U.S. investigates market-making operations of Citadel, KCG
reuters.com
reuters.com
You can never get a worse fill than what you ask for but the internalizer can make money off your order and you can miss out on market moves that occur in the second after you place your order. if you are fine getting price that you put your limit order in for then as far as you are concerned, you'll be fine.
As with all things market micro structure related, nanex.net has a good writeup on this. I'm not sure how often this scenario occurs as I've never worked for an internalizer but its
http://www.nanex.net/aqck2/What-Every-Retail-Investor-Needs-...
Check out the section called "Marketable Limit Orders".
> For example: the Wholesaler receives a retail order to buy 2000 shares at $10.03 or better (lower) when the SIP shows a total of 2000 shares offered at the best offer price of 10.01.
1. Wholesaler buys 300 from one exchange at $10.01, and immediately, 1700 shares that were available on other exchanges disappear, causing the best offer price to move to $10.02.
2. Wholesaler buys 400 at $10.02, and again, sell orders on other exchanges disappear, causing the best offer price to move to $10.03.
3. Wholesaler sells short remaining 1300 shares at $10.0290 to retail investor, providing a $0.0010 price improvement relative to the $10.03 SIP offer price at time of execution.
4. Within seconds, stock reverts back to $10.01 offered.
5. Wholesaler covers short by buying 1300 shares at $10.01.
> By quickly influencing the price of the stock, perhaps by less optimal routing, then directly filling the order at an execution price away from the original NBBO at time of Order Receipt, the Wholesaler profits from the $0.019 change on 1300 shares ($24.70), less any "price improvement" given to the retail investor, less the $4.00 Payment for Order flow paid to the Retail Broker (2000 shares x .0020 per share) and $6 in exchange fees (maximum SEC fee $0.0030 per share x 2000 shares).
0. The internalizer tells the brokerage that they want to buy 2000@$10.03.
1. Customer places a buy order 2000@$10.03.
2. Brokerage looks at the internalizer and at the NBBO. Brokerage realizes the NBBO is better than the internalizer and so routes a 700@$10.02 to the public markets as per their legal obligation due to Reg NMS.
3. Brokerage routes the remaining 1300 shares to the internalizer, where they are filled at $10.0290.
Customer trades are 300@$10.01, 400@$10.02, 1300@$10.029, resulting in (as you noted) a price improvement to the retail customer.
The only this doesn't benefit the customer is if (as kasey_junk notes) the brokerage is using an incorrect NBBO. If you believe a brokerage is doing this please report them to the SEC for violating Reg NMS.
As someone who has actually seen the scenario described above play out, I'm not so sure: stock doesn't move for minutes at a low volume time of day, the moment I place my limit order the price is suddenly higher than my limit; rinse and repeat. I've seen this happen enough times to make me believe Reg NMS is just another regulation they've found a plausible deniability around (BTW, before you tell me "prices going up when you place a big order is how the market works", I've tested this with 100 share orders, so there was no order split going on. I still got priced out.)
That said, in your scenario (you were prevented from trading) how does anyone make money? You were (probably) paying for a trade (which did not happen) and you were using infrastructure that costs money (placing orders), and no one was able to trade against you. That is not a great business model...
Also, none of the groups we are talking about are banks.
Oh, and BTW, order cancellations cost money. Cents, sure, but when you are making money on a 1c spread, you would also benefit from a 1c cancellation.
Other than that, I've been tempted (and many times succumbed) to raise the limit just to get that trade. And then you see the price go up again before your order goes through. It'd be hilarious if it wasn't infuriating.
If the order is coming from a retail brokerage that's a terrible guess - how often do you think someone goes into ETrade and tries to move $20M of GOOG?
Big orders come from institutional investors. Those are explicitly the orders that market makers are trying NOT to trade against, and that's why they try to buy order flow. Order flow is known to come from players who aren't making big orders.
If the ask price for a security is, say, 5.51 and I put my limit at 5.51 the chances of getting the order filled were pretty low. If I was willing to go to 5.53 the chances would increase, if the order was 5.55, I'd probably get a fulfilled order at an average price of 5.53 or so. Not a huge difference on small orders (say 1000x2c = $20) but enough to make a "big" volume, small spread strategy pretty much useless.
All you've discovered is that when you sell a stock, you lose both the upside and the downside.
Once you've accepted the order you are legally obligated to deliver in T+3 days. If you fail to deliver you pay big penalties, your broker shuts you down, all sorts of bad things.
The purpose of order internalization is to trade against uninformed participants (Joe 401k), without risk of accidentally trading against bigger informed players (Bill Ackman). If you are trading against uninformed players, your risk is lower and you need to offer a better price to get order flow.
Read these two blog posts for a bit more detail on price discrimination on the basis of delta toxicity (albeit not in the order internalization context).
https://www.chrisstucchio.com/blog/2014/fervent_defense_of_f...
https://www.chrisstucchio.com/blog/2014/mark_cubans_hft_idio...
If you want to argue that something nefarious is going on, go ahead and describe the mechanism. But whatever it is, it's almost certainly not the mechanism chollida described.
I clearly don't understand what's going on under the table as well as you would, and because the strategy I was toying around with at the time was basically big volume on small spreads ("big" for a retail guy, obviously, I don't have millions at my disposal) it was shot to crap by these magic changes.
So I did the sensible thing and closed my account. Off to invest in less volatile assets, if I can find any, heh.
Is there an order option to prohibit wholesalers from trading before you?
Basically any time you send a "smart" routing order, it is actually the least smart thing to do (unless you REALLY need the liquidity). Your trade order gets sent to all the exchanges, and ah I don't really feel like explaining it.
tl;dr Someone intercepts your data packet to one exchange, and alters the liquidity on the other exchanges, so you get a partial fill and then adjust your order at the slightly worse price.
In the scenario presented, the wholesaler receives your order, then executes its own orders, then fills your order based on its own executions (which you are unaware of).
How can a wholesaler guarantee the exchange price in the face of disappearing orders?
I don't think they can. Instead, the wholesaler fabricates an order at the limit of your order, then reports the sale to you.
Edit: [rate-limited] Reply to tptacek comment below:
> I can't even tell if you're talking about market or limit orders.
Can you view the context of this thread? I am talking about the explicit steps listed in the comment I first replied to. Specifically, step 3 [0].
In chollida's description the orders are limit.
Step 3 appears to be, exactly, front-running as explained in my post [1] immediately above this one.
I asked if anyone could explain how that behavior was not front-running. You responded; indicating that the behavior was according to regulation.
You claim that the wholesaler must give you the best "exchange price", but I claim that such a guarantee is generally impossible to fill (time-distance-information problem), and that in the specifics of chollida's described scenario, the wholesaler is actually front-running you by examining your unfilled (limit) order and then filling it at the limit (as in, calculus) with it's own fabricated (perhaps, synthetic, but front-ran, nonetheless) order.
If such a guarantee is generally impossible, then the wholesaler must be cheating, the law is incompetent, or, both.
[0] https://news.ycombinator.com/item?id=11668835 [1] https://news.ycombinator.com/item?id=11669193
T0: Bob->Schwab: Market BUY 100 ISSX
T1: Schwab->Citadel: Forward Market BUY 100 ISSX
... and so on? I can't even tell if you're talking about market or limit orders.
You aren't entitled to a better price than your limit. If you think you are, can you provide a sequence of trades in which someone else captures a premium where they don't take downside risk?
The fact still remains that the wholesaler has fabricated an order on the knowledge of a customer's pending order, which affected the execution price of the customer's order, before the customer could even know that it happened.
Even if the wholesaler sent their order to an exchange and still matched with their customer, it is still front-running, as the wholesaler is using knowledge of their customer's order to, essentially, eliminate all possible price improvements.
So, what is being called a "limit order" is just code for "we might just fill your order at your limit [when there are no market orders], if we think we can make money off [front-running] your order with our own".
It would perhaps be acceptable if the wholesaler offered some kind of kick-back on any profits made, but that would need to be a different kind of order and I'm not aware of anywhere that does it.
1) match the order with the NBBO. 2) send the order to the exchange.
If your limit is better than the NBBO, then they are required by the law to price improve it. Further, the nature of their agreements with your broker are such that they are required to maintain a price improvement level (that should be better than NBBO compliance). That is, incidentally also largely the requirement the exchanges operate under as well.
A limit order doesn't have anything to do with the existence or non-existence of market orders, it has to do with your limit and the NBBO. The way they make money is not by changing the market against your interests, but instead booking the spread (and in fact the reason they like retail order flow is that it is naturally uncorrelated so the spread is more even).
The customer does in fact know that this is happening as it is a regulatory requirement that they disclose it.
I suspect that lots of whole sellers would be happy to kick back profits, if the retail customer was also on the hook to back the losses. Instead, they aren't and they get heavily discounted (to the point that it is now free to trade) trading costs instead.
>In the scenario presented, the wholesaler receives your order, then executes its own orders, then fills your order based on its own executions (which you are unaware of).
>The way they make money is not by changing the market against your interests,
Oh, but they do. They fabricate an order in response to yours. If they did not act, your order would not have filled at $0.001 under your limit.
>but instead booking the spread (and in fact the reason they like retail order flow is that it is naturally uncorrelated so the spread is more even).
You can call it whatever you want, the wholesaler is manipulating the market to their advantage. The wholesaler knows the price is likely to improve, so it arbitrarily truncates your order and infills its own account with the improvements.
Is it limited risk? Of course, that's their job.
Finally, I'd ask, what is your point? That the law is flawed? OK, then work to change it. But know that lots of people have done their own research and come down in favor of the execution cost benefits of wholesellers.
Of course they are. Had they not acted, you would have had a chance at price improvement. Instead, they took your chance for a token payment.
> Wholesaler sells short remaining 1300 shares at $10.0290
whereby the wholesaler takes a speculative position wrt the original order?
Are you suggesting that the wholesaler giving you 0.0010 profit on your trade guaranteed, is working against you? Without regard to your execution costs?
Can you suggest a single chain of messages where you make money on that trade? What are the chances where that chain of messages is likely?
The wholesaler is making off-market trades and treating them as if they are on-market. The wholesaler would not trade, if it did not think it would profit. The wholesaler's profit is the difference between their buy-price and what they actually paid you.
Are you suggesting that if the wholesaler wasn't interdicting orders the average price improvement for these orders wouldn't be close to the average profit the wholesaler makes on each instance of such a trade?
They do this because they are taking on the risk that the market will eventually allow them to work out of their short position at a better price than they paid you. But they don't know that it will do that.
One of the reasons they pay for retail flow is that it on average goes back and forth, making it more likely that this trade works to their advantage.
None of the profit of that trade came from you the limit order provider. It came entirely from the average spread.
Or, you know, improve beyond the $0.001/sh the wholesaler paid.
At the point that they filled you at 10.299, the market is set at 10.30. 1 of 2 things can happen. Either there is enough volume to fill you at 10.30 that your order is fully filled at 10.30 (worse than the wholesaler gave you) or there isn't and some of your order fills at 10.30 (worse than the wholesaler gave you) and you have now set the new market level and rest your order. Once your order rests, it will not be improved and will either fill at 10.30 (worse than the wholesaler gave you) or it will be cancelled (you didn't get what you wanted).
Most people placing a limit order like that would prefer the outcome that the internalizer provides, if you don't, great use a broker that allows you to route direct. But its certainly not front-running because the internalizer acted directly in your stated interests, not against them.
Then again, this discussion is on a thread about the US DoJ investigating wholesalers for potentially breaking the law when it comes to what price the required to provide (though apparently a different requirement than the NBBO, specifically, the "best execution reasonably available".)
A legal requirement doesn't mean its going to happen in practice, it means that there is, at least in theory, a remedy available if it doesn't.
(Then again, not meeting or beating the NBBO seems a bit more obvious than not providing the best execution reasonably available, so I'd be somewhat more surprised if this wasn't happening simply because it would seem hard to get away with except in some extreme edge cases.)
An excellent explainer can be found at http://www.nanex.net/aqck2/What-Every-Retail-Investor-Needs-...
No idea if this actually happens on those exchanges, but:
Suppose you place a limit order to sell STCK for $x and I only execute it (giving you $x) if the price goes above $x+1. Someone's making $1 on that spread and it definitely isn't you.
Conversely you place a limit order to buy STCK for $x and I only execute it (costing you $x) if the price drops below $x-1.
Funds like Citadel and KCG pay the banks and brokerages( Schwab, Vanguard, etc) hunderes of millions per year to get their order flow. Here's a reuters article about Schawb's pay for order flow:
http://www.reuters.com/article/schwab-ceo-proposal-idUSL1N0O...
> Trading companies, including market makers such as UBS , KCG Holdings and hedge fund Citadel LLC, are willing to compete for retail investors' orders because they are considered "dumb money" that shows the professionals where markets are headed.
> In reporting first-quarter earnings last month, Schwab said that it expects to earn about $100 million this year from selling client orders, the first time it gave out a specific number, and higher than the estimate it had given a few weeks earlier to Reuters.
and here is the breakdown of their order flow:
https://www.schwab.com/public/schwab/nn/legal_compliance/imp...
The funds then get a free look at these orders where if they are beneficial to them they can fill the order, or if they aren't, then they can pass the order onto an exchange.
If you thought flash orders were bad, these are much worse and flash orders were discontinued years ago.
If you thought HFT funds were bad then consider atleast HFT funds have to put their orders out to the market and take the risk that they'll get rolled over when the market moves. ie they put their money where their mouth is and they compete every day on a level play ground with the rest of the market makers.
These internalizers often hold orders up for up to a second, before they choose to fill them or let them go and when you consider that the IEX delay of 350 micro seconds, not milliseconds, is considered contentious you get a good idea of just how scummy things have gotten.
in an unrelated piece of news Reuters today wrote a piece showing the top paid hedge fund managers. Citadel's founder Ken Griffin can in first, taking home 1.7 Billion last year. This is really the DOJ taking on the biggest players on wall street.
http://www.reuters.com/article/us-hedgefunds-compensation-id...
Do you mean microseconds? Because 350 nanoseconds is only long enough for light to travel about 250 feet in fiber.
That said, I much prefer my free trades from Robinhood that are the outcome of payment for order flow, than I did the $25 it used to cost me to trade. For the kind of trading I do, it is a much better deal to let the internalizers capture the spread (even if it is not the true spread) than it is to not have internalizers.
> Vanguard Brokerage Services does not receive compensation for directing order flow in equity securities.
From http://vrs.vista-one-solutions.com/Disclosure/disclosure.tar...
https://personal.vanguard.com/pdf/v719.pdf
> Vanguard brokerage receives compensation for directing certain options securities to our execution partners.
This is at least evidence that Vanguard (the brokerage with the best reputation for consumers) thinks that the internalizers are a better way to route (securities) than the other options on their own merits.
Though Vanguard probably (I don't know this) also uses at least one of these firms for their execution services on the mutual fund side as well, if only because that is the cheapest way to get good execution rates. I wouldn't be surprised if they pay for this, not get paid for it though.
This is simply false. For example, IB describes the mechanism here:
Liquidity Provider Relationships: IB has entered arrangements with certain institutions under which such institutions may send orders to IB at or near the NBBO. These orders are held within the IB system and are not displayed in the national market. If another IB customer order could be immediately executed against such an order held in the IB system (at the NBBO), the orders may be crossed and the execution reported to the National Market System...
http://www.arcacapitalinvestments.com/new/arcacapitalinvestm...
So no, the liquidity providers don't get a "free look". They get to place orders (before you do) and your brokerage fills them if there is a price match. The price is always at least as good as in the national markets, modulo the limitations of physics/CAP theorem/etc.
The purpose here is to enable price discrimination - you pay a lower price for liquidity (since you have less adverse selection), Goldman pays a higher price (since they have more adverse selection).
Well to be fair, i never mentioned IB, you did. I'm correct on this issue. Let's leave IB out of this as what you describe does happen but is a completely separate issue from selling order flow.
First off Schwab has already publicly admitted they sell their order flow. I've talked to 4 Canadian banks who do the same.
The internalizers who buy the flow are then allowed to either fill the order or pass it along, to the market, they get to see the order before it gets to the market, this is the free look i was talking about.
You may not be aware this happens but it does. I've literally talked to people on both sides of this who freely admit what they do. its currently legal, but I don't think it will be for much longer.
If you have evidence otherwise, by all means post documentation describing the specific mechanics of how matching works. Simply citing news articles that use the word "sell order flow" doesn't count.
I just cited IB because their docs are very detailed and because I know them well.
The KCG and Citadel business is called wholesaling. A broker like Schwab has market orders or marketable limit orders from customers. They make an agreement with a wholesale market maker like Citadel to fill all of those orders up to a certain size possibly, within the NBBO, in exchange for a fee and some price improvement requirements.
So imagine you're Citadel. You have a stream of incoming orders that you commit to fill at the current market prices with a bit of improvement. If the order flow is small, you just sit on the delta waiting for flow in the other direction or trade out patiently on exchange to make a spread. If the flow becomes largely imbalanced, you can liquidate the delta aggressively on the public market to hedge at a very small cost. Say the NBBO is 100.50 bid 100.51 offer and you just sold 500 shares at 100.5099 and bought 2000 at 100.5001. You can dump the 1500 residual on the exchange bid or in dark pools, whatever, and all you lose is the price improvement you gave and fees, which are smaller than the spread you make on a "good" trade.
In effect, the wholesaler is cherry picking and getting a cheap option to make a spread. All the small retail trades that are easy to capture spread on get siphoned off the exchanges, and they're left with only the imbalanced toxic ones. This is good for the wholesaler, good for the retail trader (he gets a bit of price improvement and lower trading commission), bad for the exchanges, and bad for on-exchange market makers like most HFT firms.
It's arguably "fair" to price uninformed flows differently, it's very common in some markets like FX, but locking that up behind opaque pricing agreements where only a handful of firms participate isn't. Most market making firms would gladly compete to offer even more sub-penny price improvement to retail trades, and there are even a few exchange programs that tried to do this but never got much traction.
I think we are in agreement that the main purpose of buying order flow is to price discriminate based on delta toxicity.
Hate to link to the Themis blog, but they cite the SEC here describing how internalizers, not retail brokerages, routed sell orders out to exchanges on May 6th: http://blog.themistrading.com/2010/10/the-internalizers-and-...
Does the practice described in this article and what you are discussing ultimately have a negative impact on Joe Investor who is taking the Bogleheads approach and sitting on several index funds, dollar-cost averaging his contributions, and occasionally rebalancing (but ultimately following a "buy and hold" approach)?
Could the outcome of this come back to hurt the Joe Investors of the world?
As a regular investor who is making his way through "A Random Walk Down Wallstreet," it is hard to look at how Wallstreet has behaved over the years and not feel paranoid that I'm being taken advantage of at every turn. Right now I like the Bogleheads approach to investing, and the logic behind it, but am interested in how even that approach is potentially being taken advantage of, and what those real risks are to investors of that philosophy. Or...is that even something to be worth getting concerned over and is there a bigger threat I should be focused on learning about?
Wholesalers make things even better for the retail buy and hold guy by filling them inside the bid/ask spread. The kickbacks they pay to brokers mostly translate into very low brokerage fees. Now you can trade stocks for a few dollars at a penny spread but it was $50 or $100 in commission years ago, with spreads of eighths or quarters, which can really eat into your savings when you're investing a small amount.
The main things I'd say to be wary of are:
a.) Trying to compete with professional intraday traders. Even if you have a good feel for the market, someone faster and smarter is going to beat you to the punch most of the time. The markets are much, much tougher for day traders, and the complex structure makes it easier for sophisticated automated traders to take advantage of them. It's just a waste of time and money.
b.) Trading when the markets are in a crisis mode. Automated market makers don't have as much capital as the big banks did because there's simply less "juice" in the business, so they won't be there buying forever in a falling market. No market maker ever would, but when they were earning quarters instead of pennies, there was more incentive to take on risk because the market had a built-in edge. Computers also have a lot of safety logic that can make them leave the market due to bad data during extreme events. You should always set a reasonable limit price on your orders, avoid trading around volatile times like the open or economic releases, and avoid market stops or other order types that can make you trade in an illiquid market at the worst time.
c.) Putting your money into things that are either risky like individual "hot" companies where you're competing with the best hedge fund analysts or products that are inherently bad to buy and hold (leveraged ETFs, commodity/volatility ETFs often have time decay or lose money rolling futures).
So if I'm largely a buy-and-hold kind of guy with my investments, there's not much to worry about from these practices then it sounds like. I do still worry what the risk is for buy-and-hold investors flocking to index funds as that seems like it is ripe for a problem of some sort, but given the nature of what an index fund is, I'm having trouble identifying how that could be a big issue.
1. Send a limit order for 1000@$20 to my brokerage. <- This part I understand.
2. "Lean on your bid?" <- Not a valid message in FIX.
3. Do flips. <- Acrobats do this, not brokerages.
4. Make a profit
1. send limit order 2. broker sells said order to timberhill (ib) 3. timberhill places limit order on exchange -- let's call it the nonsense exchange since that's exactly what it is. 4. timberhill bids the same price or say 1/100th ahead of my order (leans on me) on real exchanges. 5. timerbhill takes the next trade at my price while leaving me unfilled (my order still hasn't executed even though it should have). 6. timberhill offers the shares i should have bought at price inside the current best offer.
it's funny; i checked your blog posts about hft, and your solutions are the opposite of what's required to end the bullshit. the market would be much better served as a real marketplace if it enforced the penny increment rule (brokers currently sub penny the fuck out of every order on the market in the name of phony "price improvement"), and it consolidated venues (most trading venues today are holding places for customer order flow to be ripped off in the manner i've already explained).
1. Send limit order to broker.
2a. If timberhill's resting order is at or better than the NBBO, your brokerage informs him that he just got a fill.
2b. If timberhill is worse than the NBBO, the order is routed to the public market and timberhill doesn't hear about it until INET/ARCA/whoever puts it (anonymized) in the quote stream.
Once in the darkpool they placed a resting order in that darkpool. I posted a link to IB documents elsewhere - I suggest you read them.
Doing it off-exchange for a minimal improvement is less clear to me. In some markets regulators have limited off-exchange trading to block trades or substantial price improvement as a way to incentivize displayed liquidity and price discovery. That seems more fair but also subsidizes informed traders at the expense of uninformed ones. Spreads tighten as market makers compete to trade with some retail flow, but informed spread-crossers also get cheaper trading with these tighter spreads since market makers lose their option to price differently.
This is contradictory, if retail is "dumb money" it's not indicative of the future direction of the market.
The basic deal with internalizers is that they just have to fill you subject to Reg NMS rules, i.e., at or inside the NBBO. The catch is that the NBBO is set based on the lit markets, where as a market maker you have to take flow from any counterparty. As a result the spread that gets offered reflects that built-in risk that some flow you get will be adverse. The internalizer on the other hand has the great deal that they can filter down their flow to retail only, which is on average much less adverse. This gives you some options - you can offer a tighter spread to entice business based on better execution quality, or you can use the extra headroom to pay brokers to drive volume to your business, or both.
Argument for internalizers: knowing the risk profile of their customers should let them improve execution quality (offer tighter spreads) over the lit market.
Argument against internalizers: transparency is king, if retail flow was driven to lit markets there would be less adverse flow there in general, and spreads would be tighten as market makers competed for the new flow.
As I work for a non-internalizing market making operation, I selfishly like the second option, but it's not really clear who is right. My guess is nothing is going to come of this investigation unless there some real shady stuff going on behind the scenes we don't know about - the basic premise of the business model is legal, even if maybe not optimal for the market.