Instead of matching orders in the order book by price then time, what are the consequences if orders are prioritised by price then fee paid per share in the order?
An idea similar to the way transactions are prioritised in blockchain.
3 Answers
$\begingroup$
$\endgroup$
1
I think this would be equivalent to having an infinitesimal tick size, since you could always improve your execution priority by increasing the price you offer.
-
$\begingroup$ ...which probably isn't a great idea either, since the race to be first would surely result in prices being submitted like "99.99999999999999999999999999999999999999999999999999999999" which isn't great for readability, storage or processing from the PoV of an exchange. $\endgroup$ Commented May 6, 2019 at 18:06
$\begingroup$
$\endgroup$
3
Would it be fair?
On the one hand, the total price paid for a security is simply the asset price + execution fee, and if one is willing to pay higher than another party then so be it, that's their edge. On the other hand the market is no longer transparent. Exchanges display bids and offers and you place your order with some inherent knowledge about your position in the stack, with much higher clarity (at least) relative to the people coming later than you. When the exchange offers priority for a fee they are introducing a variable to the stack that is unseen.
Is this practical?
No, not in my opinion. Exchanges have very specific tick sizes and priority rules (different for differnet products) to encourage an orderly market and good liquidty. Examples can be cited where the Exchange has changed the rules, seen liquidity fall off and then reverted back to their original structure. I suspect having a fee priority structure adds such an awfully opaque feature to the market that it would drive customers away and dwindle its own revenues (which the fees would presumably be trying to increase).
Thinking about it if Bitcoin had this type of pricing embedded it might have been supportive of its recent price decline, if its anything like financial market liquidity.
I will see if I can look up an old answer on this topic, but also this would make the market more susceptible to spoofing since the spoofer might place a bid (or offer) and together with a zero commission fee, with the expectation that he is sufficiently deficient in the stack and unlikely to trade at all, but still have the order in the book.
-
$\begingroup$ Arguably, in a regular time/priority queue, your position in the queue isn't very transparent either, and spoofers can easily achieve the same low execution probability by continuously modifying the size of their order thereby sending it to the end of the queue. $\endgroup$ Commented May 6, 2019 at 10:38
-
$\begingroup$ @wildbunny agreed, and there is often some randomness applied as well. One major difference though is that a genuine bidder who might have 5000 bids in front of him will have assured confidence that after the others have executed (or modified) he we gain a higher priority and/or be front of queue. Whilst not necessarily transparent, it is different to an ordering whereby you can be perpetually at the back of the queue solely by the price you have entered and the price other bidders have entered, but unaware of it. $\endgroup$– Attack68 ♦Commented May 6, 2019 at 10:57
-
$\begingroup$ Yes, I agree it would remove the deterministic nature of queuing at a given price level. $\endgroup$ Commented May 6, 2019 at 11:06
$\begingroup$
$\endgroup$
2
Yes, that's fine, though I think, it's better to think in terms of minimal price increments: essentially, you're suggesting to replace 1 cent minimal increment with something much smaller. So, say, an order to buy @ 10.01 with 0.002 fee is the same as an order to buy @ 10.012. This will encourage price discovery within the one cent spread and will make the HFT race to get an order well-placed in the queue unnecessary.
-
1$\begingroup$ Your observation is right, but I think the fundamental difference, being a market with smaller tick sizes, and a market where the recorded price transaction is not actually the price paid (due to hidden fees), is important. To take it to the other logical extreme why not have much wider official tick sizes (which record transaction) and permit a whole scale of intermediate determined prices by necessarily varying fees contributed? It seems inherently disorderly and far from transparent, which is why I dont believe it would be successful in most markets (and partly because the race is ruined) $\endgroup$– Attack68 ♦Commented May 6, 2019 at 12:30
-
1$\begingroup$ I feel, you are arguing about semantics. What is that fundamental difference, that you think is there? The only difference I see, is that the fees are hidden, while with the smaller tick size, the fees one pays for the transaction will be printed on the tape. It's actually, already often the case - you can see sub one cent execution prices for retail orders. If you keep them hidden, that will reduce transparency and will only harm price discovery. One cent minimal increment is there mainly for historical reasons. $\endgroup$– LazyCatCommented May 6, 2019 at 13:48