Hacker News Reader

Comment by tomp | Hacker News Reader

I disagree with this description.

Retail flow is primarily uninformed - "noise", not "signal" (i.e. "alpha" in finance speak). Retail traders might trade based on Twitter news, Reddit suggestions, weather, gut feeling, sudden money needs, ...

Institutional traders are big and slow and they have long-term alpha (if any), I don't think they'd generate much worry for HFTs/market makers... The real worry is other HFTs, small, short term (well-funded, more rational) traders that do have "alpha". There, you're faced with negative selection - by trading passive (market making), you should assume that your order is more likely to be filled when the counter-party has better information (more alpha/signal than you) so you'll lose in the short term.

View on HN

That makes sense too (avoiding competition). But it's not what the quoted article says.

Institutional trading (like mutual funds) certainly is a problem for market makers:

> Sometimes, when a customer buys 100 shares at $100.01, it then buys another 100 shares at $100.02, and another 100 shares at $100.03, and keeps going until it has bought 10,000 shares and pushed the price up dramatically. The market maker who sold it the first 100 shares—and who is probably now short and needs to go out and buy those shares at a higher price—has been run over.

> This is a risk of being a market maker on the public stock exchanges: Sometimes you sell 100 shares to a small retail investor and it’s random noise; other times you sell 100 shares to Fidelity and you get run over. But if a market maker can guarantee that it will only interact with retail customers—if it can filter out big orders from institutional investors—then its risk of adverse selection goes way down.

Indeed, this isn't a comment on your summary (which I have no reason to not believe to be accurate). I'm commenting on the ideas behind the article itself.

Still, I disagree. Even a large order won't "move" the market by itself (unless you're the Fed). At best, it might trigger a flash crash (we've seen a few of these in the past few years), where liquidity temporarily dries up (HFT market makers remove their passive orders, until they figure out what the hell is going on, precisely because they already predict cases like this) but comes back as soon as human traders figure out that nothing is going on (and the price has no reason to move).

More likely, a large order might blow through a few layers of the order book, which is good for market makers - instead of selling it at $100.01 (which is the "best", i.e. "lowest" price), you're selling it at $100.02 (i.e. you're making more money than your fellow market makers). In general, if you're an "institutional" (or otherwise big & slow) trader trading "large" orders (relative to the standard market volume), you're trying hard to disguise your intents. If HFTs know that you will buy the next 100 shares, they'll just sell it to you for $100.09 or more!

This is not Japan, the FED is not trading equities.

And continuous movement of price in one direction is bad for market makers. Read the linked article, it explains why.

> Even a large order won't "move" the market by itself

Certainly, the market moves when two parties agree on a price, so one order doesn't move the market by itself; however, a large order in combination with normal market behaviors will result in the market moving higher.

When you buy the book, the market makers are going to replace their sell limit orders at higher prices. If you have a lot more shares to buy, you're going to buy out those too. If you've done a fair bit of buying this way, and market makers had been selling short, they're going to want to buy to cover, which adds to or sustains the price increase.

This price movement isn't great for the institution or the market makers.

Market makers would rather trade against retail investors -- they don't make a lot of large trades, so they don't move the markets very much, and there's not the same risk of getting run over.

Institutional investors would rather trade in ways they can get a fair price without influencing the price -- if they can find counterparties to trade with on a volume weighted average price basis, they prefer that -- or they try to structure their orders to avoid hitting the market all at once.

Does Robinhood sell data on various order types like limit, stop-loss, etc to High-Frequency Traders? That would allow the HFTs to extract profits out of Robinhood's users since this information does not appear to be non-public and hence can be used for front-running.

Reply on news.ycombinator.com