So, has anyone else reviewed my trade logs from this Simulation or the other simulation I performed (75/25/ Monthly RSI - posted shortly thereafter)? I have reviewed the code, spot checked some of the trades on charts, and it all seems accurate to me. Additionally, the data seems solid as well. The code is a complete system that follows the 3-pronged entry/exit approach of the algorithm, automatically uses calculated Fib Extensions (out to 100 levels, if need be), uses RSI monthly filters, and has a Flag option to apply an Initial Stop Loss (this has been disabled for the simulations, to parallel the Philip’s trading algorithm precisely).
I would like another pair of eyes looking at these Trade Logs…pick your favorite CCY pair, back-test on your charts, and let me know if you dispute any of the trades taken or think some are missing, and we can go from there.
Yeah I can’t see the trade logs in txt nor do I seem able to convert them to Excel to double check. I’m assuming you also checked and found the 2 EURUSD pair trades you took in the first system.
This is very puzzling to me. I think it could well be the performance of the system over 11 years.
Philip, save the Trades11c.TXT file to your computer. Rename the file to Trades11c.CSV. Any version of Excel should be able to open this .CSV file at that point… Sorry, but this forum doesn’t permit native Excel files to be added as attachments, so I had to save as a .TXT file in CSV format. The trade log is too long to embed as a graphics image.
To answer your question, I did find those EURUSD trades, EXCEPT for the fact that the new 11-year data stops in Jan, 2015, whereas the first data set went through mid-March, 2015. So, please keep that in mind.
I don’t want to be a downer, but I think things are pretty accurate, but I want another set of eyeballs, in case I made some mistakes somewhere. Again, I used the same EXACT code base on BOTH data sets. It could well be accurate over the long-term, and if it is, better to find out now.
Also, awaiting the results of the other guy implementing his back-test algorithm in his proprietary back-testing platform - it may take you/he some iterations to flush out any of his coding issues, similar to the process that we went through, but should be a good sanity check for us all. Several others have mentioned back-testing as well, but I don’t know what that means - if they are using a meaningful back-testing platform/solution, I hope that they share results with the rest of us.
ok, i’m onto this. Starting with AUDUSD for now. And just for the sake of clarity, i’ll be using Tradingview’s charts so there could be a bit of deviation here and there. Will get back on AUDUSD trades shortly.
Hey guys, I’ve got a question about the stochastic:
I know the rules for the entry (at least I think so).
For example for a long entry:
MA-Cross
Stochastic (enters and) leaves overbought-area
Stochastic enters oversold-area
%K-Line of the stochastic crosses the %D-Line from below
–> entry
My question is: is it necessary that both stochastic-lines (%K and %D) entered the oversold-area in this example? Or would it be enough if just the %K-Line enters that area, the %D only comes down to 30 and reverses and after that the %K-line crosses it from below?
@jw1981
Would you mind to post your trades of the two weeks you mentioned in this post?
No charts or something, just the pairs and the time.
Bad news. Tradingview’s H4 data goes back to 2012.
Managed to check the last 3 trades however
[B]My conclusion[/B]: Your logic for the strategy is bang on target as Phil outlined. Only point missing is the aspect of moving SL to BE after price closes below 100% Fib (in which case the two losing trades would have been exited at 0.
[U]AUDUSD - Checked last 3 trades[/U]
[B]My Personal View[/B]: Of the three trades checked, two would have been winners, one would have exited at BE.
The two winners would have been if there was a partial take profit at 100% Fib & 127.2% Fib and trailing the stop to BE when price closes below 100% and trailing from BE to 100% when price closed below 127.2%.
To put it another way, the net PL -819.27 would have in fact been +ve based on partial exits
[B]Some questions[/B]
Why is there a huge gap between the trades?
Some profitable trades missing include:
Long position on 12th Sep 2013
Short position 4th Nov 2013
[U]EURUSD - Checked 5 trades[/U]
4 trades closed, 1 running.
Of the 4 trades resulting in -ve, 3 could have exited at BE and some even at a profit if booked partially at 100% and trailed SL’s accordingly.
Sorry couldn’t be of much help. I’ll check on other pairs and get back.
If you are open to modifications, i’d love to see how the system works with partial exits by automatic plotting of fibs between the high before crossover and low thereafter (or opposite) with TP’s at 100, 127.2 and 161.8 and trailing to BE accordingly.
Thanks, jw, for that. Sounds like my Fib Extension logic needs some review. I will look at it tonight, and get back to you. Very useful analysis, thanks. Believe me, I’d love my Fib logic to be wrong at this point…Also, I’ll provide answers to your questions, specifically, the trades that you believe are missing.
Just updated EURUSD as well. Same stuff… The exit criteria should also include ‘[I]Move SL to BE when price closes below 100% Fib’[/I] and you should see some significant improvements in the overall results even without using the more complicated partial exits at the Fib levels.
The recent two big failures on AUDUSD and AUDJPY got me to look at this system more closely. Although i’m sure both trades would be recovered, using a simple concept of divergence before entering a trade would have signaled traders to exit before the two massive rally (on shorts) and declines (on longs) would have happened.
Take a look at the attached chart, which shows bullish and bearish divergences. No wonder those who entered these trades got hit so badly.
[QUOTE=“jw1981;692547”]The recent two big failures on AUDUSD and AUDJPY got me to look at this system more closely. Although i’m sure both trades would be recovered, using a simple concept of divergence before entering a trade would have signaled traders to exit before the two massive rally (on shorts) and declines (on longs) would have happened. Take a look at the attached chart, which shows bullish and bearish divergences. No wonder those who entered these trades got hit so badly. <img src=“301 Moved Permanently”/>[/QUOTE] I took a beating on this trade.
Its a great point btw. I tried to use divergences (rsi mainly) with the trading system but I ultimately failed.
I look forward to you making it work hopefully.
My first feed back and hopefully you can prove me wrong or at least find a work around is that divergences wont necessarily work; Divergences mean there will be a retracement. The retracement did occur, back to the oversold area and up signalling a buy.
The point I’m trying to say is that the system accounts for divergences.
What would be great as that you go at the losers this year for example and show us that there were divergences. Then may be go to the winners and show us there was no divergence. This way its a filter for us all.
Jw1981, I have an answer to your questions above. The Long and Short AUDUSD positions that you mentioned were classic standard trades, EXCEPT for the fact that both violated our Monthly RSI(14) filter rule. The 9/12/2013 Long didn’t happen because Monthly RSI was at 23.768 (well below the required 66.67 filter) and the 11/4/2013 Short didn’t happen because Monthly RSI was at 33.8362 (just shy of the 33.33 filtering requirement). If you look at my 75/25 Simulation that I posted, this latter trade would have executed. So, that explains those questions. The Monthly RSI filter is used to filter out the Chop which will hopefully eliminate a lot of losers in non-trending environments. However, as to be expected, it may result in filtering out a few profitable trades here and there as well. And we may want to extend the 66.67/33.33 values to 70/30 or event 75/25, to get even better results. Once I review Fib logic and correct anything that might be wrong, I’ll run overall Sims again using these different Monthly RSI’s to see what the results bring.
Again, I will review my Fib Extension logic tonight. Thanks for posting this stuff; it forces review of actual trades/no trades, which is nice.
Exactly!! I saw you mention Cardwell’s RSI. If i’m not mistaken, he (or was it someone else) who debunked the theory of divergences as its widely put. Lower Low/Higher Low and LL/LL is taken as bearish, when its in fact a correction (or retracement) as you put it.
I totally missed the aspect of divergence considering the indicators and the system is trend trading, so divergences are but obvious.
And yes, you are right about divergences. some work some just keep on going their way until the market makes up its mind to respect the divergence.
Matter of fact, the ‘recovery factor’ in this system is nothing but based on these very divergences if i’m not mistaken (at least some of them, if not all).
NZDJPY long was another trade that went into -ve… and there was a divergence on that too.
The overall results of this system (and especially avoiding the massive losses on AU/AJ/NJ for ex) is to keep an eye on divergence. At the very least, giving divergences the benefit of doubt, traders could enter at a smallest lot and then scale in when the trend is established.
A good live example is Gold H4 charts. It gave a buy very recently but stochs made a lower low while Gold made a higher low, ergo enter long on the smallest trade size due to this divergence, but shore up on a second long signal on H4.
Ok. It would be nice to see a backtest on the system with the improvements on Fibs & exits but without the monthly RSI filter, just to get an idea on how it works with no additional indicators or filters.
Yes, agreed, let me look at Fib stuff first. I have flags in my program that allow me to easily enable/disable:
RSI Monthly Filtering
Initial Stops
Also, RSI Monthly filter values can easily be changed. Only issue is that on my clunky server, it takes a Sim about 1/2 hour to run, with 11 years of Major CCY data.
