Frosty's trading bot goes live part 2
- Thread starter frostengine
- Start date
All the stuff you describe can be written in Easy Language. You can even develope your own bars [any kind of bar ever thought of by the mind of man, or Cynthia Case in the case of woman] to work with in the arrays and develope your own reporting sytem for the test, been there done that.
The idea that C+ programming skills and a stern do-it-yourself approach will produce a working system is what I believe that Frosty is here to disprove.. and so far he seems to be.
People here have disparaged my statements about statistics which I learned from Phd's, are talking about better programming and backtesting skills as "the way", it's ridiculous. If you can trade it live and make money with it, then autotrade it, if you can really reduce what you do live to a clearly defined set of rules. Now that is a challenge, believe me, getting a clearly defined set of rules... with no conflicts, ever, understandable by a machine.. try it some time. I would say that the backtesting should occur after you are trading live manually and profitably, it might find some better settings for something, otherwise it is a big impediment to learning the market you are trading.
Today -$119.60
Week to Date: -$983.60
Since going live: -$3868.6
Total days: 19
Avg per day: -$203.61
Up Days: 9
Down Days: 10
Week to Date: -$983.60
Since going live: -$3868.6
Total days: 19
Avg per day: -$203.61
Up Days: 9
Down Days: 10
Stats are very important. It's the first step to be taken when you manage risk. Quant. Risk Analysis is a necessity, if not how does one set the contingency for the system? How does one set the optimal "stops" for the system?
Better programming allows you to work within the arrays and create your own reporting system, smoothly. Better programming allows you to calculate and provide you with significant stat. values, backtesting or on a real-time basis. Better programming allows you to break limits and test efficiently, whether it be speed (TS is slow) or chronological mutiple market/system tests (TS/Rina/PushPop runs each system individually then patches them up as one later... a major problem for real-time risk allocation issues).
In terms of the ability to trade manually. I started off as a discretionary prop. trader... And I'm in a middle position for this. There are Ph.D. quants who starts off developing systems then successfully learn how to trade discretionary.
Which ever the case, the Ph.D.s have strong quantitative backgrounds. I had a strong discretionary trading background before developing models... If one has no "successful" background, I find it faster to learn how to trade than going back to school.
Back to Easy Language/TS... I've been using it for over 10 years since TS 4.0. I still use it for quick-testing ideas. I'm well aware of the capabilities of TS... More-so, when I first started writing my own custom tester, I based it on TS and EasyLanguage.
The major problem TS has is speed, the package is way to slow. Another problem is code size limit for the tests that can be run at one time (This is with 2000i, it may have been corrected with the newer version).
My point is... for one to develop a system, you need to program, whether it's EasyLanguage, C/C++, and etc. If your development capability fits EL, then that's fine.
If you need C/C++, then that's fine too. I currently run tests that take eons and eons for TS to run. That's why I don't use TS as much as I used to.
Quote from walterjennings:
i wouldn't be so sure about that. the people with successful bots who are so bored waiting for them to trade, trying to find reasons to postpone further r&d, that they cling to any type of interesting conversation available?
We're all bored between tests...
As for the apple and oranges...
You can have an abstract class called Fruits.
Have derived class called Apples, Oranges, Bananas, and Others.
You can have the bot eat Fruits.
Then you can have the bot love Apples, Oranges, or Bananas.
When you find that there are Strawberries... there's nothing wrong with telling the bot to love Strawberries.
It's a matter of how well you can convince your bot why Strawberries are sooooo tasty...
You can have an abstract class called Fruits.
Have derived class called Apples, Oranges, Bananas, and Others.
You can have the bot eat Fruits.
Then you can have the bot love Apples, Oranges, or Bananas.
When you find that there are Strawberries... there's nothing wrong with telling the bot to love Strawberries.
It's a matter of how well you can convince your bot why Strawberries are sooooo tasty...
we've all forgotten about the case of fruit roll-ups. mmm.
this whole 'testing' thing takes too long. sink or swim i say.Quote from TSGannGalt:
We're all bored between tests...
Until the bot trades better, it might be more prudent to make it prefer Ramen noodles.
Quote from RichardTodd:
Until the bot trades better, it might be more prudent to make it prefer Ramen noodles.
LOL.
:eek:
One fact about Ramen noodle... Ramen noodles are consumed more by Latinos than any other ethnicity in the US. That said:
The bot should learn to speak Spanish first...
pinchi bendejo!!!!!
i'll be the first to admit my live system is'nt a black box (and never claimed such), im far too paranoid and protective of my seed capital to not be watching and turning it on and off. plus i could give a flying fruit wether i run a black/grey/white/fuschia box as long as its producing income.
and i should hope you are not implying that i made claims to be an expert ATS, nor did i tell kids to stop all the downloadn, throughout the thread i merely post thoughts, opinion and questions for the selfish purposes of modifying my own knowledge, and would never make a childish claim such as that (gee i hope you dont go quote to prove me wrong
). plus im much too fleshy to be an ATS, let alone an expert ATS.ps. whats all this bs talk about what our bots should eat? everybody knows bots eat old people's medicine for fuel. plus they have large metal claws which they grab you with, and you cant break free, because they are made of metal.