7

I am wondering about implications of the speed of intra-daily trading on the wealth of an long-term investor. I am not necessarily asking about the costs of HFT trading to society but instead I wonder whether intra-daily price fluctuations or, in the extreme case, flash crashes should concern an investor who aims at holding a portfolio for a long time.

I can imagine that the following factors are certainly important:

  • Direct effect when setting up the portfolio (Costs of liquidity, e.g. Spread, Market Depth, Market impact are directly affected by high-frequency market microstructure)

  • Increased co-movement due to HFT-Traders: Serving as liquidity bridges between natural buyers and sellers, HFT-Traders may hedge their (market-maker like) positions with correlated assets and therefore may increase correlations in general which has an effect on the optimal portfolio structure (see, for example this excellent book).

Obviously, intra-daily price movements determine the end-of-day price, and therefore the distribution of long-term returns depends on the intra-daily price fluctuations. However, fast price movements followed by recovery such as during the Apple flash crash on BATS is probably nothing, I would care about if I added some stocks to my retirement plan - isn't this myopic and instead we should be worried if flash crashes occur?

Community
  • 1
Stefan Voigt
  • 1,456
  • 13
  • 29
  • An investor can fit a linear curve to their potential asset's security's value and invest according to its trend. The number of observations fitted should roughly correspond with the length of the term the investor is interested in. – K3---rnc Feb 07 '17 at 18:03
  • Thanks for your comment, @K3---rnc! I am not sure how your remark addresses my question: Do you refer to the need of long time-series to come up with reliable estimates of future volatilities? – Stefan Voigt Feb 08 '17 at 20:55
  • I guess it dabs at implying any short-term fluctuations that don't significantly re-slope your long-term linear curve are not to be concerned with. – K3---rnc Feb 08 '17 at 23:07
  • Oh, and I think you can avoid flash crashes by setting a reasonable stop loss. – K3---rnc Feb 08 '17 at 23:09
  • @K3 stop losses should never be used if you are worried about flash crashes which quickly reverse themselves. A stop loss will sell you out at the worst possible time (i,e when the price is temporarily down). – Alex C Jun 10 '17 at 17:03

1 Answers1

2

I agree with the implicit idea behind your question that "on the paper, high frequency fluctuations of prices should not affect long term moves". One point is for sure: the volatility we have in mind when we talk about Value At Risk and similar systemic measure has nothing to do with the potential increases of volatility due to high frequency activity.

Nevertheless:

  • the cost of trading affects all investors, hence the value of a position in a portfolio. And trading practices affect this cost, via the bid-ask spread, fees (that can be designed for HFT only), post trading costs and market impact (i.e. liquidity resilence)
  • market impact affects correlations and as you underline it should matter to portfolio managers. But where is the causality? Asset managers changing the correlations via their correlated trading flows or the reverse?
  • I have heard that: "big problems like the flash crash affect the confident people in general have in the price formation process. It is not good and it can lead to less people investing, hence less liquidity". I am not fully convinced because I dream people are more rational than this...
  • Last but not least, you can imagine a very bad configuration for the 2010 flash crash. Remember it was between 2pm and 4pm NY time. Hence no other markets (Europe nor Asia) were open. Imagine that the worst point (30min after the start) would have happened a Friday, just at the close of European markets... Add than via correlations, European markets would have closed at -10% because of this... Most probably all positions in banks' books would have had to be marked to market prices on Friday night, and hence it could have been the start of a series of fire sales to cover VaR... Remember it was the start of rumours about the status of Greece debt...

This last sequence is a very rare event, but it can happen, and we should try to prevent all these bad points. What are the solutions? some ideas:

  1. reduce operational risk we need clear procedures to put in production trading systems (from exchanges' matching engine to HFT systems and brokers trading algorithms). With need norms and certifications.
  2. improve circuit breakers to stop trading when needed, and connect (to some extend) circuit breakers of different venues.
  3. put all possible liquidity on connected platforms to have as much liquidity as needed.

(by the way I like the book too...)

lehalle
  • 12,064
  • 1
  • 49
  • 89