Comment by ChuckMcM | Hacker News Reader

Comment by ChuckMcM | Hacker News Reader

Gah what a stupid article. Consider its foundational point:

"Pragma measured the effect by comparing the volume in certain stocks with the time it takes to execute an order. Longer execution times typically result in poorer results, since a stock's price can swerve away from where it was when the order entered the market. Such an effect is known in the industry as a "shortfall.""

If you are a "long term" investor you don't sell stocks to capture a few pennies here and there. You buy at price $X and hold it for a while, maybe you put in a stop order [1] so that if the shares start heading for the floor you will automatically exit. You set a value you want to see for your 'gain' and you set a limit order [2] when the stock starts getting close. The limit fires and you exit the stock. Even if it keeps rising and rising.

The basis for the claim in the article is that some HFT house might buy your stock when it hits the limit order price, because it is predicting it will go higher and then instantly resells it for a bit more than your price. You've cashed out already (closed your position) and they skimmed a bit of cream off the top. You didn't 'lose' any money at all.

For those not familiar with stock trading:

[1] A 'stop' order tells the firm holding your stock that if the stock drops below a certain price (the stop price) to automatically sell the security. So if you buy a stock at $10/share and you don't want lose more than 20% on it you might set a stop order for $8/share.

[2] A limit order goes the other way, you tell the broker that if the stock ever gets to a certain price to sell your shares. So if you are looking for a 10% return on your $10/share stock you might put a limit order in for $11/share. (or $11.25 if you want the 'net proceeds' to be $11/share).

> If you are a "long term" investor you don't sell stocks to capture a few pennies here and there.

Yes, but I'm still certain that high frequency traders are siphoning value off long term traders when the latter have to transact.

I would be interested to hear the sequence that makes this true. Lets say I put a limit in on stock FOO at $32 and it fires and I get my $32 * n dollars. What did the HFT do to siphon value off my trade? Or off the stock for that matter?

The aggregate effect of HFT might change the amount people decide to put on orders, resulting in their buying higher and selling lower.

So, in the interest of disclosure, do you do HFT? You seem to have an interest in defending the notion that high frequency traders have no overall effect.

Per the disclosure request : I am not an HFT trader, nor are any funds I invest in managed by HFT traders. Perhaps this makes me clueless :-)

To respond to your comment though,

"The aggregate effect of HFT might change the amount people decide to put on orders, resulting in their buying higher and selling lower."

I'm trying to figure out the linkage. What mechanism would connect the order pricing from a long term investor with the activities of an HFT trader? HFT works in the first and second derivative space of values and long term traders seem solidly in the linear space. Can you construct and example where the activities of an HFT trader are both visible too, and influential on, the pricing targets of a long term investor?

Oh I do have some Bank of America stock (one of the stocks called out in the article) which I bought in 2009 when they were pummeled by the Countrywide fiasco and their inability to get another CEO, sold half of it for a 100% gain, (net 50% gain on the total investment) and have watched the remainder go up and down. I got my 50% return (and that cash is working elsewhere) and I've got some extra shares that I could sell for anything over $0 to add to that gain (although if you want to do the annualized computation it gets lower if I holder it longer etc etc) but holding that since 2009 pretty much defines a 'long' holding.

Basically you cam imagine a market with only long term traders, there is a certain amount of money being generated by companies which feeds the profits of the market.

Now add some HFTs, these make a profit (there is empirical evidence for this claim). Now money is leaving the market without going to long term investors. Thus the long term investors are making less profit.

Now the big unsubstantiated assumption I have used above is that HFT adds no value to the market. I do not feel I am in any position to argue about whether this assumption is good or not. The above argument is presented in the hope that it makes things clearer and points towards what I think is the fundamental point which needs to be discussed: "Do HFT traders add value of a market?"

I'm not involved in HFT, but I imagine the following scenario is plausible:

You place a limit order to buy FOO at $32

Someone else offers FOO at $31.90

A HFT algorithm buys FOO at $31.90 and immediately offers it at $32

You buy FOO at $32 from the HFT algo

So you have lost potential profit on the transaction even though you technically hit your limit price.

You place a limit order to buy FOO at $32

Someone else offers FOO at $31.90

At this stage, the matching engine observes that you want to buy at $32, and someone is willing to sell at less than $32. You trade directly with that person at $32.

It's actually illegal for any matching engine to match the $31.90 bid, they must cross trades at the NBBO.

  You place a limit order to buy FOO at $32
  Someone else offers FOO at $31.90

That scenario would result in a locked market, which can't possibly happen under RegNMS. The exchange that saw the offer for $31.90 is required to route-out to the exchange with a bid of $32.

  A HFT algorithm buys FOO at $31.90

That also can't happen. Even without RegNMS, the exchange's matching engine would have paired the value investor with the offer of $31.90, though the execution price would actually be $32. The HFT participant won't even see the ask price in this scenario.

I disagree. Most HFT is market making (passive trading with limit orders only), and they make most of their money from short-term and impatient traders, who account for most of the volume and repeatedly pay the spread. Long-term traders, by definition, don't trade very often so they're a small fraction of the total volume.

And its the market makers who provide the liquidity (with limit orders) for the long-term traders when they do decide to trade, not the short-term traders (who take liquidity with market orders).

I can't comment otherwise but I'd note that HFT is producing the illusion of market making without real market making.

By definition, HFT jump into liquid market and count on exiting illiquid markets faster than anyone else. That means that they don't actually bring a greater assurance that a trade will happen, in contrast to the traditional "market makers" of NYSE. The "flash crash" can be seen as a simple illustration of this but so could be the greater volatility we have seen in the last few years.

> The "flash crash" can be seen as a simple illustration of this but so could be the greater volatility we have seen in the last few years.

Which "flash crash"? The one about a year ago or the one in the mid-60s?

There are thousands of flash crashes a year in single stocks on the market, usually explained by algorithmic trading or mistakes but could also just be blind panic as the rational market adjusts to news, rumour and FUD. Usually these stocks are suspended and any problem trades reset.

Whether these are exacerbated by hft is hard to tell, but I can't see how it would help damp volatility. Probably they have more to do with momentum investors and algorithmic trading, though they can easily happen with only human trade too, just at a slower pace. Would be interesting to see stats on volatility going back decades, but it's probably quite a complex subject for a layperson.

Here are some examples: http://www.usatoday.com/money/markets/2011-05-16-mini-flash-...

I agree the liquidity seems to be an illusion. I imagine most HFT AI's have boundary conditions to stop trading once significant price changes occur to avoid losing money in unpredictable scenarios.

HFT market making is about being more informed about the true price at the micro structure level than your counter-party. Simple as that, no black magic.

If someone trades with a HFT market maker its because

a) they are un-informed about the current true price

b) they choose to optimize execution time over best price

Its the same market game that`s been played for decades, just with different players and tools.

I would expect long-term investors to be least affected by HFT. Can you provide a concrete example of how HFT would hurt a long-term investor?

>"I'm still certain that..."

It seems like most people are "certain" about the impact of high frequency trading on an a priori basis as well. If facts wouldn't be inconvenient, however, run a Google Scholar search on the impact of high frequency trading or look at Chris Stucchio's discourse on the matter [1].

Consider two markets, one with high-frequency and human market making and one with only human specialists against whom you trade. Run similar value-style strategies on both; if the former outperforms the latter portfolio you pay up $10 000 (or whatever number keeps this interesting for you), else I pay you.

Fact is you'd be silly to take that bet, as would I. We don't understand the impact of HFT enough for either of us to be "certain" of anything.

[1] http://www.chrisstucchio.com/blog/2012/hft_apology.html

Reply on news.ycombinator.com