PacMan (my first hybrid) - page 11
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Beware of the illusion
I have a different signaling process comming along nicely. Its up to 79% on an 8 pip equity target. I will have 80%+ soon.
I added an overbought/oversold filter to keep it from placing new orders after TP of old ones durring extreme levels. And also an added decision process to keep it only trading when the time is right. But still takes frequent trades.
DaveI want to warn everybody of the wrong impression that he may have found the "perfect" indicator/filter combination. I don't want to say you haven't (I don't know, it's not up to me to decide), but their are simple probability/statistical arguments why it can be very difficult to judge:
- If the settings use a higher pipstep value than TP and max trades is relatively high (> 4), the power of the martingale already gives you a high success rate, even if you walk down random street ...
- Example: Pipstep=30, TP=23 , MaxTrades=5 => 98% probability that you hit TP at latest at level 5, even if your signal is totally random (edge=0).
- In order to know whether your signal is worth anything, you need a lot of trades and a lot of backtest/forward testing data to test these remaining 2%
- We all think that we don't have a (log)normal distribution of prices wich corresponds to the random walk, otherwise we wouldn't play with indicators ...
- However, our fat-tailed distribution can bite back nastily (those counter-spikes)
- We typically optimize indicators and add filters etc. until we can avoid blowup, SL etc. => However, usually it only means reducing trade rate and being lucky to not have hit the wall this time (because now we should simply test for a longer period)
Conclusion: Depending on the settings used, different indicators may only look different by pure luck during testing. The power of the martingale is fine until it isn't, so it is still better to have an edge ... This is much easier to test if you have smaller pipsteps, maybe higher TPs and consequently higher trade rates, such that our "disaster" happens from time to time and we learn how to handle it.
Still waiting for bluto to publish anything ...
Alassio,
- Example: Pipstep=30, TP=23 , MaxTrades=5 => 98% probability that you hit TP at latest at level 5, even if your signal is totally random (edge=0).
This is a very good point. This simple gambling-system works fine but those remaining 2% are going to wipe the profits of month if they happen to occur.
Still waiting for bluto to publish anything ...
Mind if I post something? :-) I already posted my version in the Goblin-Thread and nobody seemed to care. Prolly because I didnt give it a fancy name
I tried to adress this issue with a few simple but very effective rules
-> Pipstep 15
-> Only 1 Trade per 20 Minutes allowed because I saw alot of Martingale-EAs (Goblin, 10point3) going nuts during news and opening large positions.
-> Bollinger-Filter added to confirm signal
-> Smart StopLoss and TP adjustment. The logic behind this is as follows: If we have a large trade opened we want to exit as fast as possible to reduce the risk. Thats why the TP is calculated like this: TakeProfit = 30 - Openorders * MaxTrades. This way we get out of risky trades fast without losing money on sudden crashes.
I'd be very happy if someone could test it with 90%+ modelling quality. I'm running it live on all pairs with nice and steady profits
If you want to test it remember to put it on H4 Timeframe or it won't work well.
Best wishes,
Heiko
...........Mind if I post something? :-) .............
Hi. Just gave this a go and it doesn't backtest - it generates 1 trade which it then closes, and that's it! Everyting is where it should be and there are error messages. One oddity is that though the 'Lots' are set to 0.01, the trade is for 1 lot. I have an older version of Goblin which backtests fine btw...
Alassio,
Mind if I post something? :-) I already posted my version in the Goblin-Thread and nobody seemed to care. Prolly because I didnt give it a fancy name
I tried to adress this issue with a few simple but very effective rules
-> Pipstep 15
-> Only 1 Trade per 20 Minutes allowed because I saw alot of Martingale-EAs (Goblin, 10point3) going nuts during news and opening large positions.
-> Bollinger-Filter added to confirm signal
-> Smart StopLoss and TP adjustment. The logic behind this is as follows: If we have a large trade opened we want to exit as fast as possible to reduce the risk. Thats why the TP is calculated like this: TakeProfit = 30 - Openorders * MaxTrades. This way we get out of risky trades fast without losing money on sudden crashes.
I'd be very happy if someone could test it with 90%+ modelling quality. I'm running it live on all pairs with nice and steady profits
If you want to test it remember to put it on H4 Timeframe or it won't work well.
Best wishes,
HeikoHi Heiko,
Thanks for posting your EA with different modifications. It's a nice idea to put time filter, additional bolliger filter and different tp and sl adjustment. I will have a look at the code thoroughly and put it on test next monday. By the way, could you post your current forward testing stantement as comparison? Also, what's the setting for the EA? Is it fine with default setting?
Thanks again
Cheers,
-> Pipstep 15
-> Only 1 Trade per 20 Minutes allowed because I saw alot of Martingale-EAs (Goblin, 10point3) going nuts during news and opening large positions.
-> Bollinger-Filter added to confirm signal
-> Smart StopLoss and TP adjustment. The logic behind this is as follows: If we have a large trade opened we want to exit as fast as possible to reduce the risk. Thats why the TP is calculated like this: TakeProfit = 30 - Openorders * MaxTrades. This way we get out of risky trades fast without losing money on sudden crashes.
I'd be very happy if someone could test it with 90%+ modelling quality. I'm running it live on all pairs with nice and steady profits If you want to test it remember to put it on H4 Timeframe or it won't work well.
Best wishes,
HeikoThanks for posting you EA, rainx! What currencies do you trade?
...
Mind if I post something? :-) I already posted my version in the Goblin-Thread and nobody seemed to care. Prolly because I didnt give it a fancy name
I tried to adress this issue with a few simple but very effective rules
-> Pipstep 15
-> Only 1 Trade per 20 Minutes allowed because I saw alot of Martingale-EAs (Goblin, 10point3) going nuts during news and opening large positions.
-> Bollinger-Filter added to confirm signal
-> Smart StopLoss and TP adjustment. The logic behind this is as follows: If we have a large trade opened we want to exit as fast as possible to reduce the risk. Thats why the TP is calculated like this: TakeProfit = 30 - Openorders * MaxTrades. This way we get out of risky trades fast without losing money on sudden crashes.
...Backtest for 2007 looks ok, but 2006 gives horrible results.
It makes sense to adjust TPs to reduce your profit goal with higher orders, however our problem with spikes is that they initially have no retracement. We have to have a solution for that (at least a stop).
Backtest for 2007 looks ok, but 2006 gives horrible results. It makes sense to adjust TPs to reduce your profit goal with higher orders, however our problem with spikes is that they initially have no retracement. We have to have a solution for that (at least a stop).
Alassio,
Did you use H4 Charts and the default-settings? I've been backtesting 2006 and 2007 without any large drawdowns. COuld you please post the statement and chart and point me to any errors that I might have made?
thanks for your time
Hi. Just gave this a go and it doesn't backtest - it generates 1 trade which it then closes, and that's it! Everyting is where it should be and there are error messages. One oddity is that though the 'Lots' are set to 0.01, the trade is for 1 lot. I have an older version of Goblin which backtests fine btw...
Hi Omelette,
did you use it on H4 charts and a Broker that does support 0.01 Lot-sizes? (Micro/Minilots) If not then it won't work for you with the default settings. You have to set "Lots" to 0.1 then.
Please try again
Hi Heiko,
By the way, could you post your current forward testing stantement as comparison? Also, what's the setting for the EA? Is it fine with default setting?
my forward testing statement looks exactly like my backtests. Slow but steady profits. The statement isn't large enough to be representative so I won't post it yet - in 2 weeks I will. The default settings are just fine (If your broker supports 0.01-lots!)
Hey David, here are my results so far. Looks great
Just wondering if your working on the fixes this weekend and will have the fixed version by Sunday so I can forward test again. Can you let me know what settings I should change from the default so I can test the EA on a $500 account size? Thanks