System from scratch!

Looking forward to more results.

The method I currently use for trend determination is as follows:

Take a timeframe higher then what you are trading (if you trade 60m take 240m).
Draw a trendline, only redraw the trendline when now highs or lows are made.
If the trendline gets pierced wat for the first swing below or above the trendline.
If that swing high/low is flipped from support to resistance or vice verse then I consider the trend changed.
Do look out for fake flips, it's not so easy to change a long term trend.

Thanks for the update Narn - I did read your post on the trend line drawing.

I have done the 2006 results - Still a nice result using same size position sizing per trade.

http://img26.imageshack.us/img26/1866/results2y.jpg

Im gonna crash now, but this week i'll finish 2007 for this setup. I think if your looking for approx 10% per year, this system is consistant.

So 2007 is now tested, the results are not bad.

http://img641.imageshack.us/img641/2098/results3.jpg

The interesting thing about this system is the recovery factor. I had a string of 15 losses at the start, followed by a string of 12. Which forced the first 5 months of 2007 to be ugly as hell - The biggest drawdown over the period was 5% of account.

So to re-iterate on the setup: This was using Hook entry (outlined previously), Added in 4hr trend identifier (using ma's) and taking only entries in the direction of the current trend. SL 25pips / TP 100pips.

This system seems consistent and robust. If your psyche can take 5 months of draw down you could trade this.

BUT

it does not comply to my needs / wants.

So where do I go from here? I've really got 2 options - Modify the system to try get it closer to my needs / wants. or alternatively Shelve the entry as it is not identifying the moves i'm looking to consistently capitalise on. The size (i.e number of pips) does not matter. The size of the move in relation to stoploss is critical when combined with a average success %.

For Example: If your entry picks up on a 1000 pip moves , however to ensure a good accuracy rate your stop has to be 1000 pips away you end up with a system returning 1:1. If your accuracy is only 50% you will breakeven. Add in slippage/brokerage/spread and you will lose overall.

Alternatively

If you have a system that can pick a 45 pip move, and to realise a good accuracy your stop is 15 pips away. your system shoots for 1:3. If you get 50% accuracy you will steadily grow your account.

Even tho the other system picks up on massive moves its not the size of the move that matters. It is the rest of the system which makes it a winner.

Your leverage / position sizing will ramp up your growth - depending on how aggressive you want to be. System testing can help you recognize how much you want to risk on each trade. You can see your largest loosing streak and your biggest loss and largest drawdown period - this information is VITAL. You can then position yourself so your worst tested drawdown would only loose a % that you choose (i.e 20%). Please note however this will change your % growth / % loss figures and it is worth testing again with the position sizing set.

So.... I put it to the community: What should I do:

1) Ditch the current entry - test a new entry (start from scratch)
2) Keep the entry - Add in a different filter (if so - suggest what I should try next).
3) Cry myself to sleep.

;)

Quote:

Originally Posted by doczero

The biggest drawdown over the period was 5% of account.

it does not comply to my needs / wants.

Hé Doc

Results look nice.

The drawdown of 5% vs an end result of 9.2% is good. If you feel confident about the system you might risk double (with a DD of 10% make 20%). After all, if you check the results of 2006 the DD was even lower.

In one of the first posts you want a profit of 50%. That's huge. Most money managers can only dream of this. Is your goal realistic?

What you could do is get more trades. Use a smaller timeframe for the entry chart. You got about 1.5 trades a week using this system. What if you run it of a 15m or 5m chart? Those smaller timeframes are a lot more noisy so it could be a bad decision.

How about running it on multiple instruments (and generate more trades)?

Quote:

Originally Posted by doczero

Profit probability %: approx 50%.

I should have probably worded as win / loss ratio. Im really looking for a coin flip - the reason why is a coin flip is prone to win streaks (as well as loosing) however over a long sample size comes back to 50%.

Basically if I can emulate a coin flip with better odds (not 1:1 but 1:2 or even 1:1.5) with a high signal rate, I can then apply an aggressive money management tactic. My idea is to take advantage of a winning streak - i'll go into this later tho.

your right though - most hedge funds are happy to post 10 or 15% growth - but i'm looking for something I can exploit with compounding interest (using fixed % position sizing).

If I was to look at smaller time frames i'd have to adjust my SL / TP accordingly - as in my previous post I can come back to 10SL / 40TP but your right, the smaller you go - the more prone to whipsaws you are.

But awesome food for thought! There is only one way to tell... and thats to test!

I might throw up some 15min / 1 hr and adjust the stops (may even start with 10 SL / 40 TP)

Oh and the other Idea - multiple pairs: that is my plan. your right I havent tested that either.. Hmmm I might be testing that first before the lower time frame.. See which one works out best.
sounds like a plan! As always valuable input mate!

Narn - have you read that mystic genie thread I mention in one of my first posts?

Macrotactics

Its a 1 in a million scenario : still have a read and if you want find the thread with google. Its very interesting reading.

I just read it. Like you said it, it's a one in a million scenario. I wouldn't look at this as something achievable for somebody who want's to make a living out if trading.

Here:

Quote:

If he had of had a string of losses he would have burnt his account up in a handful of trades.

Most important part of that article. :)

I know a guy IRL who doesn't know shit about trading. Yet he made hundreds of thousands of $ when he bought and hold a stock in the dotcom bubble. Split after split he saw his money double. He also blew up a few accounts...

A serious trader takes everything he reads with a grain of salt. 50% return a year with a healthy risk ratio is very hard to achieve (I don't say it can't be done). Especially with a system that trades only 1.5 times a week.

Hi Narn,

50% isnt my target for return per year , its my target for accuracy (as per above).

and regarding that article:
Yes your point is valid about blowing up his account (please note this is Forex. Dotcom bubble + buy and hold = different game of skill), but the interesting thing is that he starts with such a low account size $400 usd the first time (so LOW initail risk). If you blow up a $400 account - who cares. So by turning $400 into $60k he achieves a low risk / high reward scenario. I consider $400 a month play money - If i came across a system that could support that method - i'd definitely put $400 a month towards it for a year to see what happens. lol after lots and lots of testing of course.

But I thought you'd be interested in the read.

Oh and I love your statement -

Quote:

A serious trader takes everything he reads with a grain of salt.

I believe this also...thats why I- TEST TEST TEST!

Quote:

Originally Posted by doczero

50% isnt my target for return per year , its my target for accuracy (as per above).

Sorry, I'm reading to fast... 29% accuracy isn't very high. But the results are positive. For years. Accuracy is only half of the equation (but you know that).

Quote:

Originally Posted by doczero

and regarding that article:
Yes your point is valid about blowing up his account (please note this is Forex. Dotcom bubble + buy and hold = different game of skill), but the interesting thing is that he starts with such a low account size $400 usd the first time (so LOW initail risk). If you blow up a $400 account - who cares. So by turning $400 into $60k he achieves a low risk / high reward scenario. I consider $400 a month play money - If i came across a system that could support that method - i'd definitely put $400 a month towards it for a year to see what happens. lol after lots and lots of testing of course.

Sorry, you can't convince me about this :)

I'm sure his accuracy will be even lower then 29%.

What if you blow up the account on day 2 of the month (I read he uses insane amounts of leverage)? SOH for 29 days? Or you gonna start over?

I know of two types of methods:

Very high accuracy but R:R sucks (stops are even bigger then targets).
Low accuracy and big R:R (like the method tested)

Never found/saw a system with high accuracy and high R:R.

You should stick to the system you are backtesting. Think of a way to give it a little higher accuracy (would you settle for 40%?)

Lol - sure I think what I describe is very rare. However I would like to create something that is prone to large consecutive win streaks (win streaks of 8-10 consecutive wins is my target / year). if you double $25 by 12 times you end up with $51200.

To be honest, if this system didn't produce large consecutive loss's - i'd probably trade it as is with proper position sizing. It is definitely workable as is.

My next step is to throw it across 3 pairs and test a year. That will really be interesting.

I'll try get that done shortly ;)