stevehopwoodforex.com • Dutch Goldfish needs a super-coder to make this work! • Print view

Author:

garyfritz [ Tue Oct 30, 2012 10:41 pm ]

Post subject:

Re: Dutch Goldfish needs a super-coder to make this work!

OK, I can backtest in Tradestation much quicker and more accurately than in Empty4.

Here's the same chart you showed in your first example, with TP = 10 pips. Does it look right?

Goudvis.gif

Here is the result of trading it on M5 EURUSD with TP = 10 pips:

Goudvis2.gif

I tried TPs from 2 to 50 pips. Smaller TPs lost less, but all lost. I tried it on M5 and M15 and the results were similar. (And this test does not include costs for commissions, spread, slippage, etc, so actual results would be worse.)

Then I tried it on a year of H1 and that looked a bit better. Still sideways for the first 8 months, but better after that:

Goudvis3.gif

I tried it on H4 and D1 and the results were similar to H1.

I tried it on a dozen different pairs. Most went sideways, several lost steadily. Only GBPUSD showed fairly consistent profits.

So: this worked about like I expected, though the run-up for the last 4 months was a surprise. But the last 4 months isn't enough to convince me this is a workable strategy.

If you're getting good profits on this in hand-backtests, I think you're either missing slight crosses of MACD (so you missed a lot of whipsaw trades) or you're subconsciously filtering out bad trades.

I can post a list of trades if you want to compare it to your hand test. Let me know if you want to see M5 or M15 trades, and what TP to use.

Author:	Generaal Goudvis [ Tue Oct 30, 2012 6:07 pm ]
Post subject:	Dutch Goldfish needs a super-coder to make this work!
Hi All, A message to all coders united, help is needed by a dutch goldfish.... I have backtested visually a very simple trading technique....and it seems to work for me. normally I trade on the 4hr chart with pure price action and live money...so not as fancy, but working...and for the smaller timeframes I am looking for a strategy to be automated.....I came up with this..... on 5 M or 15M timeframe MACD crosses zero line....entry Stop Loss....x pips take profit: 50% when it hits x pips, remaining 50% cross MACD and other MACD line for the clarity a visual representation is enclosed. The Bollingerbands and the MA are just my -additional- price action settings... Looking for the super-coder to code this and share this to the community (yes a true dutchman "sharing"...they do exist And feel free to make it better or more sophisticated..but for me this already works warm regards Generaal Goudvis

stevehopwoodforex.com https://www.stevehopwoodforex.com/phpBB3/ Print view

Dutch Goldfish needs a super-coder to make this work! https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=918	Page 1 of 1

Author:	garyfritz [ Tue Oct 30, 2012 7:10 pm ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
Greetings General Fish, My concern on this would be whether you have actually been trading/testing a strict interpretation of the rules, or whether your trading ability caused you to subconsciously filter out some trades? I've seen this many times, and the trader is never aware of it. The reason I'm suspicious is because I can't believe this works. It is such a basic application of a very basic indicator that 1000 people would have figured it out already. The examples you selected look fine, but does it work that nicely all the time, or at least often enough so you win more than you lose? If it really works with the rules you listed, with no discretionary filtering, I will be VERY surprised!

Author:	Generaal Goudvis [ Sat Nov 03, 2012 4:34 pm ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
Hi Gary, Many thanks for your help. I think I have to write down more accurately my trades and get back to you. Also trying tradestation would be helpful...perhaps. Your quote: If you're getting good profits on this in hand-backtests, I think you're either missing slight crosses of MACD (so you missed a lot of whipsaw trades) or you're subconsciously filtering out bad trades. I think it is more like your first point.....as I do it nowadays included in my price action trading...I do a lot of filtering based on the h1...and that would filter out all smaller and perhaps bad trades. To be more precise: I see a MACD cross, watch also BB move up and cross middle BB by a MA, and then I wait for the third sign...crossing the zero line. And on top of that I filter (looking for confirmation) with candlesticks....That is something that you can only do on the larger timeframes.... Could you post the M5 and M15 both on EURUSD and GBPUSD? I will retest and be more accurate in writing the trades down.

Author:	garyfritz [ Sat Nov 03, 2012 4:40 pm ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
What TP? Be aware that TS tests with the Bid, as I think Empty4 does. With small TPs you will get more cases where the Bid/Ask differences result in "winning" trades in the TS test that would actually lose in real trading. E.g. if you're short and the market approaches your SL, the Bid might not quite touch it (so TS thinks you didn't hit the SL) but the Ask touched it and exited for a loss. Since you're entering on a MACD cross instead of a stop, those cases will be reduced, but they still happen.

Author:	Generaal Goudvis [ Sun Nov 04, 2012 7:00 am ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
on 5 M MACD crosses zero line....entry Stop Loss....10 pips take profit: 50% when it hits 10 pips, remaining 50% cross MACD and signal line I will start backtesting on M5 EURUSD:starting 1/10 ending 1/11...just for the feeling and keeping your remarks in mind.

Author:	ross [ Wed Nov 14, 2012 10:13 am ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
Generaal Goudvis wrote:Truth or dare? Truth....I agree a 100% to what your saying.....Mostly for that reason I want to backtest not with my discipline but with the strict rules of an EA. If you do a search on Hopper-trade you will find something similar. But they will charge you and something so obvious is either not working or to simple to charge money for...So that's why I believe in this forum...sharing is what makes this world beautiful. What I did is I backtested it manually for two months on EURUSD on 15 min...took me some hours to write it down . that gave me a clue on the "goldfish-feeling", but two months in Forex is too short to say. So join me on this journey...... I think in the Hopper trade they are using the signal line - but maybe it is the MACD line - anyway it is the faster of the 2 lines - they weren't using the histogram (unless they changed it since I last watched the video) Also, they were trading it at the start of the US or London sessions. They go for 5 pips TP with 15 pips SL on the 5M - but larger TP/SL on higher TF's.... I suspect they are just using this system to sell their course - this is what Booker does - shows some simple system that will work about half the time to sell his courses. Here is an indicator that will alert when you get crosses.

Author:	Generaal Goudvis [ Wed Nov 14, 2012 10:44 pm ]
Post subject:	Re: Dutch Goldfish needs a super-coder to make this work!
Hi Ross, Many thanks for your answer, insights and also the indicator....it's funny although with same settings your MACD and the Hopper MACD have minor changes which could result in changes over different bars . At least this gives me the possibility to paper test and play with the settings. And for jennifer and Rob...it looks easy, but too easy to be true....and I think that is the reason why they don't show the results over a longer period in written.

All times are UTC	Page 1 of 1
Powered by phpBB® Forum Software © phpBB Limited