Debunking HFT Myths

I have seen an increase in media attention to high frequency trading lately, following the so-called tweet crash, an article in the Wall Street Journal and sudden price spikes in GOOG and SYMC. Negativity abound. HFTs are stealing your money. HFTs liquidity is fake. Etc. Etc. Etc. We have heard it all before. Please allow me to debunk some of the myths regarding HFTs using layman’s terms so that next time you read or write an article you can put things in perspective.

HFTs do take risk

All strategies carry risk and involve winning and losing trades: in straightforward market making there is a risk of an adverse price movement, in ETF vs underlying arbitrage there is execution risk, pairs trading and statistical arbitrage strategies have correlation risks and so on. The view of HFT firms enjoying a free lunch is wrong; they just try to be right slightly more often than they are wrong – and repeat indefinitely.

Front-running of orders

Contrary to popular belief amongst HFT bashers, no-one can see your order before it reaches the exchange and buy a stock at a lower price ahead of your bid. On the other hand, if you decide to show your hand by resting orders in the market rather than crossing the bid/ask spread then that is a choice you make. You risk the chance of the market running away from you for the potential payoff of getting a better fill. 

Cross-exchange arbitrage opportunities

Buying a stock on one exchange to sell it on another exchange for a penny higher is a perfectly logical strategy and is no different than Starbucks selling you a bottle of water that they bought at CostCo next door. HFT firms take advantage of the ignorance of the counterparty not to use the optimal venue for their trade.

Quote stuffing

A strategy in which a firm sends out an overload of order messages to slow down their competitors’ processing speed at the exchange. The effectiveness of this strategy is debatable as it often tends to slow the order sender down more than anyone else, and well-designed exchange systems won’t be impacted. Exchanges are cracking down on this behaviour as it has no positive impact on price discovery or market liquidity – which makes perfect sense to me. The impact on all non-HFT trading is negligible – no retail trader will ever get hurt from HFTs stuffing quotes.

Dark pools aren’t evil

Dark pools are essentially private silent auctions.The benefit they have is that normal exchange price rules don’t hold up so that investors who think sub-penny spreads are still too wide can get filled at a sub-sub-penny price. All determined by the fairest economic principle of all: the auction. I am struggling to understand how people can be against dark pools. I am struggling even harder to understand how people can be both against HFT and dark pools at the same time, almost by definition someone against HFT should be in favour of dark pools.

HFTs often claim to lower spreads (true) and provide liquidity (true, most of the times). They say that as a justification of their activities to the public; everyone knows they are in the game to make money – oh wait… just like every other trader. HFTs just invested some more money in technology and employ more advanced trading strategies. Sure, it is silly to send tens of thousands of order messages across exchanges in a 5 second window, but if the HFTs think they can make money of it, then let them: it’s not hurting you, is it?

I invite anyone who has a compelling argument against HFT and/or can provide specifics on how retail traders are at a disadvantage to post in the comments.



This is not a loophole

The WSJ attempts to describe a loophole in the CME:

‘Firms can use their early looks at CME trading data in several ways. One strategy is to post buy and sell orders a few pennies from where the market is trading and wait until one of the orders is executed. If crude oil is selling for $90 on the CME, a firm might post an order to sell one contract for $90.03 and a buy order for $89.97.

If the sell order suddenly hits, the firm’s computers detect that oil prices have swung higher. Those computers can instantly buy more of the same contract before other traders are even aware of the first move.’

Once a trade happens, the CME has to send out three types of messages: 1) it has to send a trade confirmation message to the buyer and the seller, 2) it has to send an update of the last traded price to all market participants, 3) it has to update the order book for all market participants.

Common sense implies: one message has to be sent before the other. Like most exchanges, the CME chooses to send the trade confirmation first, then the last traded price and then the order book update. The so-called loophole is that the participants in the trade are able to deduct from their trade confirmation messages what the new bid/ask spread is before the order book is sent out to all market participants – gaining maybe 1 millisecond on their competitors.

As a result, using the example above, they might be able to bid 90.00 before their competitors and hope to lock in a quick profit. Obviously, there is absolutely no guarantee they will get filled at that level.

That’s it. Now what is all the fuss about?

Impact of HFT regulation on markets

Financial Times today:

EBS, one of the two dominant trading platforms in the foreign exchange market, is suggesting scrapping the principle of “first in, first out” trading, which it says gives an unfair advantage to the fastest computers and has led to an arms race of spending on technology.

Instead, under the plan, incoming orders would be batched together and dealt with in a random order.

For now, let’s ignore the question whether any regulation is necessary, and discuss the impact this particular change of rules will have for the market microstructure if they were accepted universally.

Liquidity providers, i.e. market makers, will have a harder time figuring out their priority in the order book and may be exposed to pick-offs of resting orders in the case of a news event – depending on whether cancellation messages are prioritised over trade messages by the exchange. Bid/ask quotes can only widen based on these new regulations.

Trade agressors, i.e. market takers, will have less benefit of an advantage in speed. Any self-respecting trading firm can get their orders to the exchange in milliseconds, so when multiple firms run similar strategies (and they certainly do), speed won’t be the deciding factor anymore. These regulations are effectively discouraging competition and will lead to a more level playing field among HFT firms and thus to a slow-down in the technological arms race.

It is up to the regulators and exchanges to decide whether the wider bid/ask and the forced reduction in competition justify the cost savings. Personally, I am all for free markets and wouldn’t regulate when it isn’t needed – but this proposal is as feasible as I’ve seen. It’s certainly a less ludicrous idea than introducing a financial transaction tax or a minimum holding period.

Please stop blaming HFT for the Flash Crash that never was

After the tweet-hack incident last Tuesday caused a one percent down spike in S&P futures, it was only a matter of time before the perpetual pessimists at Zero Hedge, ignorant NBA team owners and retail punters would blame high-frequency traders for all that is unfair in the current marketplace. Nothing new there. What caught my eye is that some of the more nuanced and respected bloggers were clinging on to this idea too: the Flash Crash of 2013 was born. Josh Brown at The Reformed Broker mentions:

‘[HFTs] provide a glass of water in a monsoon, but when you really need them to absorb sell orders, they’re on the sideline, buffering. It’s a joke.’

And I’m not even sure what Howard Lindzon (founder of StockTwits) exactly means with the following sentence – and who he wants to punish – but it’s safe to say he’s not happy:

‘I don’t like the idea of hackers and market manipulation, but if this is what it takes to get the algo’s away from the machines with their bullshit bids and false sense of liquidity, all the better. We need to punish these type of hackers with as much speed and breadth as we can.’

After the release of the infamous AP tweet, 175 million twitter users can read that the White House has been attacked and President Obama might well be dead. Every prop desk, hedge fund and anyone with an e*trade account and a twitter feed is one mouse click away from selling the S&P future. And somehow, at this point, it is expected of HFT firms to keep their bids in the market and ‘absorb the sell orders when you really need them’? I would like to call double standards on this one. For all we know, the leader of the free world has been attacked, and simply because HFT trading firms are ehm.. HFT firms (?).. they are supposed to buy your S&P futures? Using technology and social media to gather information is great, but acting upon new information to avoid trading losses is ‘a joke’?

Regarding that disappearing liquidity: have another look at that much referenced Nanex chart. What commentators seem to ignore is that people who really wanted to get their trade done managed to do so, as indicated by the spike in trading volume. The 1% up and down move was on the back of 300,000 contracts – almost 20% of the day’s volume in a matter of minutes. Seems like decent liquidity to me. Sure, the slippage may be bigger than usual, but oh yeah, let’s not forget that THE PRESIDENT HAS JUST BEEN ATTACKED.

Bloggers and journalists seem to idealise a sort of utopian past in which all trades got done at mid-market, with unlimited levels of liquidity at any point in time, and a complete lack of market manipulation – a world that supposedly ended with the rise of the evil HFT firm who does nothing but stealing your money. Think about the ultimate carnage that would break out on the NYSE floor after the release of such a news message. No floor trader would honour their bids and everyone will scramble to get short – most likely against their own clients.

There is nothing else the HFT guys could have done other than pull their resting orders. Any real-person trader would have done the same. Even though HFT algos are programmed by Physics PhDs, their thinking isn’t very different to those old school floor traders. It’s mainly the execution that has evolved, the basic market making strategies not so much. It is pretty easy to program the behaviour of traders in calm market surroundings as their instincts are fairly predictable. However on unexpected, never-seen-before occasions as these ones all predictability goes out the window – and so does the algo. Innovative human thinking has to take over at this point; which it did. That is why the S&P future went down when everyone thought a disaster had happened, and why it came back up when it was clear that there was no disaster.