Welcome Guest ( Log In | Register )

Outline · [ Standard ] · Linear+

 Quant Trading, gather some like minded people

views
     
TSklehfeh
post Apr 16 2021, 01:21 PM, updated 3y ago

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
Hi all,

I have been passively been trading and investing for the last 1 year and recently got hooked with using statistics and mathematics to trade rather than looking at charts and Technical Indicators.

I am just wondering whether they are any like minded people who like to explore trading using quantitative methods and what are your experience ?

For example , many retail traders use MACD , RSI etc , but there are other ways to look at the market , we can use ADF (augmented dickey fuller test) and Hurst Exponent to measure whether the market is mean reverting or trending and apply the algo/strategy accordingly,

We can use more reactive measurement with noise filtering technique to measure the average price instead of simple moving averages (sma) which is overly lagging.



James1982
post Apr 16 2021, 01:34 PM

New Member
*
Newbie
7 posts

Joined: Oct 2013
QUOTE(klehfeh @ Apr 16 2021, 01:21 PM)
Hi all,

I have been passively been trading and investing for the last 1 year and recently got hooked with using statistics and mathematics to trade rather than looking at charts and Technical Indicators.

I am just wondering whether they are any like minded people who like to explore trading using quantitative methods and what are your experience ?

For example , many retail traders use MACD , RSI etc , but there are other ways to look at the market ,  we can use ADF (augmented dickey fuller test)  and Hurst Exponent to measure whether the market is mean reverting or trending and apply the algo/strategy accordingly,

We can use more reactive measurement with noise filtering technique to measure the average price instead of simple moving averages (sma) which is overly lagging.
*
I have some questions
1. All these math term and name you quoting here, do you really understand it or just google copy and paste here?
2. do you understand how expensive is the raw data (real time and historical) for you to do all these calculation
3. Normally people doing this they are good in both quantitative and coding, so what is your industry or what is your strength
TSklehfeh
post Apr 16 2021, 01:47 PM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
QUOTE(James1982 @ Apr 16 2021, 01:34 PM)
I have some questions
1. All these math term and name you quoting here, do you really understand it or just google copy and paste here?
2. do you understand how expensive is the raw data (real time and historical) for you to do all these calculation
3. Normally people doing this they are good in both quantitative and coding, so what is your industry or what is your strength
*
1) average only in maths but i like it not that i am interested in the theory but interested in applying it to solve real world problems smile.gif i have went into the statistics rabbit hole and yes, only understand maybe 30% but enough to apply it smile.gif
2) yes aware of the price, once bought 10 years of historical data and cost me 500 usd sad.gif , so if anyone has some good data , maybe can share with me ? wink.gif
3) I am in ERP system space and comp science trained.


2malaysia
post Apr 16 2021, 02:07 PM

Getting Started
**
Junior Member
161 posts

Joined: May 2012
I think what James said in (2) is correct, the raw data is too much and too confusing. For forex trades, the price volume data based on that company data only not the whole
world with all the aggregate data. This you can see when yesterday tick chart and today tick charts are different as more data merge with old data the following date in any online
brokerage firm.

In addition, there is the CFD or certificate of different products which mean each brokerage company is having its own market maker to distort the market for they advantage
for both stock indices and forex. Of course you have heard about stop hunter in these brokerage company where the retail trader set stop and the market maker try to adjust
the price to execute their buy sell price then reverse to reach their stop value to take their money...

Who is managing the 23 hours-24 hours trading of all online financial products ? I think each brokerage house has to manage/manipulate its own order especially during non trading hours...

Online forex and Indices trading can depends on price and volume for 80% successful trade. But if you go too deeply into mathematical formula, you are like trying to hit and Toto 4d number with its
statistical data which not only waste your time but also your focus.

Overall I think you are just procrastinating, you avoid trade and try to find a holy grail ...But I would be sad to tell you there is no 90% right methods may be you can get at best 8 out of 10 trades.

Do not throw out a complicated methodology to waste your time. Start practicing real trading now and learn from experience. Start small and choose something easier to trade with less fluctuation.

(procrastinate is the opposite meaning of motivation)

Good luck.

Just a last tips, the most profitable trade are those in the opposite direction of the trend before it reverse so market maker usually want to catch the follower that jump into the trade more than the high risk taker ..

This post has been edited by 2malaysia: Apr 16 2021, 02:11 PM
waghyu
post Apr 16 2021, 02:09 PM

Look at all my stars!!
*******
Senior Member
3,389 posts

Joined: Sep 2019
QUOTE(klehfeh @ Apr 16 2021, 01:21 PM)
Hi all,

I have been passively been trading and investing for the last 1 year and recently got hooked with using statistics and mathematics to trade rather than looking at charts and Technical Indicators.

I am just wondering whether they are any like minded people who like to explore trading using quantitative methods and what are your experience ?

For example , many retail traders use MACD , RSI etc , but there are other ways to look at the market ,  we can use ADF (augmented dickey fuller test)  and Hurst Exponent to measure whether the market is mean reverting or trending and apply the algo/strategy accordingly,

We can use more reactive measurement with noise filtering technique to measure the average price instead of simple moving averages (sma) which is overly lagging.
*
forex is easy money game. I withdraw regularly my profits to local bank.
James1982
post Apr 16 2021, 04:13 PM

New Member
*
Newbie
7 posts

Joined: Oct 2013
QUOTE(klehfeh @ Apr 16 2021, 01:47 PM)
1) average only in maths but i like it not that i am interested in the theory but interested in applying it to solve real world problems smile.gif  i have went into the statistics rabbit hole and yes, only understand maybe 30% but enough to apply it smile.gif
2) yes aware of the price, once bought 10 years of historical data and cost me 500 usd sad.gif , so if anyone has some good data , maybe can share with me ? wink.gif
3) I am in ERP system space and comp science trained.
*
1. Thats the thing, even you are really understanding about 30% it which is already way higher than normal people but you are putting your money on something you only understand about 30%? Is it gambling or investing?
2. This one I have to give it to you at least you did spend on that to do some research instead of just talking about it
3. Well think I think you have no issue on algorithm trading at least
4. When someone told you that he makes money from forex 9.5 out of 10 they are just bluffing whistling.gif
TSklehfeh
post Apr 17 2021, 10:23 AM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
QUOTE(2malaysia @ Apr 16 2021, 02:07 PM)
I think what James said in (2)  is correct, the raw data is too much and too confusing. For forex trades, the price volume data based on that company data only not the whole
world with all the aggregate data. This you can see when yesterday tick chart and today tick charts are different as more data merge with old data the following date in any online
brokerage firm.

In addition, there is the CFD or certificate of different products which mean each brokerage company is having its own market maker to distort the market for they advantage
for both stock indices and forex. Of course you have heard about stop hunter in these brokerage company where the retail trader set stop and the market maker try to adjust
the price to execute their buy sell price then reverse to reach their stop value to take their money...

Who is managing the 23 hours-24 hours trading of all online financial products ? I think each brokerage house has to manage/manipulate its own order especially during non trading hours...

Online forex and Indices trading can depends on price and volume for 80% successful trade. But if you go too deeply into mathematical formula, you are like trying to hit and Toto 4d number with its
statistical data which not only waste your time but also your focus.

Overall I think you are just procrastinating, you avoid trade and try to find a holy grail ...But I would be sad to tell you there is no 90% right methods may be you can get at best 8 out of 10 trades.

Do not throw out a complicated methodology to waste your time. Start practicing real trading now and learn from experience. Start small and choose something easier to trade with less fluctuation.

(procrastinate is the opposite meaning of motivation)

Good luck.

Just a last tips, the most profitable trade are those in the opposite direction of the trend before it reverse so market maker usually want to catch the follower that jump into the trade more than the high risk taker ..
*
yes, aware of the complexity in aggregating and cleaning the data . Previously, we did cleaning on 10 years of Futures Data , we have to make it continuous, the rollover every month involved calculations and a lot of cleaning. Still fairly new to currencies but yes, we would never get the perfect accurate price and volume because of diff brokerages and their diff spreads and volume .

yeah , i honestly am not aware of stop hunting , but i understood after you explained. I am aware of market maker technique because part of my algo is to mimic their actions but "attempt" to do it better than them ( am using a technique to determine when the "informed" traders are making a move )

of the mathematical formula part, i am consciously reminding myself that my result may be an act of randomness ( especially after reading the book, fooled by randomness by Naseem Taleb) , however, to eliminate that my result is not purely on randomness, I make sure the backtest is at least 3 years and i did a monte carlo simulation to at best , try to ensure the performance dont go to ruin . Of course, there are other pitfalls , like overfitting , look ahead bias, data snooping bias etc but as much as possible, i try to eliminate them in my backtest and to factor in slippage and commission.

of the procrastination part , yes , i am trying to "go-live" real soon ( the tough part is now selecting a broker who is an ECN/STP, with good integration and low fees, a unicorn smile.gif ) , and i agree, the best learning ground is to trade live! , but i think will put the algo for paper test for 3 mths first, to catch any "bugs" and ascertain the performance.

I do also agree that there is no holy grail , and i have thought of a way to get closer to it (but nowhere near it) . It is really to have different strategies running on multiple "asset classes" (diversification is the next best free lunch) , I would put mean reversion strategies as the lowest hanging fruit ( think pairs trading , etf pairs, futures pairs ) , couple with a few momentum strategies and have an excellent risk management system (think kelly criterion and CPPI ) . Of the problem with black swan event, i am devising an algo that runs on fat tails event ( think of the VIX ) whereby it earns slowly in normal market condition and reaps profit during high volatility period ) . So i think with a mixture of Mean Reversion + Momentum algo , with an algo that reaps on fat tails event, it should be a more stable portfolio . (nowhere near the holy grail , but should generate smoother returns) ,, and then to the moon ! ( just kidding ) smile.gif


thanks for the tips! I think the whole market is based on this fundamental volume accumulation and distribution of the big sharks . Also , i also tend to like this phrase , "The market can stay irrational longer than you stay solvent " tongue.gif

may i know what you trade on? smile.gif


SUSxander83
post Apr 17 2021, 11:48 AM

Blast off like a rocket
*******
Senior Member
6,427 posts

Joined: Jan 2003
From: Autobiography!!!
QUOTE(James1982 @ Apr 16 2021, 04:13 PM)
1. Thats the thing, even you are really understanding about 30% it which is already way higher than normal people but you are putting your money on something you only understand about 30%? Is it gambling or investing?
2. This one I have to give it to you at least you did spend on that to do some research instead of just talking about it
3. Well think I think you have no issue on algorithm trading at least
4. When someone told you that he makes money from forex 9.5 out of 10 they are just bluffing  whistling.gif
*
7 out of 10 more realistic but it depends of the number of trades

TS should start putting the money to work rather than waste time in predicting the futures with mathematician models as markets shows that 70% failed because of sentiments doh.gif
TSklehfeh
post Apr 17 2021, 12:08 PM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
QUOTE(James1982 @ Apr 16 2021, 04:13 PM)
1. Thats the thing, even you are really understanding about 30% it which is already way higher than normal people but you are putting your money on something you only understand about 30%? Is it gambling or investing?
2. This one I have to give it to you at least you did spend on that to do some research instead of just talking about it
3. Well think I think you have no issue on algorithm trading at least
4. When someone told you that he makes money from forex 9.5 out of 10 they are just bluffing  whistling.gif
*
1) by means of 30% , it's relative . I learnt that we need to know what the numbers indicate. The basic is really understanding
characteristics of time series and basic statistics ( standard deviation, kurtosis , skewness, etc) . By going to the statistics rabbit hole, I learnt also other techniques that supposedly is practised in the quant world in the 1990s like ARIMA, Garch , but sadly, not all of them worked when I tried it out in matlab

2) yeah did some research by coding it out but like Malaysia2 mentioned, the ultimate test it to trade it live smile.gif

3) I started out my career as a java and sap abap programmer, but haven't been coding or the last 10 years, am mildly surprised that I can still code and also realized the programming paradigm has changed quite a bit , especially now with python in the easier way to deal with data structures and the impressive machine learning libraries that comes with it

4) yes , indeed , unless it is HFT , I believe HFT can achieve high win rate because they are doing stat arb . However , I don't believe high win rate equals high returns , I believe we need to look at expectancy , ( it could be low win rate of 30% , but each win is big and the loss are all contained and small , this is a classic characteristics of a momentum strategy )

May I know what you trade on ? smile.gif


mohdyakup
post Apr 17 2021, 01:17 PM

Look at all my stars!!
*******
Senior Member
4,287 posts

Joined: Oct 2009
From: Bintulu, Sarawak


Parking in this interesting thread
James1982
post Apr 17 2021, 09:07 PM

New Member
*
Newbie
7 posts

Joined: Oct 2013
QUOTE(klehfeh @ Apr 17 2021, 12:08 PM)
1) by means of 30% , it's relative . I learnt that we need to know what the numbers indicate. The basic is really understanding
characteristics of time series and basic statistics ( standard deviation, kurtosis , skewness, etc)  . By going to the statistics rabbit hole, I learnt also other techniques that supposedly is practised in the quant world in the 1990s like ARIMA, Garch , but sadly, not all of them worked when I tried it out in matlab

2) yeah did some research by coding it out but like Malaysia2 mentioned, the ultimate test it to trade it live smile.gif

3) I started out my career as a java and sap abap programmer, but haven't been coding or the last 10 years, am mildly surprised that I can still code and also realized the programming paradigm has changed quite a bit , especially now with python in the easier way to deal with data structures and the impressive machine learning libraries that comes with it

4) yes , indeed , unless it is HFT , I believe HFT can achieve high win rate because they are doing stat arb . However , I don't believe high win rate equals high returns , I believe we need to look at expectancy , ( it could be low win rate of 30% , but each win is big and the loss are all contained and small , this is a classic characteristics of a momentum strategy )

May I know what you trade on ? smile.gif
*
haha regarding 4, seem like you have no idea what is HFT at all, there is no way an individual investor can do HFT sweat.gif
2malaysia
post Apr 18 2021, 03:34 AM

Getting Started
**
Junior Member
161 posts

Joined: May 2012
For anyone who do not know

1. What financial products they want to trade or that can suit their style of trading
For example many years ago TSLA has made a lot of people losing money because of volatility and swear they never touch this counter again but for newcomer starting today, they think if they can buy many years ago
they are many hundred times richer etc.

2. Which brokerage to use ?

3. Which software to use to trade ?

4. Planning Part-time trading with full time job in other area not related to financial products he is trading.
You need insider information if possible, like if you have a good friend working as forex trader for bank negara to give you the tips..

5. Acting alone without any guidance by real trader.

My suggestion is please stop thinking of online trading in forex and indices or stock to make money.


But as i expect nobody will listen to me so if you must do it, then I advise you to stick to something simple with a lot of common sense and not mathematical calculation like fibonacci or indicators or creating new indicator etc

I recommend you go through the following youtube first (you need to know mandarin to understand)



However, talking about how to execute trade and make money this and that with back test etc are all unrealistic because you are only providing explanation with HINDSIGHT as you already know the right side of the chart.
For example, let said you have 5 conditions to execute a trade, If 1 conditions is meet, then 5 conditions contradict, would you take the trade. But in most youtube Guru presentation, they never explained why they ignore
the 4 conditions that are not met but continue to speak about how his condition 1 is so useful that it work and make money this and that..

In real trading without hindsight, your real experience is needed...And experience consist of practice, observation and learning, mental preparation, money management etc so this is why I say it is not easy and definitely not a part-time job.

TS is not even a python or C# programmer then he say this and that.... rclxub.gif

This post has been edited by 2malaysia: Apr 23 2021, 12:00 AM
TSklehfeh
post Apr 18 2021, 03:56 PM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
hmmm , interesting, I guess no one here yet have traded using algo with automation. So, should be a fairly new thing here.

Anyone here has experience trading purely on automation ?

Some of the most successful hedge funds in the world trade entirely using maths and statistics ( Medallion Funds by Renaissance Tech ) Medallion Funds

Also curious whether has any firms try HFT on FKLI or FCPO on Bursa?


just a sharing of what Hurst is , it can measure the price series whether it is mean reverting or trending . This is an example Hurst Exponent



Yggdrasil
post Apr 18 2021, 09:58 PM

Look at all my stars!!
*******
Senior Member
2,208 posts

Joined: Jan 2018
Malaysian brokers don't have the tech. Some foreign brokers like IBKR allow you to put your own API/algos to trade for you.

Algos aren't guaranteed to be profitable. One mistake and you lose all your money. Most algos earn money using arbitrage/delta neutral strategy.

If you want to trade on technical analysis (directional bet on delta), you have to be more right than wrong to profit. You also need large amount of money to profit. RM10,000 won't get you anywhere.
Yggdrasil
post Apr 18 2021, 10:05 PM

Look at all my stars!!
*******
Senior Member
2,208 posts

Joined: Jan 2018
QUOTE(klehfeh @ Apr 18 2021, 03:56 PM)
Some of the most successful hedge funds in the world trade entirely using maths and statistics ( Medallion Funds by Renaissance Tech )  Medallion Funds
*
You are talking about backtesting. If the 40 year return of S&P500 is 6% p.a. then we can assume the next 40 year average return should also be 6% p.a.
There is a weakness in this because past returns does not guarantee future returns.

Also, there is something called fat tail risk. A risk that has a super small probability of happening but if it happens the effect will be huge.
A good example is nobody expected oil futures to become negative because it didn't happen in history before.
It was the first time when covid happened that oil futures dropped to negative.

Most brokers use historical statistics to help their clients design strategies.
IBKR allows clients to estimate their probability of winning in a particular options trade based on historical statistics.
They have something called Probability Lab too (patent pending) which allows you to design strategies based on your own probability (i.e. what you think will happen).



2malaysia
post Apr 19 2021, 12:43 AM

Getting Started
**
Junior Member
161 posts

Joined: May 2012
A search through the web discover the followings video tutorials on Algorithm trading via python Using AI and machine learning also. Definitely not for 1 person team.

Attached Image
Attached Image
Attached Image
Attached Image

devil.gif

Where to get it ? biggrin.gif

This one did mentioned about Quant trading
Attached Image

Here is how to start learning Algo trading
https://www.freecodecamp.org/news/how-to-ge...ding-in-python/

1. Learn Python Programming
In order to have a flourishing career in Data Science in general, you need solid fundamentals. Whichever language you choose, you should thoroughly understand certain topics in that language.

Here’s what you should look to master in the Python ecosystem for data science:

Environment Setup — this includes creating a virtual environment, installing required packages, and working with Jupyter notebooks or Google colabs.
Data Structures — some of the most important pythonic data structures are lists, dictionaries, NumPy arrays, tuples, and sets. I’ve collected a few examples in the linked article for you to learn these.
Object-Oriented Programming — As a quant analyst, you should make sure you are good at writing well-structured code with proper classes defined. You must learn to use objects and their methods while using external packages like Pandas, NumPy, SciPy, and so on.
The freeCodeCamp curriculum also offers a certification in Data Analysis with Python to help you get started with the basics.

Blah Blah Blah..

You need many months or years. Good Luck !..

This post has been edited by 2malaysia: Apr 19 2021, 01:02 AM
2malaysia
post Apr 19 2021, 03:25 AM

Getting Started
**
Junior Member
161 posts

Joined: May 2012
Last Friday Nasdaq Live trade stream. Hello TS, this is what you expect..in algo trading, lots of alert, quick move, very scary if you are ill prepared or when your algo is wrong and lot of money lost in seconds...



The Vinny Trader create some algo but he still made losses -4500 before climbing positive to +8500, he said NDX 06-21 is difficult and does not work well with his created algo...

This post has been edited by 2malaysia: Apr 19 2021, 03:36 AM
waghyu
post Apr 19 2021, 03:30 AM

Look at all my stars!!
*******
Senior Member
3,389 posts

Joined: Sep 2019
QUOTE(2malaysia @ Apr 19 2021, 03:25 AM)
Last Friday Nasdaq Live trade stream. Hello TS, this is what you expect..in algo trading, lots of alert, quick move, very scary if you are ill prepared or when your algo is wrong and lot of money lost in seconds...



The Vinny Trader create some algo but he still made losses -4500 before climbing positive to +7500, he said NDX 06-21 is difficult and does not work well with his created algo...
*
More on risk management.

My setup are ultra high end low latency system, rate for HFT and other trading strategies.
TSklehfeh
post Apr 20 2021, 04:21 PM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
QUOTE(Yggdrasil @ Apr 18 2021, 10:05 PM)
You are talking about backtesting. If the 40 year return of S&P500 is 6% p.a. then we can assume the next 40 year average return should also be 6% p.a.
There is a weakness in this because past returns does not guarantee future returns.

Also, there is something called fat tail risk. A risk that has a super small probability of happening but if it happens the effect will be huge.
A good example is nobody expected oil futures to become negative because it didn't happen in history before.
It was the first time when covid happened that oil futures dropped to negative.

Most brokers use historical statistics to help their clients design strategies.
IBKR allows clients to estimate their probability of winning in a particular options trade based on historical statistics.
They have something called Probability Lab too (patent pending) which allows you to design strategies based on your own probability (i.e. what you think will happen).
*
Fat Tails are my fav research subject these 2 months, i recommend pple who are interested in fat tails and in trading to read the book "The black Swan " and "Fooled by randomness" by Naseem Taleb. He always favour fat tails bet ( limited downside, huge upside , provided your position is small )


yes , i have personally used IBKR TW for investing (not for trading) , and find it quite good. Surprisingly, i am able to call their support and some Eastern European dude answered. I have tried paper trading on calendar rubber futures pairs and find they have a reasonably well setup.

Yes, i been studying probability theory and hidden markov model , very important in managing risk ( value at risk ) especially market is random, and i think it will be awesome if IBKR has that lab.



thanks for chiming in!


TSklehfeh
post Apr 20 2021, 04:50 PM

Getting Started
**
Junior Member
67 posts

Joined: Nov 2020
QUOTE(2malaysia @ Apr 19 2021, 12:43 AM)
A search through the web discover the followings video tutorials on Algorithm trading via python  Using AI and machine learning also. Definitely not  for 1 person team.

Attached Image
Attached Image
Attached Image
Attached Image

devil.gif

Where to get it ? biggrin.gif

This one did mentioned about Quant trading
Attached Image

Here is how to start learning Algo trading
https://www.freecodecamp.org/news/how-to-ge...ding-in-python/

1. Learn Python Programming
In order to have a flourishing career in Data Science in general, you need solid fundamentals. Whichever language you choose, you should thoroughly understand certain topics in that language.

Here’s what you should look to master in the Python ecosystem for data science:

Environment Setup — this includes creating a virtual environment, installing required packages, and working with Jupyter notebooks or Google colabs.
Data Structures — some of the most important pythonic data structures are lists, dictionaries, NumPy arrays, tuples, and sets. I’ve collected a few examples in the linked article for you to learn these.
Object-Oriented Programming — As a quant analyst, you should make sure you are good at writing well-structured code with proper classes defined. You must learn to use objects and their methods while using external packages like Pandas, NumPy, SciPy, and so on.
The freeCodeCamp curriculum also offers a certification in Data Analysis with Python to help you get started with the basics.

Blah Blah Blah..

You need many months or years. Good Luck !..
*
thanks for this ! i think this will benefit for those (including myself) who are interested in using quant methods to trade. I strongly believe in quant still as i think mathematics has a way to understand/measure not just market but nature and the universe.


well, actually i been coding in python to test out a recent mean reversion strategy on the Dragon ( GBP/JPY ) , it is a scalping technique and we been using Hidden Markov Model technique and statistics to automatically detect the setup and execute the trades with real time risk management strategy (based on predicted probability expectancy of trade , market condition ) .


user posted image


I know it is backtest result and absolutely mean nothing , so , let me report back the live performance in 6 mths. Another interesting research we been working on is using Machine Learning technique to learn on your own past historical trades ( I emphasize on ... the machine learning your own trade, not to predict the market cause we have tried predicting and failed miserably ) and give you a probability on your next trade to assist your Cut Loss strategy.


thanks for collating the materials.


If all goes well, i would be keen to start a quant hobby group to invite maths and programming talents so we can share ideas and strategies .






2 Pages  1 2 >Top
 

Change to:
| Lo-Fi Version
0.0208sec    0.37    6 queries    GZIP Disabled
Time is now: 29th March 2024 - 07:22 PM