioned back in May 2020, one method for computing prices for depth would be to compare the amount of data in a depth feed to the data in top-of-book.
Based on our estimates at the time, that could make a “fair” price for depth around 10-times the price of top of book.
Top-of-book data now costs $75 per professional user per month. Based on that, if competing consolidators charged for depth at 10x the costs currently charged to professional users for top-of-book, the cost of depth to professionals would be $750.
Chart 1: Relative messages on SIP vs. depth
Although it’s also important to remember that even when depth becomes “core” data, buying it is voluntary. Most retail traders, with their orders much smaller than the size at the NBBO and their fills inside the spread, have little need for depth.
That means that even at 10-times the cost, the revenue collected by the SIP would not be anywhere near 10-times the SIP top of book revenue.
What does depth cost now?
In reality, very few investors need depth or sub-millisecond latency. So a rate sheet that allows investors to choose appropriate to their usage is also important to make sure small investors don’t subsidize large professional traders.
Consistent with that point, proprietary data feeds are currently also voluntary to buy. Data shows that customer choices result in the market solving competitively for the price of proprietary data for each exchange.
In fact, we can see that in Chart 2 below, where we compare the cost of each proprietary data feed to the SIP weighted market share of each exchange. Importantly, the chart shows:
- All exchange groups with mature data pricing models sit almost perfectly on the diagonal line – meaning there is a consistent charge per unit of liquidity provided by that exchange (large bubbles).
- We can use this charge per unit of liquidity to estimate the value that “free data” exchanges are bundling into other services (orange lines)
- We could also use it to argue that CboeOne Premium, a partial depth product, is also clearly priced at a discount compared to its full depth products in order to attract other revenues.
- It also seems that although exchange groups’ prices are on the line, separate venues are not dynamically modified for market share shifts within the group. Perhaps the SEC’s process to file and approve all proprietary data rate changes contributes to that (small bubbles).
Chart 2: Cost of proprietary data vs. market share
Using this unbundled market price of depth data calculation, we can estimate what the market-based fair value of a consolidated depth feed should cost.
Some might say it’s unfair to unbundle like this. But note that the new NMS-II rules mean those exchanges are also likely to receive revenues for their depth data going forward, regardless of their current public stance on data or business model.
So what would a consolidation of prop data cost?
Calculating the market-based fair value for all prop data fees using this approach shows that the unbundled consolidated value of all prop feeds would be $455 per professional user (Table 1 below).
Table 1: Actual and unbundled depth data cost estimate
But depth data only goes five levels deep
Clearly, the new core data offering, which only adds five levels of depth (and odd lots) to the top-of-book product, shouldn’t cost the same as a full-depth product. But how much less should it cost?
One way to think about that is to look at how much of the time investors actually need more than five extra levels of depth. Based on current liquidity consumption in the market, the answer would be around 12% of trades need more than five cents of depth (although we’d note that most of the 10% comes in high-priced stocks).
Chart 3: How much does each level of depth add to total liquidity the market needs through the quote
Adding this all up
If competing consolidators simply passed on market prices for proprietary data costs currently charged to professional users, the cost of depth would be $455.
If we discount that because the new core product cuts off at only the 5th level of depth, by 12%, it gives a cost of $400 for professional users.
However, the new proposal also sets separate rates for auction and odd lot data, with auction data adding to $30 per professional, and while odd lots are free once you buy depth. Removing that means the remaining SIP depth data would be $370.
That’s much lower than a fixed cost per message.
The new proposed fee for depth feeds is actually $297 for professional users.
Adding the NBBO for $75 brings the total to $373, almost exactly the same as the fair costs for just proprietary data now across all exchanges.
To be consistent, the same ratio of depth to top-of-book was applied to retail investors – which itself is consistent with the ratio among many exchanges.
Retail investors get a much bigger discount on all data, as they trade much less. They now pay at most $3 per user per month for top-of-book SIP. So using the same 4-times ratio ($75 Top; $297 depth for professionals), the new depth fee for retail is just $12 per user per month. Not bad for a retail customer doing lots of free trades.
If you (really) want depth data, the economics need to support it
No doubt, some say this is still not fair. But at least it’s closer to market-based.
It’s also important to remember that if we want the market to produce public data, the public has to pay a price that allows exchanges to offer incentive programs that make it economically attractive for buyers and sellers to post. There are already a lot of hidden orders in the market.
