" Hidden Orders: A useful tool in a trader’s tool box to defend against predatory HFTs Ever since the publication of the book Dark Pools in 2012, it became public that ECNs and exchanges were accommodating HFTs firms by offering them special order types that most market participants don’t get. These order types enable HFTs to decrease the amount of “adverse selection” they are exposed to and to be at the top of the queue, key components of profitable market making. Most HFTs firms have dozens (if not hundreds) of special order types at their disposal by exchanges and ECNs. Retail traders who want to be able to fight back and cut down their trading costs should learn and implement useful order types that are available to them. There is an interesting tool that is not very well known by traders that can help you decrease the amount of HFT “gaming” your orders suffer. Hidden “native” orders are orders that sit on an ECN or exchange and that, in theory, are not seen by anyone. I say in theory because the exchanges/ECNs have broken the rules before in order to accommodate HFTs. It’s possible but unlikely that some of them have leaked information about hidden orders at some point. These orders are different from iceberg/reserve orders, iceberg orders show up on the Level 2 window and on the direct feed data that HFTs receive. They usually appear as 100 shares and give HFTs an opportunity to “penny jump” the order by going in front of it by 1 cent. This exposes the order to adverse selection, meaning they will get out of the way when they see a shift in the short-term supply and demand and stay ahead of you by 1 or a few cents when they don’t. You tend to miss out on fills that you want and get fills that you don’t want. Lots of traders report the experience of having a stock move down or up right to the point that their limit order is not get filled by one cent (or get partially filled). Sometimes this happens due to randomness, but a lot of the time this happens because HFTs use passive non-HFT orders as insurance. If they are bidding for a stock, they know that if they get ahead of you by 1 cent, they can get their fills and try to sell on the ask. If all goes well, they make the spread, if it doesn’t and the stock looks to drop, they dump their shares into your orders. You get your fill but now you are also underwater in the trade (or you left money on the table for a buy to cover order while exposing yourself to a squeeze). A useful way to diminish this passive order gaming is to use “native” hidden orders. With these orders, the HFTs tend to only find out where your orders are after they have been executed (whether in full or partial). Let me give you an example: Let’s say you are short 1,000 shares of stock ABC, it’s trading at $35, it then proceeds to drop as you expected. You decide to take profits in the $34.60-$34.65 area because you think there will be some support at $34.50. You put a hidden buy order for 500 shares of ABC, native to ARCA at $34.64, 200 shares at $34.63 native to NSDQ and 300 native to NYSE at $34.63. When I say native to the ECN/exchange, I mean that it will only interact with the order flow of that ECN/exchange. For instance, if someone sells 500 shares of ABC at $34.64 at NYSE, you will not get filled because your hidden order was sitting at ARCA. Hidden orders can even be traded through, meaning, you can see prints bellow the price of your buy order. This disadvantage can be countered by using what I call a hidden order “net”. You use several ECNs/exchanges that have the most volume, instead of just one. That way you have multiple order flows to interact to. By having your orders distributed that way, when the stock drops and stops in the $34.60s area, you will avoid being used as insurance for HFTs looking to make markets. If the stock really drops quickly, say from $35 to $34.50, it won’t make much of a difference whether you are using hidden or lit (displayed) orders. It’s on the marginal situations where your orders are on the bottom or close to the bottom that hidden orders can decrease your trading costs by exposing you to less adverse selection. With the lit orders, having the HFTs penny jump you and prevent you from getting filled at or close to the bottom can cost you significant profits as the stock reverses higher and you are forced to hit the ask in the $34.70s area. With the hidden orders, the HFTs can still game you in a similar fashion but they can do it only after they find out that there are hidden orders at a certain level. In a fast moving stock with good volume, by the time they find out, it’s too late. Your order is likely to at least have been partially filled. In addition, HFTs don’t like hidden orders because even though they can find out that they are at a certain level, they don’t know when they get cancelled or entirely filled and the free insurance is gone. They do like to use a trick against these orders, odd lot trades. Sometimes, in order to find out if there are hidden orders in an exchange/ECN they will send out orders for a tiny amount of shares (usually less than 50). When they discover the hidden liquidity, they can penny jump it or just keep sending odd lot orders to annoy and/or increase the trading costs of the trader behind it. Routing a Hidden Order Net Which routes to use to best increase the change your hidden order net will capture most of the volume in a particular stock? You have to be aware of where, currently in the US, most of the volume is going through. As of September of 2014, volume is distributed as follows: Source: BATS Exchange. By the table, you can see that in a NASDAQ stock (Tape C), NASDAQ and ARCA do most of the volume (over 37%). For a NYSE stock (Tape A), NYSE and ARCA do most of the volume (about 31%). For AMEX stocks (Tape C), ARCA and NASDAQ do most of the volume (about 38%). When routing your net in a NASDAQ stock, the better routes are NASDAQ (sometimes still called Island by some brokers) and ARCA as a secondary route. When trading a NYSE stock, the better routes are NYSE and ARCA. For AMEX, ARCA is the main one and NASDAQ is a secondary route." due character limit of ET rest of the article at: http://www.beathft.com/?p=88 Anyone agrees or disagrees with this? Any technical flaws in the article?
I don't see these as good as you say. First, you won't get NMS protection, so you may miss out on fills that way. Second, you pay higher commissions since the rebates for displayed liquidity are several times higher than for hidden liquidity. Lastly, like any limit order, if you don't react very quickly to changing market conditions, you'll still be subject to adverse selection when the market moves through your price. If you're not as good as the HFT guys, arguably taking liquidity has about the same cost in terms of spread as does the adverse selection you suffer from getting run over with a limit order. Your specific suggestion of having lots of orders on different ECNs is also problematic. How many shares to you put on each if you want to buy 300? 100 on each of 3 exchanges, or 300? Even in the good case where you don't get run over by a market move, you're likely to only get filled one order at a time since each trade usually only happens on one exchange. So now it takes 3x as long to get filled, and with more commissions for more orders. If you put 300 on each of 3 exchanges, you take a lot more risk since you can easily get filled for more size than you wanted.
Hidden orders have sucked for years. hft goes out of their way not to fill them unless its at a very bad price. hidden orders would only be useful if there was 1 consolidated exchange, the way it is now its a clusterfuck getting filled.
I agree that its not a panacea but it is still quite useful. As far as rebates go, its not a big deal, the article mentions that you are supposed to use in big spread stocks. If the spread is 20 cents and you are trying to not pay it, a little rebate won't matter. Also you get a higher rebate with a displayed order but with more adverse selection (which is a cost), how much is that cost? Probably more than a little rebate, at the very least it should be a wash. As far as using multiple orders, that's to avoid the issue you mentioned earlier (NMS protection). I wouldn't run the risk of overfilling. Usually i just use the most liquid ECN or exchange or 2 orders with the most the and second most liquid place. In the scenarios mentioned: your orders end up at the turning point of the stock or the stock is a big spread and you want to free shot to get in (otherwise its unprofitable), hidden costs can save you some money.
The points that you bring up are the basic idea for using darkpools. I used to rely heavily on hidden orders and dark pools when I traded in dry stocks, they made life much easier for avoiding the whole HFT BS and also preventing my order from holding the market from going in my favor (as it sometimes happens with reserve orders) To prevent the HFT's from sniffing your order easily, you can pick a few tricks from the darkpools playbook, like cancelling the order when the spread is too narrow (or using an order pegged to the far side of the spread) or limiting the maximum % of the volume represented by your fills with TWAP-like logic.
I agree with you on dark pools. I wrote an article about that and sent to traders magazine, they published it in their site but they botched it so badly in the editing that the article is almost unreadable. The original version is here http://www.beathft.com/?p=75 The tough thing for retail traders is to get access to these dark pools in the liquidity adding side. IB has IEX along with some other brokers but the other big dark pools is tough to access. Brokers should enable them more
Very good article. CS Crossfinder used to be one of my favorite darkpools... I guess they changed their ways towards the dark side (pun intended) since I stop trading illiquid stocks. One advantage of using the hidden orders from displayed exchanges, is that you don't have to pay the spread as you usually do in the darkbooks and you can just sit one penny inside the spread from a displayed order to get your fills from orders that are trying to hit the displayed book. About the fills with odd lots that the HFT's may use... if I recall correctly there is a flag on the FIX to set minimum fill size, although I don't know how widely adopted it is nowadays. Another alternative is to sit under the displayed liquidity (that's usually 100 or 200 shares in the dry names), hoping for an order to swipe the book.
I've been using hidden nasdaq midpoint pegs to add liquidity in most cases. The goal is to avoid HFT with the ability of price improvement. The price improvement makes up for no rebate ($0 fee/rebate). Sometimes you miss fills, but the trade off is worth it imo. I use PDQ for taking liquidity. They hit the most dark+lit destinations, and you can set a fixed cost with the broker that is far less than the usual 32 mills for taking lit liquidity. Once you have a relationship with them they can also develop custom routes to suite your needs. There used to be tons of price improvement in darks, but not so much anymore..hft has filled the gap.
I'm curious, what typically is the spread size in the stocks that you used these hidden mid-point pegs? I find that in big spread stocks (like GPRO), when I go into the middle of the spread (displayed), a lot of the time I get hit instantly
I like them for the thicker stuff (1-2c), but you can use them on anything. The risk in the bigger spread stocks are manipulative games that change the nbbo to give you the worst price.