opinions on strategy discovery | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Opinions on this strategy? 3 replies

Infomercials --- any opinions 25 replies

opinions 1 reply

Opinions of USD. 10 replies

My 8 new books for $60 (EBAY), please give opinions. 1 reply

opinions on strategy discovery

  • #1
  • First Post: Mar 30, 2007 9:05am Mar 30, 2007 9:05am
Hello, this is my first post...

I have been optimizing one of my expert advisers and came across an interesting discovery. I discovered predictable situations for placing profitable orders with a maximum of 10 pips, 100% profit accuracy.

Now the bad news: These situations only happen 5-8 times per year tops, sometimes taking 4 months between orders. Traders would have to keep the computer online 24/7, year-round. I've tested this on MetaQuotes data and Alpari data bank. I've went through the data by hand. It all works out the same.

Is this a significant find IYO? Or does the lack of activity in between become so discouraging that one would have a hard time letting it play out? Also, knowing a situation is a certainty, to which level would you take advantage of it? If you knew beyond a doubt that you would earn 10 profitable pips, how much margin would you use?

Looking forward to your input!
  • #2
  • Mar 30, 2007 9:17am Mar 30, 2007 9:17am
No trade can ever be a 100% certainty, if it was you could put all your wealth on it each time and very soon own the World.

If this only happens 5-8 times a year how many years worth of records have you got.

I reckon you need at least 200 trade records to show consistency which is about 30-40 years in this case, if you just have a couple of years I don't think it is meaningful in any statistically relevant way.
  • #3
  • Mar 30, 2007 9:22am Mar 30, 2007 9:22am
Quoting exptrade
Hello, this is my first post...

I have been optimizing one of my expert advisers and came across an interesting discovery. I discovered predictable situations for placing profitable orders with a maximum of 10 pips, 100% profit accuracy.

Now the bad news: These situations only happen 5-8 times per year tops, sometimes taking 4 months between orders. Traders would have to keep the computer online 24/7, year-round. I've tested this on MetaQuotes data and Alpari data bank. I've went through the data by hand. It all works out the same.

Is this a significant find IYO? Or does the lack of activity in between become so discouraging that one would have a hard time letting it play out? Also, knowing a situation is a certainty, to which level would you take advantage of it? If you knew beyond a doubt that you would earn 10 profitable pips, how much margin would you use?

Looking forward to your input!
Operating from the premise that you actually have found a 100% accurate trading opportunity, yes it would still be worthwhile. IF and ONLY IF it were 100% accurate, you could trade it at 100:1 or 200:1 leverage and earn 10-20% ROI per event. Based on your trade quantity figures, your looking at 50-160% a year (non-compounded). Thats not bad money.

BUT I would be VERY cautious with your expectations. Backtests are cool but they are in no way a perfect representation of the true market. You may not be able to get filled at the prices your backtest shows that you could. You Could also have a skewed expectancy due to a low number of events in your sample. It is very hard to accurately predict what a true winrate is with less then 500 samples. Having such a low qty per year, you need around 100 years of data to truly develop confidence in your results.

My suggestion would be to allocate maybe 10-20% of your total stake to the system and just let it run all out with max leverage. I believe you said it was an EA, so it's not like you need to really do anything beyond fund an account and click go. If it really does what your backtests imply, in 5 years you should be sitting on a nice nut.

With all that freetime you have, keep looking for something a little less reliable and a lot more frequent.
  • #4
  • Mar 30, 2007 9:24am Mar 30, 2007 9:24am
Why would you reckon 200 trades as being accurate? I'm not doubting you, but I've noticed that we like to put arbitrary numbers on statistics.

I agree, the 5-8 times that have occurred so far probably isn't enough to claim 100% accuracy, but you could easily increase your pool by testing it out on more currency pairs. If you come back with 100% accuracy there, I think it would be cause to at least try it.

At this point, I would be skeptical only because we're talking about one currency over a 1 year period...

To scorpion's point, if you had any kind of significant data point (1 year is just too short)...like say even 10-15 years OR 10-15 currency pairs and they ALL return 100%, believe me, people would listen.

The 1 and 1 thing (pair and year) is just too small a data set.

Quoting scorpion
No trade can ever be a 100% certainty, if it was you could put all your wealth on it each time and very soon own the World.

If this only happens 5-8 times a year how many years worth of records have you got.

I reckon you need at least 200 trade records to show consistency which is about 30-40 years in this case, if you just have a couple of years I don't think it is meaningful in any statistically relevant way.
  • #5
  • Mar 30, 2007 9:28am Mar 30, 2007 9:28am
I completely agree with Darkstar's point.

For YOU to trade it, you have to be comfortable with your findings. For others to trade the system, they would have to be comfortable with your findings. It is obvious that Scorpion and I don't share your optimism for the system. That's not to say that it isn't a great system or the beginnings of one...but you obviously have to be comfortable enough and confident in your findings in order to trade it.

For me...like I said, I'm pretty risk tolerant and I wouldn't trade based on only 1 year and 1 currency pair.

Quoting Darkstar
Operating from the premise that you actually have found a 100% accurate trading opportunity, yes it would still be worthwhile. IF and ONLY IF it were 100% accurate, you could trade it at 100:1 or 200:1 leverage and earn 10-20% ROI per event. Based on your trade quantity figures, your looking at 50-160% a year (non-compounded). Thats not bad money.

BUT I would be VERY cautious with your expectations. Backtests are cool but they are in no way a perfect representation of the true market. You may not be able to get filled at the prices your backtest shows that you could. You Could also have a skewed expectancy due to a low number of events in your sample. It is very hard to accurately predict what a true winrate is with less then 500 samples. Having such a low qty per year, you need around 100 years of data to truly develop confidence in your results.

My suggestion would be to allocate maybe 10-20% of your total stake to the system and just let it run all out with max leverage. I believe you said it was an EA, so it's not like you need to really do anything beyond fund an account and click go. If it really does what your backtests imply, in 5 years you should be sitting on a nice nut.

With all that freetime you have, keep looking for something a little less reliable and a lot more frequent.
  • #6
  • Mar 30, 2007 9:28am Mar 30, 2007 9:28am
Quoting scorpion
No trade can ever be a 100% certainty, if it was you could put all your wealth on it each time and very soon own the World.

If this only happens 5-8 times a year how many years worth of records have you got.

I reckon you need at least 200 trade records to show consistency which is about 30-40 years in this case, if you just have a couple of years I don't think it is meaningful in any statistically relevant way.
I only have 1999-to-present data so my trade records would definitely fall below your recommendation.
  • #7
  • Mar 30, 2007 9:29am Mar 30, 2007 9:29am
So you've backtested all of those years?

Can you post your results?

Quoting exptrade
I only have 1999-to-present data so my trade records would definitely fall below your recommendation.
  • #8
  • Mar 30, 2007 9:35am Mar 30, 2007 9:35am
I appreciate your responses. I actually am very comfortable with it. I'm just not used to the timespan in between orders. I like to see things happen on a daily basis otherwise I'll probably lose my drive. My posting here wasn't really doubting the findings, it was more of a request of opinions to find to what extent others would go if they knew of a small profitable certainty. How much to risk, etc.

And to clear things up it's 100% from 1999-2007

Quoting mrmikal
I completely agree with Darkstar's point.

For YOU to trade it, you have to be comfortable with your findings. For others to trade the system, they would have to be comfortable with your findings. It is obvious that Scorpion and I don't share your optimism for the system. That's not to say that it isn't a great system or the beginnings of one...but you obviously have to be comfortable enough and confident in your findings in order to trade it.

For me...like I said, I'm pretty risk tolerant and I wouldn't trade based on only 1 year and 1 currency pair.
  • #9
  • Mar 30, 2007 9:38am Mar 30, 2007 9:38am
Quoting mrmikal
So you've backtested all of those years?

Can you post your results?
Since this was discovered while modifying another EA, I used that EA to find the signals for placing the orders this strategy would place, then hand back-tested the after effects. I'm putting together the EA to specifically place these orders and I'll be happy to post the results here.
  • #10
  • Mar 30, 2007 9:41am Mar 30, 2007 9:41am
I only plucked the 200 trade figure out of thin air, just making the point as others have that you need a good sample, as suggested if it gives 5-8 trades per year on one pair try testing it on say 10 pairs to give a fairly good 50-80 trade record, then 10 pairs over 5 years which would give at laest 250-400 trades, then you will at least know if you are onto something.

As also said, remember that live trading is different, you need to get filled and be able to take profit at the price you want, if you have a lot of leeway with the '10 pips' then thats maybe not a problem.
  • #11
  • Mar 30, 2007 9:46am Mar 30, 2007 9:46am
Quoting scorpion
I only plucked the 200 trade figure out of thin air, just making the point as others have that you need a good sample, as suggested if it gives 5-8 trades per year on one pair try testing it on say 10 pairs to give a fairly good 50-80 trade record, then 10 pairs over 5 years which would give at laest 250-400 trades, then you will at least know if you are onto something.

As also said, remember that live trading is different, you need to get filled and be able to take profit at the price you want, if you have a lot of leeway with the '10 pips' then thats maybe not a problem.
Right now I'm only working with EURUSD. I'll download the Alpari data for some other currency pairs and see if this process works for those as well.
  • #12
  • Mar 30, 2007 10:22am Mar 30, 2007 10:22am
I'm not entirely sure why this has moved to the forex "beginner" area. I have over a dozen EAs written, have been trading for years and am using this topic is about the development of a new trading system... its original forum was correct.
  • #13
  • Mar 30, 2007 10:31am Mar 30, 2007 10:31am
I don't think there's any reason to take offense for the move. It's not that you're a beginner, but you started the thread with a rather benign question. From what it sounds like, you're at the "beginnings" of a system. It's not so much a statement of what you're expertise levels are but as to the maturity of your soon-to-be-explained system.

It would be a new trading system, except for the fact that you've yet to divulge exactly what that system entails.

At this point, it could (and should) be classified as either a forex discussion or a forex beginner's thread.

Quoting exptrade
I'm not entirely sure why this has moved to the forex "beginner" area. I have over a dozen EAs written, have been trading for years and am using this topic is about the development of a new trading system... its original forum was correct.
  • #14
  • Mar 30, 2007 12:19pm Mar 30, 2007 12:19pm
Quoting mrmikal
I don't think there's any reason to take offense for the move. It's not that you're a beginner, but you started the thread with a rather benign question. From what it sounds like, you're at the "beginnings" of a system. It's not so much a statement of what you're expertise levels are but as to the maturity of your soon-to-be-explained system.

It would be a new trading system, except for the fact that you've yet to divulge exactly what that system entails.

At this point, it could (and should) be classified as either a forex discussion or a forex beginner's thread.
The reason it disturbed me is because I expected far greater input/involvement in the trading systems area. Maybe I'll just relist this as a trading system after I get those results I told you I'd get.
  • #15
  • Mar 30, 2007 1:57pm Mar 30, 2007 1:57pm
Exp,

I understand your concern...and to be honest, you probably would get more help in that area, but you'd have to divulge your system before it can actually be listed in the trading systems area...otherwise, it's more a Q&A thing anyway.

That's a great idea. Relist your work once you have your system rules and appropriate results to share.

Thanks...I'm sure there are already a lot of interested people.

Quoting exptrade
The reason it disturbed me is because I expected far greater input/involvement in the trading systems area. Maybe I'll just relist this as a trading system after I get those results I told you I'd get.
  • #16
  • Mar 30, 2007 11:40pm Mar 30, 2007 11:40pm
Quoting exptrade
The reason it disturbed me is because I expected far greater input/involvement in the trading systems area. Maybe I'll just relist this as a trading system after I get those results I told you I'd get.
It may have been influenced by your post count. Don't let the move bother you. It is still being read.

Not sure if you can do this with the platform you are using but you may want to consider this.

Example Only
Pick a New Pair.
First Data Date Available: March 2001
Load Data From March 2001 to March 2004.
Backtest Loaded Data and record optimum Inputs.

Load Data From March 2004 to March 2007.
Apply EA using Inputs achieved from Backtest.
Analyze results.

hth
  • #17
  • Last Post: Mar 31, 2007 10:15am Mar 31, 2007 10:15am
I moved this to the this forum, because there is no system listed and the claim of 100%. We have seen this type thing before and it always has developed into a person trying to sell something. (ie, Zeekans thread, he eventually tried to sell by PM)

The trading Systems forum is for systems that are spelled out, not for a discussion of a method that will not be available...

I hope that clears it up. Thanks Scott
0 traders viewing now