« first day (3375 days earlier)      last day (1942 days later) » 

23:00
Have any of you guys heard of dwarf fortress?
Not sure what ebnf is
Extended Backus Naur Form
It's essentiall:
$A \mapsto aBC | A$
$B \mapsto cc$
Defining a language of strings that is acceptable in the language you're defining
A grammar is an intuitive and compact way to specify an infinite language.
Then there are programs called parser generators that take their own version of EBNF as input, and create part of a program that does the parsing for you.
Which is reading in and validating user input
Expecting a certain form, or erring when the form is not met
Huh
Let me look at that
Yes, you can define even your own programming language
People usually create DSL's or domain specific languages for an app they're righting. The language then extends the capabilities of the app
*writing
I don’t think I understand it man. Are you saying your combining all the languages into one by interpreting each different one into a mega one.
Me, what I'm doing is I need to determine whether stuff like $a_1$ is entered etc
but when you're designing something, you tend to generalize
so what if $a_{i + j}$ is entered and so on
I need some classes to represent expressions and variables
Then the Expr class will have a .variables() method that recursively determines all occuring variables
then I can do math on the expressions
At least variable substitution
But first I need to get the text into the form of an object hierarch
y
23:11
Is it possible to conbine all the languages into one like google did with google translate but with programming languages or is that negating what each one was designed for
So I need to parse, and writing parsers is difficult by hand, while parser generators make it somewhat easier
Variable substitution can be done in excel
@JacobMcMichael no it's not, because just to combine two languages would take weeks of manual work
That's nice that excel does it, but my app is standalone from excel
It's a full standalone PyQt5 app called BananaCats
It already does visual diagramming
Wow let me look at that
What is the end goal?
The end goal is to make diagram chasing proofs about 10x funner / accessible
23:13
I can’t find the app with a simple search
Drawing large diagrams for me on paper is cumbersome and I constantly have to redraw
It's not published yet
I have the source code available but it's probably for an older version
Are you a coder?
You would not believe how much geometry code goes into getting the diagrams to do that above
Also the app supports persistent undo redo
Even if you were to restart your machine
You can undo / redo from whereever you left off!
Ok so step one puts symbol in a list
step 2 defines the symbol
step 3 puts it in the end list
?
What are you referring to?
The diagram
Oh, it was for a question I had, that's just an old screenshot
but the language will be extendible through diagrams themselves
to define further logic
but internally, I take all text expressions and store them more intelligently than just strings
It's possible to do just string-based, but would probably perform poorly, and people would go wow that's kinda cheap etc
And it might be more difficult to code
23:18
Yeah I mean I code enough to do what I need to to scrape and trim data for a database an a few automation codes. I want to learn more I haven’t actually sat down and read a whole book on python or vba like I need to
I could teach you Python
I never read a book on it
I just started coding on a project
Shouldn’t I just learn from a book?
If you already know some programming, you don't have to
Not disssing your teaching skills
it's just like any imperative language
like C, except syntax change
And you have auto garbage collecting
23:20
There’s just a lot of commands I don’t have memorized I need to make a database of commands to refer to
everything is a reference to an object
And typing is not static
Yeah, it's called googling
when you google a specific coding task, usually it pops up on SO
I know but googling has its limitiations honestly
People don’t often talk about advanced stuff
Well, there are all the documentation to python and it's libs... on the web. Not in any book
What would you like to do in Python?
You have to have a goal first
23:22
Or if they do it’s to fit a different purpose and you can barely recognize how you can adjust it to your purpose
I agree you have to have a goal
I can teach you how to parse something sense that's what I'm currently doing
Basically I want to beat the stock market lol
Yes, will not work
They already did that
It’s possible with advanced math models to do it
Bill Williams did that with HMM (Hidden Markov Model) introducing advanced algorithms into trading
That was 30 years ago, and so much more has been tried, all the working algorithms are proprietary.
23:24
I had a sample size of 5 years of a data with 500 trades left after using a custom filter and it returned 1 -3 % a week
Sure you can apply advanced math, but if you always have that goal lingering, it will distract you from the math itself. You'll likely become attached to some math subject while studying for that goal, and then forget about the goal.
Not amazing but it’s a start and I can get much more advanced and a much bigger sample size and go farther back in history. Even create fake history off of all the aviliable data
You sound like my friend Rhett
He was also talking about making up fake history
I don’t sound like anyone I know
but those signals... there's nothing in them.
You have to trade based on news events tempered by those signals
23:26
But this place would be where I would find some people I a can find stuff in common
But the news gets quantified by the past historical prices
Sure, that's true, but how exactly, is an open question
All that stuff is possible and people are working on advanced correlations to sentimental data
Sure, but they've got a team of 1,000,000 software engineers, and you are a company of < 10?
So you'd have to beat Bill Williams' HMM (also a former mathematician with PhD)
It’s easy BatchGetSymbols in Rstudio scrapes the data from yahoo finance you just have to plug and chug
but you'd have to not only do that, you have to beat a million other Bill Williams since 30 years ago
23:29
A professor from Mexico made that program I want to send him some money when I get it figured out
It's an interesting research area
I’m not gonna lie it’s hard
what's an intersecting researhc area?
Forex algorithms
But there’s not as many people working on this as you think
23:30
There are too many people working on it and not seeing fruit. Pure math is different in that the math itself is the fruit.
Most of the real mathmateticans that work on wall st won’t talk about the real math strategy’s that work they are contracted not to and they have no incentive to share info
So that means if you can figure it out without them there’s no real compettion
Sure, except for comptutational resources
they're right near the trading head quarters
What's the best model at the moment?
23:32
Like none of the retail investors really use advance math I know I didn’t for 4 years and I never heard of anyone else using it
so the speed of light in an optical fiber cable comes into play - that's how fast they're working
That's how fast their computer resources are making messages
Buy low sell high on stocks with high standard deviation. That are far from their 2 month high ( so sold off)
So you want to take say < 10 PC's and make enough to turn that into 100 PC's and so on.. until you can afford their real estate and hardware.
It's possible, but then you have to ask yourself, would it be worth it?
That’s only slightly above 55% win rate but it’s with a very specific mathematical filter that’s run on the data so even if I told you the numbers there’s no way to copy unless you set up the excel sheet to get that data
You could honestly study analysis, predictive analysis, probability, etc for 20 years and still fail at that game. Therefore just make the game math itself.
23:35
That’s what I’m saying shine
I love math now lol
:>
Whatever gets you hooked
now you're ours :)
Right haha
What textbook is next guys lol
Read Matsumura, Lurie, Grothendieck, Hartshorne
Like try finding a stock that has a 42 day standard deviation of 2.05 (not actual data but an example) you would have to run that filter that day to get that data so there’s no way to copy people or fact check people even without setting it up
What is that
Copy that ultra dark what are they about I can look them up but why do you mention them
make sure to get the audio book version
23:44
I don’t mind sending the code I use to you guys honestly I don’t know how to share that on here
It took me 2 1/2 months of working like a madman and 1000 google searches
Omg if you guys look at it and make a good strategy pls share with me so we can move on to robotics
We need money to fund the robot army ahaha
I can run 450 filters on 7500 stocks with some stretching back 30 years.
i just can’t do that much right now because my computer would take several hours to run any mouse click

« first day (3375 days earlier)      last day (1942 days later) »