Quants Are Losing Their Minds...

[Option_Attack]

Jon Quigley says he probably should have known something big was coming -- even if his risk models didn’t.

Just a day after the Great Lakes Advisors manager watched CBS’s “60 Minutes” about America’s unprecedented efforts to deploy a vaccine when it comes, Pfizer Inc. revealed significant progress on its pandemic cure.

That revelation spurred the biggest moves ever in Quigley’s $3.9 billion portfolio. While stock benchmarks cheered the news, Wall Street’s most popular styles of quant trading got hit by a historic storm.

“Events happened that statistically never could happen,” said the chief investment officer of disciplined equities in a telephone interview from St. Petersburg, Florida.

Quigley spelled out the odds to clients in a note. As he computed it, the crash in the momentum factor was so rare that writing out the chances of occurrence on any given day required a 16-digit number -- followed by 63 zeroes...

https://www.bloomberg.com/news/arti...d-happen-hits-wall-street-models?srnd=premium

 

[Laissez Faire]

Definitely been some fat tails and n = 1 type of days this year.

 

[JamesJ]

Those quants were very wrong when they bought up NQ right after vaccine news, be it as momentum or spread trade.
There is just no historical data explaining the divergence between ES/ Russell/YM to NQ that day.
ES up, Russell Limit up and Nq deep red.

 

[Laissez Faire]

Monday opened up + 3,96 % or 138,75 points on ES. That's a fairly big gap.

However, since 2005, we've had 4 opening gaps which have been greater than that. 2 of them occurred this year. The last one on the 24th of March (+ 5,86 % on the Open).

Maybe the quants need to re-write their models after 2020.

The chart below shows the distribution of opening prices since 2005 with Monday highlighted.

 

[thefuturestrader]

Nice chart, what did you use to make this? R or Python perhaps?

 

[JamesJ]

Monday opened up + 3,96 % or 138,75 points on ES. That's a fairly big gap.

However, since 2005, we've had 4 opening gaps which have been greater than that. 2 of them occurred this year. The last one on the 24th of March (+ 5,86 % on the Open).

Maybe the quants need to re-write their models after 2020.

The chart below shows the distribution of opening prices since 2005 with Monday highlighted.

View attachment 244078

If you look at ES-NQ spread on open it definitely is many standard deviations off.

 

[Laissez Faire]

Nice chart, what did you use to make this? R or Python perhaps?

Thanks. I made it in Excel.

If you look at ES-NQ spread on open it definitely is many standard deviations off.

So was Monday's Open. I don't track the ES-NQ spread, but I ran some additional numbers on the dataset I posted. It seems like Monday's Open was about 5 standard deviations from the mean.

Note: This data isn't quite normally distributed, though. Too many fat tails.

 

[Cuddles]

“Events happened that statistically never could happen,” said the chief investment officer of disciplined equities in a telephone interview from St. Petersburg, Florida.

Quigley spelled out the odds to clients in a note. As he computed it, the crash in the momentum factor was so rare that writing out the chances of occurrence on any given day required a 16-digit number -- followed by 63 zeroes...

https://www.bloomberg.com/news/arti...d-happen-hits-wall-street-models?srnd=premium

 

[Russell Shuffle]

lol... yeah I kind of laughed to myself when I read that quote. Anyone who has been around a minute knows the next Black-Swan is right around the corner, it will be worse then the one before it and not as bad as the one that follows.

 

[NoahA]

Those quants were very wrong when they bought up NQ right after vaccine news, be it as momentum or spread trade.
There is just no historical data explaining the divergence between ES/ Russell/YM to NQ that day.
ES up, Russell Limit up and Nq deep red.

Isn't this always the issue? Data doesn't explain anything. It's up to the human brain to understand the data.

It makes perfect sense why the ES would do much better than the NQ after you learn about the rotation out of tech stocks, used as a proxy for "stay and work from home" businesses. But how do you program an algo to be aware of this? Certainly you can program in the ability to check other markets and see how they are moving, but ultimately, this move must have been human operators just using their algos to either buy or sell, and knowing which to turn on in which market.