8 ms·

There are several explanations for this phenomena: * The largest moves in price occur after quarterly earnings announcements, which are released after-hours. S

There are several explanations for this phenomena: * The largest moves in price occur after quarterly earnings announcements, which are released after-hours. Same with other material announcements. * Prices are more volatile after-hours because there are fewer market participants. Because the order book is smaller, the same sized trade will have a larger effect on price after-hours compared to during trading hours. * Some traders at hedge funds are limited on how much $ they can hold overnight by risk management. This is because there is less liquidity and it is difficult to unload if a position implodes overnight. To reduce risk, they sell end of day, and buy back after market open. * For the above two reasons, holding overnight confers more risk, which will be rewarded by the market with higher returns. * (personal speculation): The "positive returns for holding after-hours" effect reflects the general trend of the stock market. Stocks have trended upwards for the past several decades, so the effect is currently positive. If stocks were to trend downwards for several years, I'd expect this the after-hours effect to also be negative. I have not backtested this.

Panzer04 4y ago
Night is generally just longer too, so you're taking on more risk regardless of liquidity as well (more time holding the stock, more time for good/bad news, on average you get rewarded for that risk)
logicalmonster 4y ago
Another reason is that the circuit breakers that halt wild stock movements during the day don't apply in after hours trading.
magwa101 4y ago
[dead]
evrydayhustling 4y ago
Add to all of these that the return is uninvestable, so the market cannot eliminate the arbitrage. 1016x over 30 years is a gain of less than 0.001% per trading day. You would have spent much more than this through fees and market impact to buy and sell the stock each day, especially at any real volume. The market is full of uninvestable or tiny capacity trades like this. A reliable pattern gets created by trading flows like the ones described in other replies, and it doesn't get arbed out because it is uninvestable or has tiny capacity.
- dumbfounder 4y ago
  Imagine how many HFT algorithms are out there trying to exploit every micro trend to the extreme. I am not in that game, but I can imagine all the things I would try, and I know there are really smart people and obviously a ton of money invested in it. Trying to find something simple that consistently makes money for a long time would be nearly impossible because the patterns would be found and exploited by many until they scrubbed out any potential gains. (But then again if someone did find it they wouldn't share). So you need to get novel and pour all sorts of external data into your models to have any edge. Something this simple just wouldn't be a thing. But it is still interesting to understand.
  - evrydayhustling 4y ago
    Algo trading teams are indeed research shops that have to continually replace their strategies. Often times success is less about the cleverness of a prediction than clever engineering (technical or contractual!) that makes a known pattern investable. Or else about being earlier to identify emerging, short lived patterns so that you can profit from more of their lifecycle. Teams make long term investments in technologies and strategies that help them do these things repeatably.
  - rootusrootus 4y ago
    I think at this point the traditional stock market has so much algorithmic trading going on that making your own algorithm that can consistently win is basically impossible. But there are other less popular markets that don't receive the same attention.
    - jjtheblunt 4y ago
      What about finding a company with reliable business model and stable dividends and reinvesting those, instead of trying to do HFT?
      - rootusrootus 4y ago
        I don't think too many people would suggest algorithmic trading as a retirement strategy. Nor is investing in a single company a great plan, regardless of their dividend stability. I'm a confirmed Boglehead myself, but I do enjoy playing games with my lunch money. Options, crypto, or just individual stocks if I like the company. And yeah, it does help if they are a Dividend Aristocrat if I plan to hold that stock for any length of time. I've also considered trying some algorithmic trading at lesser known markets, because it is interesting.
runnerup 4y ago
> holding overnight confers more risk, which will be rewarded by the market with higher returns. That’s really not how risk works.
- kypro 4y ago
  Assuming an efficient market, that's exactly how it works - if there's more risk investors will require more reward to compensate.
  - runnerup 4y ago
    That’s a massive assumption. How efficient and rational does the stock market seem to you? With most of it driven by index funds and 401k contributions and even wall street traders jumping on bandwagons. Risk reduces alpha. Risk doesn’t increase alpha. Yes obviously in a perfectly rational environment there would be a “need” to reward necessary risk but the market definitely does not need to in this case. More than enough idiots will hold their positions overnight without any upside.
  - BurningFrog 4y ago
    I think the point is it doesn't work "in reverse". Taking huge risks doesn't in itself mean you'll, on average, be hugely rewarded.
- bongoman37 4y ago
  [dead]
uoaei 4y ago
Phenomenon is singular, phenomena is plural. The same is true for other words of similar form (criterion, etc.). I have an unusual sensitivity to societal grammar, spelling, pronunciation patterns vs the literal rules of language, and only recently has it been clear that most people who use those words use them very consistently incorrectly. This seems to have changed since I was a child. For all the talk about how the Internet's information educates us, we still need to acknowledge that people need to seek out and be receptive to that information.
- tristanj 4y ago
  Thanks, I'll remember the distinction :).
shawndrost 4y ago
Haven't you explained too much? Your robust explanation should hold true in 1990, 2000, 2010 but the phenomenon had a discrete beginning in 2009 (and arguably a smaller inflection point in 2002) per the graph.
- tristanj 4y ago
  The explanation is for the general phenomenon which applies to stocks part of the S&P500 https://finance.yahoo.com/news/buy-close-sell-open-114705554.html https://finance.yahoo.com/news/buy-close-sell-open-114705554... . The chart in the article only shows AIG stock.

Panzer04 4y ago

Night is generally just longer too, so you're taking on more risk regardless of liquidity as well (more time holding the stock, more time for good/bad news, on average you get rewarded for that risk)

logicalmonster 4y ago

Another reason is that the circuit breakers that halt wild stock movements during the day don't apply in after hours trading.

magwa101 4y ago

[dead]

evrydayhustling 4y ago

Add to all of these that the return is uninvestable, so the market cannot eliminate the arbitrage. 1016x over 30 years is a gain of less than 0.001% per trading day. You would have spent much more than this through fees and market impact to buy and sell the stock each day, especially at any real volume. The market is full of uninvestable or tiny capacity trades like this. A reliable pattern gets created by trading flows like the ones described in other replies, and it doesn't get arbed out because it is uninvestable or has tiny capacity.

dumbfounder 4y ago
Imagine how many HFT algorithms are out there trying to exploit every micro trend to the extreme. I am not in that game, but I can imagine all the things I would try, and I know there are really smart people and obviously a ton of money invested in it. Trying to find something simple that consistently makes money for a long time would be nearly impossible because the patterns would be found and exploited by many until they scrubbed out any potential gains. (But then again if someone did find it they wouldn't share). So you need to get novel and pour all sorts of external data into your models to have any edge. Something this simple just wouldn't be a thing. But it is still interesting to understand.
- evrydayhustling 4y ago
  Algo trading teams are indeed research shops that have to continually replace their strategies. Often times success is less about the cleverness of a prediction than clever engineering (technical or contractual!) that makes a known pattern investable. Or else about being earlier to identify emerging, short lived patterns so that you can profit from more of their lifecycle. Teams make long term investments in technologies and strategies that help them do these things repeatably.
- rootusrootus 4y ago
  I think at this point the traditional stock market has so much algorithmic trading going on that making your own algorithm that can consistently win is basically impossible. But there are other less popular markets that don't receive the same attention.
  - jjtheblunt 4y ago
    What about finding a company with reliable business model and stable dividends and reinvesting those, instead of trying to do HFT?
    - rootusrootus 4y ago
      I don't think too many people would suggest algorithmic trading as a retirement strategy. Nor is investing in a single company a great plan, regardless of their dividend stability. I'm a confirmed Boglehead myself, but I do enjoy playing games with my lunch money. Options, crypto, or just individual stocks if I like the company. And yeah, it does help if they are a Dividend Aristocrat if I plan to hold that stock for any length of time. I've also considered trying some algorithmic trading at lesser known markets, because it is interesting.

runnerup 4y ago

> holding overnight confers more risk, which will be rewarded by the market with higher returns. That’s really not how risk works.

kypro 4y ago
Assuming an efficient market, that's exactly how it works - if there's more risk investors will require more reward to compensate.
- runnerup 4y ago
  That’s a massive assumption. How efficient and rational does the stock market seem to you? With most of it driven by index funds and 401k contributions and even wall street traders jumping on bandwagons. Risk reduces alpha. Risk doesn’t increase alpha. Yes obviously in a perfectly rational environment there would be a “need” to reward necessary risk but the market definitely does not need to in this case. More than enough idiots will hold their positions overnight without any upside.
- BurningFrog 4y ago
  I think the point is it doesn't work "in reverse". Taking huge risks doesn't in itself mean you'll, on average, be hugely rewarded.
bongoman37 4y ago
[dead]

uoaei 4y ago

Phenomenon is singular, phenomena is plural. The same is true for other words of similar form (criterion, etc.). I have an unusual sensitivity to societal grammar, spelling, pronunciation patterns vs the literal rules of language, and only recently has it been clear that most people who use those words use them very consistently incorrectly. This seems to have changed since I was a child. For all the talk about how the Internet's information educates us, we still need to acknowledge that people need to seek out and be receptive to that information.

tristanj 4y ago
Thanks, I'll remember the distinction :).

shawndrost 4y ago

Haven't you explained too much? Your robust explanation should hold true in 1990, 2000, 2010 but the phenomenon had a discrete beginning in 2009 (and arguably a smaller inflection point in 2002) per the graph.

tristanj 4y ago
The explanation is for the general phenomenon which applies to stocks part of the S&P500 https://finance.yahoo.com/news/buy-close-sell-open-114705554.html https://finance.yahoo.com/news/buy-close-sell-open-114705554... . The chart in the article only shows AIG stock.