Hacker News new | ask | show | jobs
by hftrader998 2164 days ago
I work in the industry and these kind of articles are always full of bad information about order routing.

* Robinhood order flow is informed and toxic like all other brokerages. Taking the opposing side of all Robinhood trades would cause a broker-dealer to lose all of their capital very quickly.

* The "bad prices" the "novices" are trading at, are in fact, the same market price that all participants trade at (at or inside the bid/offer). If the prices were obviously bad, there is free money available to the author here by simply quoting inside the spread.

* Recall that the majority of trades on lit exchanges are from professional or institutional investors. For this reason, spreads are wide because providing liquidity means you will likely get run over. Robinhood orders do not exhibit as much short term momentum, and so trading against them is safer for broker-dealers because there is less risk. This risk profile is valuable, and you might wonder what's a fair way to allocate that value. One option is to not capture it, and send all Robinhood orders directly to the market. The author implies this makes sense (a gravy-free approach), but it does not, the retail customer actually ends up worse off. Another option, the one that occurs in practice now is for the value to get split between the counterparty taking on risk (Citadel, in the form of less toxicity on orders), the customer (the Robinhood client, in the form of price improvement over the national bid/offer), and Robinhood themselves for sourcing the flow (a commission or payment).

6 comments

> Taking the opposing side of all Robinhood trades would cause a broker-dealer to lose all of their capital very quickly

Why? This is literally the definition of order flow purchasing and market making. Flow amidst spreads creates profits.

The non-cynical explanation for Robinhood’s flow being attractive is in the law of large numbers. Robinhood’s trades are tiny. That means buying their flow gives one lots of small, idiosyncratic exposures. Institutional flow, on the other hand, is lumpy, which can leave one with a few giant positions.

To add to what you're saying about less chunky risk...people in the market-making industry also prefer to cross with retail customers because there is less negative selection bias. Retail customers are generally less sophisticated traders and they are less likely to trade fast alphas, insider information about pending news, etc.

That said, there's money to be made in providing liquidity on chunky trades, as long as the price is right.

He doesn't mean taking the opposite side and immediately offloading the risk in the market with a small bid-ask.

He means taking the opposite side as a principal - like many spread betting companies do. Which is dangerous because not all Robinhood clients are small and uninformed.

Also, as the OP believes, Robinhood's flow is being induced, i.e., manufactured by Silicon-Valley-style maximization of user engagement. User engagement is the raw material necessary for creating as much order flow as possible for sale to Wall Street firms. Order flow from engaged users is the product.
Would you mind explaining a few of those terms? "Informed" and "toxic", specifically.
If you're a broker/dealer, you spend a lot of your time facilitating other people's trades. By "facilitating" here, we mean that you don't only put their trade on the market for them, you often trade with them "on risk" by taking the other side of their trades and then unwinding them. So if you want to sell 1000 tesla I might (as a broker/dealer) just buy them off you and look to sell them myself later either at market or directly without touching the market as part of executing someone else's trade later. This is more efficient for everyone as it means you don't pay the exchange fees for those trades you can "cross off" against a colleague or another client and therefore can offer a slightly better price to the end client. It also allows traders to manage the market impact of big orders more effectively, allowing large trades to be completed without moving the market as much.

If you're large enough, your client base represents the market generally. That means your client base by definition doesn't outperform the market (ie has zero alpha). So that means that facilitating their trading earns you the commission and the trades you have to unwind have net zero alpha. This is not entirely true because it ignores some important effects around how commissions work etc (which end up meaning that broker/dealers are structurally long the market in general) but is not false enough to matter for the purposes of this discussion.

Imagine you had one client who knew the future (ie every trade they made would make them money). By definition taking the other side of that trade would therefore lose you money. Their orderflow would be "toxic" - by trading with them you would always lose money.

When someone says that orderflow is "informed" what they mean (usually) is that the people making the trades have more information than the rest of the market and therefore will trade when beneficial to them which is likely to be net/net not beneficial to you (if as a broker you're on the other side of the trade).

Now, whether or not robinhood order flow is on the whole informed or toxic is another question. Personally I would be surprised if that turns out to be true but I could be wrong.

The "large enough" argument does not work if you explicitly or implicitly filter or stratify the input to your sampling function. Just like law of large numbers does not work for heavy tailed distributions directly.

In which case your client do not represent the market, either by volume, or by number.

If you can actually know in which ways your sample is different, you can play it then against the market.

Robinhood would optimize its input for small clients with less knowledge, or clients looking for highly speculative risky plays of smaller volume. These are very specific characteristics that likely differ from general market.

I'm not talking from the perspective of RobinHood, I'm talking about the broker/dealers who provide execution services to RobinHood. RobinHood is nowhere near large enough to have a client base which represents the market. JPM, GS, MS etc are.
"Informed" in his context means based on new research/news/information, implying that the market is more likely than not to subsequently move in the direction of the trade.

A "toxic" position is one that has a high chance of moving against the holder. Taking the opposite side to informed trades might be toxic, but taking the opposite side to the first buy trade in a series of a hundred by a big mutual fund company who decided they like a stock is definitely toxic because they will continue moving the price higher with each trade -- so the brokerage doesn't want to keep these short positions on their own books but source them in the market.

> Another option, the one that occurs in practice now is for the value to get split between the counterparty taking on risk (Citadel, in the form of less toxicity on orders), the customer (the Robinhood client, in the form of price improvement over the national bid/offer), and Robinhood themselves for sourcing the flow (a commission or payment).

According to the WSJ, compared to other brokerages, Robinhood disproportionately takes that value to themselves [1]. Part of the reason why it's so lucrative for them is that they're steering their clients to options trading, where the incentive payments from the market makers are much higher. I think this quote from the WSJ tells the whole story:

One executive with a high-speed trading firm that executes orders for Robinhood said its price improvement is much worse than that of competing brokers.

[1] https://www.wsj.com/articles/why-free-trading-on-robinhood-i...

Question, does RH (or other brokers) do internal clearing/matching of orders?

That is, they clear the purchases/sales internally if they can, going to "the stock market" only if they can't fill it? Or that is a big no-no for brokers?

Yes that is perfectly legal and happens, but I’ll try to be brief and explain why that wouldn’t happen usually and why it’s still a good thing. When you go to buy/sell a security you see a buy/sell price. Everything being discussed here pertains only to price improvement. That is, payment for order flow only happens if the offer is better than what you saw on the screen when you said buy/sell. When you go to sell/buy a security you’ll notice that there are different prices to buy or sell. Filling an order internally means that the difference in those 2 prices is entirely bore by your own customers, versus the ability to go to broader markets and improve outcomes for both of your customers.

Payment for order flow and internal clearing are 2 ways to execute out of a ton of options that a brokerage has. The specifics are seen as moats / “the secret sauce” for these firms so you’re not going to see anyone spill the specifics of how they get price improvement. But there is a ton of legislation & complexity around it for sure.

They could do that, but it would be detrimental to their business. They (or their partners who pay for order flow) are better off filling the selling customer on the bid, and the buying customer on the offer. If this sounds unscrupulous to you, remember that the market-maker and order-router have a right to make a profit from their business, and wouldn't be able to operate without making a profit.

Either way, they would have to print the trade to the tape, so it's part of "the stock market" regardless.

> If the prices were obviously bad, there is free money available to the author here by simply quoting inside the spread.

Bam. Nailed it here. Instant credibility.

A pity your HN account is "new" (green color), as I would love to hear much, much more from you.

I’d take the opposite side of the trade on several things. Hertz being one of them.

Also, I think you mistook the meaning of “overtrading at bad prices.” Buying Hertz at any price in June was overtrading it, and at a bad price to boot.