I must say right away that calling myself a programmer or an expert in machine learning does not turn out to be my language, let's just say - I program better than 90% of traders and understand trading better than 99% of programmers and datascientists. This is not to say that I am such a fine fellow, but rather to the question of what hole of misunderstanding exists between the fields of knowledge, which I will try to eliminate a little.
I have a blog on a trader's site, where I describe my approaches to screw ML to trading. Despite the fact that I myself am a very beginner in the field of ML, frankly, I do not often see relevant reviews, because 90% of practicing traders have only heard about neural networks and have an idea of it as a pink pony. Equally, when I see some pure mathematician or programmer trying to implement his knowledge in relation to the stock market, I often start bleeding from my eyes.
2 , , , . . , : " - ". ( , , , ). , . , , IT ("?! !"), ML . , , , , - . . ML ? , , , , , . . (!), , , , .
? , - , - 10 ( - , , ), , , , . . ? " , , , ". , . , , , . . ? , . , - ?! - , , - - .....? , - " " , .
" ". . , 50% . 50/50. , , .
? , 5 . , - 5 . . - 5 , 5 ? . , , . +3%. ? , , , . . - 10, 15, 5. 15. .
. , " " , . - , . , - " , , ". , , . , " " : " ". "": "?! ?! ?! ?! ?!... ?! ?!". ( ), , , , - 1 + 1 = 2, , . , . . , , , , 50/50. ? - " ". 50/50, 2 2 , 100% . . "" . , , . . 10% 20% , +10% ? , ? . ? return , ? . , . , , , , , , . 2008 . , . 2009 , , . ? , , . ? , . ( ), . , . , , , . . . . ? , , . . , , .
datascientist ? , - , , , ( ), . "... !". . , ! , , " , !". ? , , ?! . . , , . - - , , , " ". . , , . - , . . , , , , , , . ( ). , - . , , . , .
10 , , , - , . , 20 150 0,5%, - ( ). 150*0,5, , , 0,15% , 50% . 150 30 , 150*(0,5-0,15) 30*(0,5-0,15), , - 30*(0,25-0,15) . datascientists . "" , datascientist , , , /. . datascientist , - , , - , , , , . , . , .
, ML, !
ML:
. datascients . , datascientes , . ? , . , 50/50. .
. 80% , , , , , - . ? - 250 . " ? ... ?!". , , , . , , , . , 255. Datascient 1. , 0. , , , . , 80% "", , , . feature engineering.
, , . 2 , , . - , , - " ". , sec.gov, 10-, , 10 , XBRL. financials.morningstar.com, 10. selenium.webdriver .
. " - " , , " " . , , D/E (Debt to Equity) , , , ROE (Return On Equity) - . . , . , . T T+1, . - , . , ? , , ? - , , / .
, , S&P500, . c profitabililty, growth, cashflow, financial health. : Cash Conversion Cycle, Inventory, NetMargin%. Accounts Payable - . , , , , , , . Accounts Payable - . ? . . - Inventory. , , 3 , 20% .
In the appendix to the American stock exchange, I tried to find something interesting in the analysis of sentiments in reports 10 - K, 10 - Q, assessed the usefulness of the patterns, something was interesting. And macroeconomic fundamentals. The CLI indicator seemed interesting, the technical issue is its availability.
I hope someone learned something useful from this text.