This is from an article truetype posted at https://www.elitetrader.com/et/thre...pen-to-outside-investors.320553/#post-4645375
-----------------
“It comes down to turning data into ideas into investments,” Mr. Tulchinsky said. “When we started out with very few alphas, we were concerned with secrecy, but now that we have millions, no one alpha is important. Our edge is putting things together, it’s the implementation.”
The idea is that with so many “alphas,” even weak signals can be useful. If counting cars in parking lots next to big box retailers has only a tiny predictive power for those retailers’ stock prices, it can still be used to enhance a bigger prediction if combined with other weak signals.
For example, an uptick in cars at Wal-Mart parking lots—itself a relatively weak signal—could combine with similar trends captured by mobile phone apps and credit-card receipts harvested by companies that scan emails to create a more reliable prediction."
So quant trading is about stuff like this? Programming metrics such as the above into a computer to predict price movement? An increase in the number of cars in a big-box parking lot does not mean an increase in sales for that store. There are so many other variables involved unknown that cannot be known or "quantified".
Such as payroll on clock that day, the internal machinations of profit margin cuts due to overstocks, or even how many of those cars left the parking lot with a purchase. And then there's, "Well, how much profit did the store do on that day selling manure when it is sunny, compared with how many snow shovels it sold two weeks before when it was snowing?" Therefore, how much money did that store make in that month? How much money did it cost in manpower to shift the merchandise around?
And the idea that one could somehow scan e-mails to make a reliable prediction...What are they scanning, and whom are they scanning?
Dudes, when same-store sales are down period after period in the company as a whole, that is going to drive the stock price down, in general.
My mind is suddenly swimming with all the variables on why this "quant" scanning is a terrible predictor of future results. It boggles my mind that people are getting paid gobs of money to try to program this stuff.
It is here I would champion the true fundamentals of a company to determine future price. Not some algo that count cars. There are WAY too many variables to consider.
I am aghast. I thought algos were all about ekeing out a living on dark pools, not trying to predict a company's future performance based on shaky and circumstantial evidence. Yikes!
-----------------
“It comes down to turning data into ideas into investments,” Mr. Tulchinsky said. “When we started out with very few alphas, we were concerned with secrecy, but now that we have millions, no one alpha is important. Our edge is putting things together, it’s the implementation.”
The idea is that with so many “alphas,” even weak signals can be useful. If counting cars in parking lots next to big box retailers has only a tiny predictive power for those retailers’ stock prices, it can still be used to enhance a bigger prediction if combined with other weak signals.
For example, an uptick in cars at Wal-Mart parking lots—itself a relatively weak signal—could combine with similar trends captured by mobile phone apps and credit-card receipts harvested by companies that scan emails to create a more reliable prediction."
So quant trading is about stuff like this? Programming metrics such as the above into a computer to predict price movement? An increase in the number of cars in a big-box parking lot does not mean an increase in sales for that store. There are so many other variables involved unknown that cannot be known or "quantified".
Such as payroll on clock that day, the internal machinations of profit margin cuts due to overstocks, or even how many of those cars left the parking lot with a purchase. And then there's, "Well, how much profit did the store do on that day selling manure when it is sunny, compared with how many snow shovels it sold two weeks before when it was snowing?" Therefore, how much money did that store make in that month? How much money did it cost in manpower to shift the merchandise around?
And the idea that one could somehow scan e-mails to make a reliable prediction...What are they scanning, and whom are they scanning?
Dudes, when same-store sales are down period after period in the company as a whole, that is going to drive the stock price down, in general.
My mind is suddenly swimming with all the variables on why this "quant" scanning is a terrible predictor of future results. It boggles my mind that people are getting paid gobs of money to try to program this stuff.
It is here I would champion the true fundamentals of a company to determine future price. Not some algo that count cars. There are WAY too many variables to consider.
I am aghast. I thought algos were all about ekeing out a living on dark pools, not trying to predict a company's future performance based on shaky and circumstantial evidence. Yikes!
