Edge

[Crude8 10]

Hello gentlemen i need your guidance in determining if trading system posses some edge.

i have been a scalper in commodities and i have incorporated a new approach which is sort of intraday positional trading and its possible to scale it up to 500 - 1000 lots based on my unscientific observations , unlike back testing or real time simulation i implemented it live on 1 lot straight away without any scientific approach , i am hoping i may get help in this forum who can conduct a thorough scientific dissection if it has some edge or its just fluke

The results are as follows - After 3 weeks

 

Number Of trades - 176

Winning trades - 86

Loosing trades - 90

Winning% - 49%

Loosing% - 51%

Average Win Amount - 1112

Average Loss Amount - 373

Average Win Trade Time - 76 Minutes

Average Loss Trade Time - 32 Minutes

Average Win/Average Loss ratio - 3.05

Profit Factor - 2.86

Expectancy - 355

 

 

Before starting i didn't used to measure any statistics and i found few online which i complied , used these formulas

Expectancy = (Probability of Win * Average Win) – (Probability of Loss * Average Loss)

Profit Factor = (Avg Win/Avg Loss)* Win%/(1-Win%)

 

My question is obviously there are traders who are more into statistics than myself , i need your help to see if my trading does have an edge

 

Regards

[BlueHorseshoe 10]

Hi Crude8,

 

Based on the figures you provide, then it is fairly unequivocal that the strategy has had an edge over the period you have been trading it. This does not mean that it will continue to perform in this way. Even random entries/exits tend to produce suprisingly long strings of profitability and drawdown.

 

I, and many others here, could test this for you over a longer period but . . .

 

(a) you have to fully disclose the strategy, which you might consider your own intellectual property and not wish to share, and

 

(b) backtested results do not always correspond to the results that can be realised in real trading due to the difficulty in attaining limit fills and/or the spread/slippage associated with other order types, all of which can have a dramatic impact on perfomance.

 

If you're trading a single contract/lot and the risks associated with this are not an issue, and you have already realised a modest profit using this strategy, then my suggestion is that you look at position sizing approaches that will allow you to continue to trade with increasing size, but remove the possibility of giving back all your profit should the strategy begin to fail. Trading in and out of your own equity curve could be one way to do this.

 

I hope that's helpful, and I'm sure you'll get plenty more advice in due course.

 

Kind regards,

 

BlueHorseshoe

[Crude8 10]

Thank you for the reply, i have traded only 1 lot live but i have implemented a position sizing algorithm which i have simulated during the course of last 3 weeks, Average Sizing in simulated environment varies between 1-4 lots depending upon my equity and average size comes out to 2.5 lots per market. Unscientifically speaking 1 lot result can be easily replicated on 100 lots per market BUT off course i am not financially ready for that yet.

I do not wish to disclose the strategy on public forum but i can share it privately. The statistics i provided to you were net of costs and slippages. Chance of strategy failing seems limited as majority amount of analysis is discretionary and even if i were to automate the execution , the speculation bias would be given at my discretion , say for example i consider the market a short sell - the algo will take only sell signals & stay flat during buy signal (rather than being long) , usually i have a large timeframe employed which ensures that i don't get scared by counter trend movement in smaller timeframes as by logic i would be flat during those times.

[BlueHorseshoe 10]

Thank you for the reply, i have traded only 1 lot live but i have implemented a position sizing algorithm which i have simulated during the course of last 3 weeks, Average Sizing in simulated environment varies between 1-4 lots depending upon my equity and average size comes out to 2.5 lots per market. Unscientifically speaking 1 lot result can be easily replicated on 100 lots per market BUT off course i am not financially ready for that yet.

I do not wish to disclose the strategy on public forum but i can share it privately. The statistics i provided to you were net of costs and slippages. Chance of strategy failing seems limited as majority amount of analysis is discretionary and even if i were to automate the execution , the speculation bias would be given at my discretion , say for example i consider the market a short sell - the algo will take only sell signals & stay flat during buy signal (rather than being long) , usually i have a large timeframe employed which ensures that i don't get scared by counter trend movement in smaller timeframes as by logic i would be flat during those times.

 

Hi Crude8,

 

Glad that was helpful.

 

You remark that "majority amount of analysis is discretionary" - this will not be possible to program, so I am doubtful that testing will be an option for you.

 

If you wish to PM me details I can try and run simple tests of the strategy when I have time and if the strategy does not require extensive coding.

 

Kind regards,

 

BlueHorseshoe

[sergso 10]

Hello gentlemen i need your guidance in determining if trading system posses some edge.

i have been a scalper in commodities and i have incorporated a new approach which is sort of intraday positional trading and its possible to scale it up to 500 - 1000 lots based on my unscientific observations , unlike back testing or real time simulation i implemented it live on 1 lot straight away without any scientific approach , i am hoping i may get help in this forum who can conduct a thorough scientific dissection if it has some edge or its just fluke

The results are as follows - After 3 weeks

 

Number Of trades - 176

Winning trades - 86

Loosing trades - 90

Winning% - 49%

Loosing% - 51%

Average Win Amount - 1112

Average Loss Amount - 373

Average Win Trade Time - 76 Minutes

Average Loss Trade Time - 32 Minutes

Average Win/Average Loss ratio - 3.05

Profit Factor - 2.86

Expectancy - 355

 

 

Before starting i didn't used to measure any statistics and i found few online which i complied , used these formulas

Expectancy = (Probability of Win * Average Win) – (Probability of Loss * Average Loss)

Profit Factor = (Avg Win/Avg Loss)* Win%/(1-Win%)

 

My question is obviously there are traders who are more into statistics than myself , i need your help to see if my trading does have an edge

 

Regards

 

Based on what other people say who are considered experts, 176 trades for intraday trading is a very small sample. You might have been lucky simply speaking and the system has no edge. If you want to find out if you have en edge you will need to do some serious statistical work like for example is done in this blog for a daily system Significance of a System For Trading SPY | Price Action Lab Blog

 

Also if you can figure out your Sharpe ratio it will be very good. It must be > 1.

[Crude8 10]

As a scalper i can make upto 500 trades in a day IF conditions warrants it in Oil/Natgas.

This system is not really a system per se more of sort of like one of my side projects for intraday positional trades with bias, as a scalper i really don't care where it is(market direction) as long as i can manage to keep my avg buy cost < avg sell cost.

Like i have mentioned before i concentrate on income rather than statistics and i have never much paid attention which i am changing now off course. I will look over at the link you have provided and report back. Although i think i have been in this business long enough to know that having a accuracy above 50% doesn't actually guarantee an edge.

In India our trading platforms don't provide much statistics from our broker end apart from simple buy/sell statements

Edited May 19, 2013 by Crude8

[Crude8 10]

I downloaded a excel sheet just now and filled in the values in it and i got this result

 

Avg Annual Risk-Free Rate - 7.5%

Average Annual Returns - 16.12%

Standard Deviation - 0.01

Sharpe Ratio - 6.40

 

Instead of years i entered weekly returns (3 weeks) , i can do this again tomorrow for each day return (14 days out of which 12 days green and 2 days red) , i will continue to monitor this and i will have more data from this week as well.

Okay gentlemen have a profitable week ahead, i am of to sleep now , i trade (sometimes overtrade :doh:) for 12-14 hours so i might not get a chance to post until next friday.

I am getting infrastructure in place so that i can understand from my scalping statistics where/when/why i am over trading and hopefully this method turns out to have edge rather than luck so that i can move above the scalping world and relax a little!

Edited May 19, 2013 by Crude8

[Crude8 10]

hope you guys had a great week , i sure did and here i have my full 4 weeks results

 

Number Of Trades - 267

Winning Trades - 135

Loosing Trades - 132

Win% - 50.5%

Loss% - 49.5%

Avg Win/Avg Loss Ratio - 2.96

Avg Win Trade Time - 75 Minutes

Avg Loss Trade Time - 30 Minutes

Profitable Trading Days - 84.21% [16 out of 19]

Loosing Trading Days - 15.79% [3 out of 19]

Profit Factor - 3.02

Expectancy - $ 6.76

 

i may be calculating sharpe ratio wrongly but it sure is coming at 6.88, is this possible ? well this week i inquired about algorithmic trading and the cost is coming around $500 per month for the server which they are saying will be installed at the broker end, not sure about other products yet but i am keen on setting up my automated routine by September ,lets see if i am able to find an optimal fit for the wallet as i think $500 per month is pretty stiff . Whats is your opinion about the algo costs ?

[SIUYA 11]

well this week i inquired about algorithmic trading and the cost is coming around $500 per month for the server which they are saying will be installed at the broker end, not sure about other products yet but i am keen on setting up my automated routine by September ,lets see if i am able to find an optimal fit for the wallet as i think $500 per month is pretty stiff . Whats is your opinion about the algo costs ?

 

Depends on how dependent on speed it is - there are algo alternatives that a lot less expensive than that.

The other factor to consider is also the overall combined cost of such algos and brokerage v a simple algo and retail brokerage. Always hard to say and it is simply something you have to work out based on volume and speed of trades - plus dont forget it is something that is moving rapidly and so always worth watching every quarter as well.

[Crude8 10]

Well i suffered a draw down of about -2.89% last week & my accuracy was 38.75% last week and i was winning 3 times i was loosing but due to costs and more looser, i ended down which i covered in 1 day (Monday) for equity & 2 days (Tuesday) for commodities & this week was one of the best weeks in my live trading for this system & i have been trading it for 29 trading days and have 80% winning days & 20% loosing days Overall accuracy is 49% while this week 60% were winners

Have a profitable next week Cheers!!

[Guest OILFXPRO]

As a scalper i can make upto 500 trades in a day IF conditions warrants it in Oil/Natgas.

This system is not really a system per se more of sort of like one of my side projects for intraday positional trades with bias, as a scalper i really don't care where it is(market direction) as long as i can manage to keep my avg buy cost < avg sell cost.

Like i have mentioned before i concentrate on income rather than statistics and i have never much paid attention which i am changing now off course. I will look over at the link you have provided and report back. Although i think i have been in this business long enough to know that having a accuracy above 50% doesn't actually guarantee an edge.

In India our trading platforms don't provide much statistics from our broker end apart from simple buy/sell statements

 

There is no edge here in averaging down etc.500 trades is for clueless traders , not skilled traders .

[Crude8 10]

There is no edge here in averaging down etc.500 trades is for clueless traders , not skilled traders .

 

Yup i am a clueless trader some of my best and rewarding trades have been done this way.

Staying at topic i have been following my system (semi-automated strategy) quite rigorously and i have come to except that what really matters is a positive expectancy over accuracy and analysis. Cheers!!

[Guest OILFXPRO]

Yup i am a clueless trader some of my best and rewarding trades have been done this way.

Staying at topic i have been following my system (semi-automated strategy) quite rigorously and i have come to except that what really matters is a positive expectancy over accuracy and analysis. Cheers!!

 

Broker and rebates and their agents spreading the holy grail eating up spreads .:rofl::rofl:

 

2 high probability trades a day will give same result., rather than 500.

[tradingwizzard 10]

Yup i am a clueless trader some of my best and rewarding trades have been done this way.

Staying at topic i have been following my system (semi-automated strategy) quite rigorously and i have come to except that what really matters is a positive expectancy over accuracy and analysis. Cheers!!

 

I agree with the last part........

[Crude8 10]

Well i am currently live on 3 strategies :-

 

Pair trade btw energy [i take 1 or 2 high probability trades in this]

Semi-Automated intra trend following [5 -8 trades per market on metals ]

Tape reading trades or scalps via dom [100-200 per market IF and only only IF crude oil is active]

 

 

3rd one the tape reading one occurs probably once in a quarter where in a day i might make 200-250 trades each in oil & natgas.

You know i have exchange membership in mcx , commissions are not to much for me