Brainstorming on trading topics

[TracerBullet]

Everything is Manual. Manual hard labour. a few thousand hours of work, per system.
My work here is done

Thats one way, another way is to try to automate some of it. This is ofc without discretion, but mech testing can give reasonable idea on what works without blindly following another's system. Very hard to backtest manually when you are designing the plan, so many options to test.

I do both mech and manual, still working on 1st plan. Worked on entry, then traded, then failed, Then worked on trade management, got better but again facing issue and now working on trade selection (stocks). Every failure is a chance for insight and to improve the system, as long as base edge is there ... i think almost there with this one now, lets see ..

 

[ncube]

For an intraday system, which trades upto 3 to 5 times a day, a minimum analysis of a minimum 3000 to 5000 trades, as it would include 4 to 5 years of data, which would have covered all types of markets, sideways, trending, very volatile, etc
An analysis of the MAE and MFE, with binning of both the MAE and MFE
A running drawdown figure
A sample of at least 100 bunch of trades, starting point picked randomly, and totalled for the next 20 and next 50 trades, and that max dd, and average dd of thes bunch of trades
Many more such things

@sumosanammain , Thanks, learnt something new today, I was not aware of MAE & MFE concept, googled for it and found that this is a very useful metric for both developing as well as optimizing trading systems. It will definitely help one to improve trading performance.

 

[mohan.sic]

How many traders here have 3 to 5 systems with a minimum of 3000 back tested traders here? Please do give your inputs people. Also, according to you all, what amount of data needs to be back tested? According to me, 4 to 5 years. And for you all?
Also, how many traders here have punched 20 trades absolutely according to system, one after the other, without missing a single trade because of self doubt? Here, I am talking of 20 consecutive trades that were triggered by your system, not 20 in total over a period of time, doing cherry picking of trades.

And guys and hopefully gals, please excuse any spelling or grammer mistakes, as I am just punching the keyboard from my heart, if i keep looking at the screen to correct spellings, my thought process is interrupted. Although, my english, when i need it to be, is impeccable

On intraday ideas, for each method I do manual testing for 3 months. This testing also includes real trades with appx 50 k value per trade. I don't go more than 4 months unless I miss the testing in between. In 4 months data, usually all kinds of market conditions will be covered.

When testing a intraday method, I test only on the current day data. I don't consider what happened yesterday. I do it only on 1 mt frame data.

Testing is done only on liquid stocks. Here along with good trading volumes, I prefer stocks with least bid-ask spreads.

I consider the technique and the test result as it is. I don't add any filters or conditions to improve its success rate. I think we should not optimize any technique. On a standalone basis if its hits +60% , on at least 200 signals, I consider it for further testing or i leave it aside. But I keep note of what I tested. Because if I don't have a record, I will re test the same idea after few months or years.

I will try to continue as we progress.....

 

[sanju005ind]

Whenever I look keep studying the charts.I form certain hypothesis.Then test this hypothesis against the past data. Rather than the quantity of trades on the back-test. I make sure that I test it against all market conditions like rallying,declining,high and low volatility markets.Then I do forward testing of about 40-50 trades in live markets with real money.I give more weightage to forward testing because in back-testing the crucial human element of emotion is not involved.In live testing it will show actual performance .This is the time also to realize what flaws are there in the system and in self. Once the test is done.I document it with performance. The risk involved and the capital requirements. Yes each system requires proper capital allocation.I do have some good systems but dont trade them for now because it does not meet my capital requirement and money management.
The kind of systems I have done for now are
1. Positional - 2 Systems
2. Swing - 3 Systems.
3. Intraday - 2 Systems.One for Vix > 15 and Vix < 15.
All this as per my rudimentary understanding of the market.
going forward which shall take about six months I am working on Python on the Quantopian platform
to learn /practice. I will be planning to develop systems based on statistical models and option pricing.

 

[sumosanammain]

One Primary thing I have not discussed here, is the instrument.
According to me, and this is just my own opinion, any instrument other than Nifty Futures is not suitable for system trading. This is because the nature of the price movement changes according to he business cycle that the scrip is going through. Since the Nifty is an average of 50 companies, the effect of 2 or maybe 5 companies does not have that much of an effect on the nature of the movement. Though this is also becoming skewed, nowadays, with banks having a major say in the movement.
By the time, you spend a few months and perfect a system on a particular company, its nature may or may not change. Thats your luck. Now, people will say, since you have tested 5 years of data, all cycles, the system muct have already accounted for that kind of movement. True to an extent, but companies can go through particular phases for years, till their slump, or run up continues. Nifty shrugs it off in a few days or a few weeks.

Anyway, logging off for a few weeks now, have a whole lot of work to do.
You can always reach me at my youtube chanel, titled, " Make 8658345638 percent per month risk free "

 

[Smart_trade]

I am not a 100 % mechanical systems person ( though my trade entry and stoploss is mechanical and happy that way ) but I was expecting to gain some inputs on mechanical systems development , so was watching silently.....I was hoping that you will take a simple system, backtest it and then add parameters, again backtest, change few parameters, backtest....then backtest on different data set ....to finally illustrate how system development is done....so I have to wait for few more weeks till Sumo arrives on the scene again.....

Smart_trade

 

[sumosanammain]

I am not a 100 % mechanical systems person ( though my trade entry and stoploss is mechanical and happy that way ) but I was expecting to gain some inputs on mechanical systems development , so was watching silently.....I was hoping that you will take a simple system, backtest it and then add parameters, again backtest, change few parameters, backtest....then backtest on different data set ....to finally illustrate how system development is done....so I have to wait for few more weeks till Sumo arrives on the scene again.....

Smart_trade

Sir, aap ke liye, Ghar aa jaata hoon. Saath mein kar lenge. Hope situation is better in your area now. Saw videos of MIT being flooded

 

[Smart_trade]

Sir, aap ke liye, Ghar aa jaata hoon. Saath mein kar lenge. Hope situation is better in your area now. Saw videos of MIT being flooded

Most welcome but it is still raining heavily......next 2-3 days heavy showers are expected....we can meet after that but please bless TJ members also with your systems knowledge......

Smart_trade

 

[TracerBullet]

One Primary thing I have not discussed here, is the instrument.
According to me, and this is just my own opinion, any instrument other than Nifty Futures is not suitable for system trading.

One way around that is to look at many stocks rather than 1 stock .. under common conditions they can behave similarly. And this has persisted for 5+ years of backtest and 2+ years of forward test ..
But things can change, edges can change or just go away - although i have not yet faced it.

 

[manny.here]

One Primary thing I have not discussed here, is the instrument.
According to me, and this is just my own opinion, any instrument other than Nifty Futures is not suitable for system trading. This is because the nature of the price movement changes according to he business cycle that the scrip is going through. Since the Nifty is an average of 50 companies, the effect of 2 or maybe 5 companies does not have that much of an effect on the nature of the movement. Though this is also becoming skewed, nowadays, with banks having a major say in the movement.
By the time, you spend a few months and perfect a system on a particular company, its nature may or may not change. Thats your luck. Now, people will say, since you have tested 5 years of data, all cycles, the system muct have already accounted for that kind of movement. True to an extent, but companies can go through particular phases for years, till their slump, or run up continues. Nifty shrugs it off in a few days or a few weeks.

Anyway, logging off for a few weeks now, have a whole lot of work to do.
You can always reach me at my youtube chanel, titled, " Make 8658345638 percent per month risk free "

Sirji acha hua aap ne koi phone number nahin diya , nahin to log kisi ko call karte aur case bhi ho jata recent movie ke tarah