During my usual back-to-front reading of this week’s Economist I happened upon this article in Free Exchange. This week’s study was a critique of high-frequency trading (HFT), with the allegation that HFT reduces liquidity; a solution from the study authors is discrete pricing intervals during market hours.
In lieu of making a comment on Economist website – somehow I was kicked out when I tried to post – I’ll put my thoughts below.
- The article cites “Flash Boys” and the study for the HFT strategy front-running. The description in the article defines the term wrong:
- First, front-running requires customer orders; HFT have none.
- Second, the strategy being described is much more like latency arbitrage. Just as Rothschild’s messengers gave an information advantage during the Battle of Waterloo, or pit traders used buddies at other exchanges (via phone) to communicate market events, today’s HFT are able to use price changes at one exchange to trigger orders at another. Nothing nefarious; just the same as before but at higher speed.
- The ‘investor’ in the article seems to be the same as the ideal modelled by other papers: that is, some organisation/person which requires liquidity in large size, but has invested for reasons other than short-term price movements (e.g. long term investors or pension funds). The latency arbitrage described above, and depicted erroneously in the article, doesn’t hurt the investor. Let me explain:
- Long-term investors require more out of winning trades than a tick or two. That’s the margin of an HFT engaged in latency arbitrage, or indeed a market maker setting the bid/ask spread. The spread is the cost of taking liquidity from the market maker. If the long-term investor really cares about that tick, they should probably just lower their fees by around 1 or 2bps for their investors.
- If the investor needs a lot of liquidity immediately, they will pay for the privilege by taking out multiple bid/ask layers in the order book. Instead of the 1 or 2 tick spread, the investor will pay perhaps 5 or 10 ticks. Seeing as pre-HFT market makers were routinely pricing bid/ask spreads in the 12.5 – 25 tick range, the investor is still better off with the HFT market maker.
- This may go some way to explain why Vanguard, one of the world’s largest investment providers, is OK with HFT.
- The article and paper cites lack of market depth for a reason to curb HFT. But with today’s technology, orders can be issued, cancelled and replaced nearly instantaneously. That means keeping orders in the deep book (e.g. away from the National Best Bid/Offer) is inefficient (as brokers frequently require margin to be posted to cover the order) and unnecessarily risky.
- The risk comes from the article’s depiction of ‘adverse selection’: if I have a choice between posting offers at 5 consecutive price levels, I’m going to be very mad if some investor sweeps the offer book and I get filled on all five. I will probably lose money. Instead, I’ll post an offer or 2, then post other offers as I see the price move.
- So though the market seems not to have depth, a big order will still likely be executed at a near-best bid/offer, as HFTs come to make markets. Again, better than before, where the book depth was all there was – no one could react fast enough to get a big order filled without a huge margin for the market maker.
- Finally, I’m not convinced by so much moralising over the wasted money on lightning-fast speed. Most authors with a beef on HFT call this a waste of money; I could say the same about any purchase I wouldn’t make. The fact is this ‘arms race’ frequently cited is actually good for everyone but the HFTs that must continually spend the money, slowly putting themselves out of business. It’s a bit like the recent OPEC announcement, keeping production high: the only ones really hurt by the announcement are governments which rely upon oil revenues to keep finances in check. The rest of us get an early Christmas gift.
The paper’s authors attended a CFA conference weekly, and discussed the topic. A summary is here. Similar to other calls to curb HFT, the solution seems awful woolly; particularly for getting rid of a system which generally works, but would require wholesale changes to obtain the desires of opponents.