I use data I have downloaded within the metatrader 4 software. I use a backtest and it downloads history from the time period I have chosen. I do not have the data in a folder or other location.
EDIT: On Google, I searched and found Download Free Forex Data – Free Forex Historical Data. I hope it will help.
Sorry, I was not clear. I already have data from 1/2004-1/2015 that I used in my aforementioned simulation. I was asking if you could do the following, once your development is complete:
Run your back-test over the same period with the 7 Major CCY’s?
Generate a Trade Log similar to what I did?
It may provide a sanity check for all of us.
Thanks in advance.
OK, will do. It will take me a couple days at least to finish the EA modifications and test everything. My data will be a bit different trades wise because I am currently working on having the EA not trade during non trending markets. I will upload a .zip file with tests for most currency pairs when everything is done along with a release of this new revised version of the EA. I will also upload a .zip with tests you requested without the new feature on of non trending markets.
If this data is true, then this isn’t the most profitable trading system lol. Anyways, I made note of this data and page where it’s located. I’ll test my EA and see if this is something that happens with the ea I am working on modifications to. Being how profitable this method is, how popular this topic is and how Philip trades this method on his own for many months, I assume this data is indeed invalid. Time will tell and a lot of hard work but anyways, I’ll post an updated EA and data by next Friday hopefully. I want to have a stable release that is bug free so whatever time it takes me, it takes. Trading during non trending markets is something I am eliminating and that alone should eliminate a majority of loss. The results I post will be of both non trending market option to not trade being on and being off plus your request of me.
Thanks, techy. On Philip’s advise, we have been using the Monthly RSI(14) as a trading filter to avoid chop. I have run my simulation w/o this RSI filter, and the results are worse. Of course, we can optimize on all of this. Good luck with your development.
I think the jury is still out on this. It appears that it has worked much better of late (last 3 years or so), but the value of these independent tests is that we provide a sanity check to make sure we are all implementing the same way, to help validate the solution. It takes a whole lot longer to develop/test by posting on these boards, but I’ve enjoyed the effort in doing so, as everyone seems to be working together nicely. I am a big believer in back-testing - have been doing this for close to 10 years - and at the end of the day, has helped preserve my accounts. It is well worth this sort of time investment, and I’m sure you concur on this.
Well I’m done with the conversions. I took sometime to add some crosses that had trades in 2014. The total number of trades was 289.
Now here is something that will have an important impact on money management later; If you count the break evens as losers, the winning% will be 43.3%. If you count them as winners, it will 58.8%. If you remove the break evens from the count all together the percentage is 51%.
EDIT: total pips produced were 6919 pips. We will see the importance of money management next.
I haven’t revised the trades but we will just assume they are right. Now we will move on to discuss money management. You all know about the 1-2% rule, so I will look to discuss three other methods next then apply them to our data.
Kh
This will be great, Philip, and u r spot on with the %'s with BE’s treated as losers. More important than Winning %, is net gain, and my crude money mgmt approach on my sim, can only be enhanced from here.
BTW, Engine gave me 11 yrs of 28 CCY pairs (including majors), so I’m going to try to expand on my sim to include ALL that data this weekend.
Could you please this time send me an excel file to my e-mail? I will send it to you because I had a very hard time filtering those trades to the criterion I wanted. So I can’t imagine what it will be like for 28 pairs.
Will do. I assume that PM’s go to Email? Not sure why you aren’t able to take my .txt file attachment, rename it to .csv, and then load into Excel…it should give you total access to it in Excel.
No when you’re done just let me know and I’ll send you my e-mail.
I will look to discuss the following in the next series of posts:
Phil, its very simple. Download the txt’s from dough. Open up excel.
Go to:
Data> select ‘From Text’ and locate the txt.
Click next on the pop up window and you have everything there.
Then select the top row, and click on data>filter
this will give you easy drop downs to show the select ccy’s and so on.
I doubt it will work but I’ll take your word for it and try it next time. Thanks bro.
[B]Summary[/B]:
Here’s the most recent simulation using 11 year’s worth of 4-hour bar data for 27 CCY’s (includes 7 majors from last Sim) Again, thanks to Engine who provided me with all this 4-hour data over the 11 years. This was run on the same EXACT code base as my last simulation. The 27 CCY’s covered in this Sim are (in alphabetical order): AUDCAD, AUDCHF, AUDJPY, AUDNZD, AUDUSD, CADCHF, CADJPY, CHFJPY, EURAUD, EURCAD, EURCHF, EURGBP, EURJPY, EURNZD, EURUSD, GBPAUD, GBPCAD, GBPCHF, GBPJPY, GPBNZD, GBPUSD, NZDJPY, NZDUSD, USDCAD, USDCHF, USDJPY, and USDSEK.
To reiterate, my version of WealthLab doesn’t natively support FX, so all trades are denominated in $'s (and are traded like shares) based on a very simplistic 10% of portfolio allocation per trade, with a starting account of $ 250,000. Hence, very rudimentary money management. Monthly RSI(14) threshold values are used to minimize the chop and are set at 66.67/33.33 respectively. Of note, Wealthlab treats all BE trades as losses, so this has a pretty significant impact on Winning %, but helps accurately depict Avg Winning Trade $ vs. Avg Losing Trade $ per trade. It appears that about 18% of all trades taken out of the 876, were BE trades. Also, note that it took 14+ months of data before we could take a trade, due to our RSI(14) data requirement.
Additionally, in the Trade Logs referenced below, I have automated and created an “Exit Trade Column” that precisely describes in text, how a trade is exited (e.g., 3-prong exit approach, BE, Fib Level 1.272, etc.). Additionally, when RSI filtering is enabled (which it is for this Sim), I also notate the Monthly RSI value at the time of all Trade Entries.
[B]Disclaimer[/B]:
Given my server speed and the code base as it is currently written, it takes almost 2 hours to run this simulation. The bulk of this processing time occurs due to the portion of the code that calculates Monthly RSI on trade-able bars. I am currently looking to improve that part of my code, but just wanted to let you guys know this, in case I’m asked to run 8 different simulations. Let’s be picky and choosey on which simulations to run.
[B]HELP[/B]:
I reviewed logic, and did some spot checking on charts for select trades. If you guys want to grab a CCY pair for a particular data range (a couple of years), and compare your findings against the attached Trade Log below, that would be appreciated!
[B]Performance Results[/B]:
NOTE: Performance can be optimized through money management techniques (still TBD), possibly tweaking RSI values, use of divergence filtering, etc. Without resorting to curve fitting, I think that all of these should be explored further.
[B]Performance Graph[/B]:
[B]Trade Logs[/B]:
Copy this Trades14c.TXT file to your computer, rename it to Trades14c.CSV, and then load it into Excel (make sure to open it as a .CSV file explicitly, if you have an older version of Excel. [B]Some people have been having trouble doing this, so if you would like me to Email the Excel file containing these trades, please PM me with your Email Address. I will put you on a BCC mailing list and send you the Excel file directly.[/B]
[B]The Purpose of Money Management.[/B]
There are two reasons why a forex trader should apply money management; 1) Restrict losses. 2) Maximize gains.
Unfortunately, you cannot have both simultaneously. You will have to choose to either restrict your losses or maximize your gains.
This is why I divided strategies to either defensive or aggressive.
[B]Defensive strategies[/B]
These are strategies concerned primarily with protecting your capital. It argues that if you do that, you will make money with time. I will not go in depth on those strategies for two reasons: 1) they are the ones you find available all over the internet. 2) It will require me to know the original stops on all the 289 trades that were backtested, which I don’t.
Examples of defensive strategies are the 1% risk per trade and the dollar stop. It’s important to note that with those strategies, it is important to diversify (take several trades of different pairs simultaneously.) This is because the returns from trading one trade at a time will be very little.
[I]The 1% rule[/I]
This strategy calls for risking 1-2% per trade. To calculate that, you measure the distance between your entry and exit in pips. Then you equate that distance to the value of 1% of your capital.
Example: Let’s say I have a balance of $10,000. 1% of my equity if $100. Let’s assume I’m about to enter a trade that has a 100-pip stop-loss. You apply this equation: Lot size (in mini lots)= value of 1% of your balance (in dollars)/stop loss (in pips).
So applying the formula: 100/100= 1 mini lots. Or 0.1 standard lots.
[I]The dollar stop[/I]
The dollar stop asks you to set a fixed number of dollars you are willing to lose on all trades. So let’s say its $100. You will apply the same formula above to this one. So in our example it will still be 100 (dollars)/100(pips)= 1.
So what is the difference between the two methods? [B]A HUGE DIFFERENCE![/B] In the 1% risk per trade your lotsize and risk will increase when winning and decreasing when losing. In the dollar stop method you will remain the same.
Example: let’s say that you have grown your balance to $100,000. Now 1% of your equity is $1,000. While the dollar stop method remains at $100.
So now lets say I’m about to enter a trade with 100-pip stop loss. I will use 1 standard lot on the 1% rule (1000/100). But I will still trade the 1 mini lot on the dollar stop method (100/100).
I hope this is clear.
[B]On diversification[/B]
The problem people face here since they have to trade a basket of currencies simultaneously is how big should that basket be? Well each person sets their own rule. The general guideline however is that you shouldn’t exceed 20% of your balance. That means when you basket trade, do not open more than 20 trades simultaneously (since each one is worth 1%.) So you should look to open at least 3 trades and at most 20 trades.
[B]Note:You will have to exit at the original stop loss of the pattern for this method. Once your stop is hit you are out.[/B]
[B]Advantages and disadvantages[/B]
Well the advantage is it will take a very, very long time for you to ever crash your account with that method. It is almost impossible. I was using the 1% rule back when Switzerland lifted the peg with the Euro. I was in a USDCHF long and I had no stop loss when the pair crashed. People lost hundreds of thousands and some brokers went banktrupt because of that trade. I lost 40% of my balance.
The biggest loser in the history of forex with no stop loss and I didn’t get wiped out. All thanks to the 1% rule.
The disadvantage on the other hand is that you will not good as a good a return as the aggressive versions. Based on the backtests Doug has done. The estimate of a trade for this pattern (with losses included) is $25. So if you open the full 20-trades basket you should expect a return of $500. You can (and you will) earn more on many times. But this what you should be expecting.
I will probably work on the Excel after I finish the money management posts.
To be completely honest for 876 trades in 11 years, I think the avg winners are about right so I don’t expect the data to be wrong at all. In fact I’m quite impressed we never got wiped out.
Yet I agree with you, we will get a much clearer picture on performance once we apply money management.
Just so you know, I have coded the algorithm where RSI threshold filtering can be enabled/disabled (and of course can handle different threshold values), and Initial Stop Loss can be enabled/disabled (my Sims have it disabled to date). Assuming that Initial Stop losses are set to Swing Hi or Swing Lo, for example, I can add this initial Stop Loss value to the Trade Entry text in the Trade Logs. Easily done, if we want to go down that route.
[B]Aggressive Strategies[/B]
Even though they are well known and turn in better results, they very rarely used. I will discuss why later on in this post.
Aggressive strategies are ones concerned with maximize profits but they just don’t give two ****s about protecting your capital. The best way to describe them is that they are the fastest way to lose money. So why use them, two reasons; 1) They are also the fastest way to make money. 2) It is actually much more methodical in losing money than the-flush-your-money-down-the-toilet approach.
[B]General notes on Aggressive strategies[/B]
One of the reasons they are rarely used (but by far not the main reason) is that using these styles requires having a clearly defined trading strategy that has been back-tested over a large number of trades (at least 250 trades).
The argument of these systems is that trading is a gambling process with an open-ended results. That’s way it requires having having a tested system because it will require a number of data from the system’s past performance to apply in the future. These data are: winning %, avg. winner-loser and worst loss. Let’s take a closer look.
[B]Kelly Criterion[/B]
Kelly criterion is a mathematical formula that provides an answer to a very simple: If the odds against a bet I will take are X, how much of my money should I place on that bet?
The comforting thing is that the formula provides a mathematical answer. In other words this answer is factual and not debatable, it is the correct amount. It was treated for forex traders as well. So here is the formula: [B][I]Kelly= W-((1-W)/R)[/I][/B]. Where:
W= win%
R= winners to losers ratio.
Now if you have been following my conversation with Doug you will know this is our first challenge; what is the winning% of the pattern over the last 11 years?
Well we have two answers: one when we will Break evens as losers and another when we count break evens as winners. I normally prefer to count them as losers but most forex systems count them in their winners. It is very debatable. A third option could be to remove them from the count, since they technically are neither winners nor losers.
Well I did all three. The answer according to Kelly was as follows: 1) Counting them as losers would render the strategy without an edge.
Removing the break evens completely would yield a Kelly of 0.12836. Counting them as winners, the Kelly was 0.257898. It is up to you which one you would want to choose. You may ask what this number is (I will answer after I finish all three methods). But let’s move to the next step.
I used 0.257898 as the kelly. To determine your lot size you simple apply this formula: [B]Lot size= (Balance*formula)/biggest loss[/B] The biggest loss for the system in the 11 years of testing was 1053 pips. This number could increase with the new data. But I will update the numbers after I check out the new data.
Just to get an idea of the difference Kelly makes you need to know this. Trading Dough’s trades with 1 mini lot (0.1 standard lots) for all trades would have yielded 6909 pips. I used 1 mini lot because the value of 1 pip is equal to 1 dollar generally.
So basically without any money management, we would have taken a $10,000 balance to $16,909.
Applying Kelly on the same set of trades yielded $33,003. The highest the equity got was $34,500 and the lowest was $7069. I could even share charts of the data. I’ll probably do that on the new ones.
Next is Larry Williams’ formula that helped him turn a $10,000 into 1.1 million in one year. It is the same formula his daughter used to turn 10,000 to a 100,000.
[B]Larry Williams’ Secret Formula[/B]
In 1987(ish), Larry Williams made history when he turned $10,000 to $1.1 million. An astounding feat. He became famous for that achievement and eveyone wanted to know his trading strategy. It was Larry Williams who first brought up the importance of money management in trading because he revealed there was no great strategy and that the answer was only in money management. The formula was revealed, but never caught on for a clear reason that I will describe at the end of my posts on aggressive strategies.
The formula is this: [B][I]Lot size= (Balance*Risk)/worst trade[/I][/B]
Larry Williams recommends risking no less than 5% (0.05 when you input it in the formula). This sounds quite risky when you compare it to the 1% rule.
Well to make matters worse, Williams actually used 25% on every trade on his road to making history.
Insane, right? Well the idea of risking so highly is that he factors in the worst trade in calculation, so he has to make up for it in risk.
To be blunt; Kelly, Williams’ formula, Optimal f and my variation all produce a fraction number. In our example with Kelly it was 0.25789…
What this fraction means is the following: It is the percentage of your equity you’d lose if you’re to encounter your worst loss on the next trade.
So basically our formula is recommending for us to set a lot size that will have us lose 25.8% of our equity if we face a trade that loses 1053 pips. No matter if we $1,000 or 1 million. If we trade the pattern using kelly formula and we lose 1053 pips in one trade we will lose a quarter of our equity. That’s what the “aggressive” part is for. But remember, this number is mathematically correct.
Applying the formula to our trade log, we would turn $10,000 to $32,242. Slightly less than Kelly but way more than no money management. The equity would have grown to $37,187 and down to $7147. So its better than Kelly in that regard.
Notice that our Kelly figure of 0.25789 happened to be very close to 0.25, hence the close result.
So Larry williams risked 25% per trade. His daughter, actress Michelle Williams, risked 15% when she turned $10,000 into $100,000.
So the secret is out. Next we will discuss the revolutionary and complicated optimal f, my favorite.