# PercivalFX 17

We are proud to introduce you a new robot based on the concept of “Fuzzy Indicators”. A “fuzzy indicator” is a set made up of multiple instances of the same indicator, each of them with different settings (usually the period).

When we started developing the EA strategy we used the fuzzy version of a single indicator (specifically the Parabolic SAR) scaling in and out trades based on our DLS (Dynamic Lot Sizing) algorithm. In other words the lot size increases as more indicators in a “fuzzy pool” agree on a Long direction of the market, the lot size decreases to zero when the indicators are equally divided between a Long and a Short outlook, and finally the lot size gets more and more “negative” as more indicators suggest a Short direction of the market. While this approach seems to make perfect sense, it turns out it is pretty difficult to find the right parameters that make it work over an extended period of time.

That’s why we experimented a different approach that looked much more promising from the beginning. Let’s say our “fuzzy pool” is a set of 11 fuzzy indicators. Let’s imagine each indicator participates in a “Round Table” and each one expresses its opinion about being Long or Short in the market. Like King Arthur’s legendary Knights of the Round Table each indicator has equal rights. Each of them gives a score (+1 for Long and -1 for Short) and we end up with a global score that can be any number in this set: {-11, -9, -7, -5, -3, -1, +1, +3, +5, +7, +9, +11}. This Round Table will be Long when score >=7 and it will be short when score <= -7. This is the basic idea behind our new robot PercivalFX.

Then we got a step further: Round Table to the square! PercivalFX actually combines multiple RoundTables, each of them being based on a different fuzzy indicator. This helps a lot smoothing the equity line over extended periods of time.

Percival was one of King Arthur’s Knights who completed the quest for the Holy Grail, that’s why we think PercivalFX is a name of good omen for our latest robot.

This is a robot that anyone can backtest and optimize quite easily (thanks to the fact that trading decisions are taken only at the opening of each new bar), and it has a lot of flexibility. Here is the list of indicators it can work with: PSAR (Parabolic SAR), RSI (Relative Strength Index), WPR (William’s Percent Range), SMAEMASMMALWMA (different flavours of Moving Averages), STOCH (Stochastic’s Oscillator), CCI (Commodity Channel Index), MACD (Moving Average Convergence/Divergence), RVI (Relative Vigor Index).

We are working on expanding this list with other standard and custom MT4 indicators.

Are you just curious and want to see it at work? If you have a couple of spare minutes you can enjoy this “pips-churning” video

This video highlights the most important features of PercivalFX EA:

• it makes new equity highs generally every 2/3 months (unless there ia a major trend happening) however each new high is sensibly bigger than the previous one
• it opens multiple trades at the same time (and many times it also hedges positions), because each strategy (based on a fuzzy indicator) works independently from the others
• it is a classic trend follower which hits big during trending periods – it is able to capture a portion of almost all the major market moves – however it is also able to capture small profits during ranging periods, thus reducing losses

The backtest you have just watched is based on a mix of the following “classic” indicators: PSAR (Parabolic SAR), RSI (Relative Strength Index), WPR (William’s Percent Range), EMA (Exponential Moving Average), STOCH (Stochastic’s Ocillator), CCI (Commodity Channel Index), and MACD (Moving Average Convergence/Divergence). Each strategy in this backtest works with 1 minilot, however the user is free to assign a specific lot size to each strategy (also lot_size=0, meaning a strategy is disabled).

We are sure each user will find his/her own best way to run PercivalFX. It only requires some time to explore the wide range of options available (currency pair, timeframe, input parameters).

But results are not really due to optimization. Please give a look at the following Optimization Graphs that clearly point out that the strategy implemented by this EA is pretty solid (click on the screenshots below for a better view):

This is just the first of only three stages of optimization. You can notice that each optimization graph is a “cloud” that extends all over the combination of input parameters.  That means that each backtest (a cloud is made up of 1694 backtests) ends with a positive result. This first stage of optimization takes into account three input parameters: period_numbers (the number of participants in a Fuzzy Round Table), open_score (the minimum score that allows opening a trade), and exit_score (the score that forces the closure of a trade if it was not yet closed with a stoploss or takeprofit level). The fact that all backtests produce a gain (and not a loss) is a clear sign that the “Fuzzy Round Table” algorithm is intrinsically valuable.

But let’s stop talking and let facts talk.

We’ve put our strategy under a long and deep series of tests and what we’ve found is that it performs very well only by doing a mild optimization for each pair/timeframe set.
So far we’ve found 3 pairs working well on the 4 hour timeframe and 2 on the 1 hour timeframe. But the search continues and you’ll be able to find your mix as well.

Here are the results:

## USDCHF (6 Years)

And if we put all the 3 curves in the 4H timeframe together here is what it looks like:

I’d like to draw 3 considerations from the picture above:

• the starting balance is 30k USD and the lot size of each strategy (there are 3 x 7 = 21 of them) is 1 minilot, therefore the maximum leverage is about 7:1 (in theory this can happen if all 21 strategies are all Long or Short at the same time, however this situation never happened during backtests). You can easily scale this strategy on a 3K USD account using 1 microlot per trade. If you are pretty risk-averse then you may allocate 5K USD of your balance to this strategy.
• most of the time the equity line is above the balance line: that means the strategy never tries to greedily bank profits while allowing losing trades to accumulate a floating loss; on the contrary PercivalFX tries to follow big trends as much as possible, therefore the floating profit&loss on the account is positive most of the time.
• there are periods when the balance line drops noticeably (the worst case is between March 2008 and July 2008 in this backtest) because PercivalFX accept losses during a market phase which is not in accordance with its trend following nature; however when trends do develop the EA quickly recover losses and makes new equity and balance highs very frequently. In general each 2-3 months we can expect a new equity high.

You may be wondering if a robot involving so many strategies that work simultaneously can be considered reliable, since it involves quite a bit of calculations on all the “scores” from a plethora of indicators and all the open trades to be managed. We can assure that we’ve coded PercivalFX taking all these concerns into account 😉 You can also stop the EA working on one computer (e.g. your PC) and put it to work on a completely different machine (e.g. a VPS) and PercivalFX will automatically recover the status of each strategy and manage all current open trades

And here’s our latest finding:

## EURJPY (6 Years)

But we were not satisfied and we wanted to see how PercivalFX would perform on shorter timeframes.
We’ve found that the 1H timeframe performs very good for certain pairs and here are the first backstests over the last two years.

## AUDUSD (2 Years)

Impressive isn’t it? We simply stopped counting pips 😀

But there’s more. We decided to start a new way to deploy our EAs and indicators. Starting from PercivalFX.

We want to reward all the people following us and we’ll sell the first 100 copies at an incredible special price. After those first copies price will almost double for the following 100 copies and then reach the “final” price.

This is because we want to involve our customers in the further development and improvement of performances. We’ll do that via private Yahoo! Group. If you’ll decide to buy it you’ll be automatically invited to join the group and that’s where it all will take place 😉

## Suggested Broker

Although the strategy can work pretty well on any reputable broker, we personally use and suggest FinFx which offers top-rated liquidity providers and very competitive spreads.

## What You Get

• PercivalFX EA
• Setting Files for each pair and timeframe we backtested successfully
• An online manual with instructions
• Subscription to PercivalFX private Yahoo! Group in order to partecipate to the performance improvement process of the EA

If you are interested in the PercivalFX EA you can buy them from here.

PercivalFX EA for MT4 – FREE

PercivalFX, the fuzzy indicators round table EA. With best settings and membership to specific private Yahoo! Group.

The EA works on every MT4 platform.

PercivalFX EA for MT4 – \$149,99

PercivalFX, the fuzzy indicators round table EA. With best settings and membership to specific private Yahoo! Group.

The EA works on every MT4 platform.

## 17 thoughts on “PercivalFX”

• Paolo

Hi Rafael, we are setting up demo accounts for PercivalFX that will be available on MyFxBook. We do not provide our live accounts both for privacy reasons and because we trade multiple strategies on them, also manual ones (and to make things more complicated we also use One4All to aggregate trades).

• Special0ne

Can i know the order will close on the TP and SL is set earlier? Or it will modify the TP or SL to close it?
Thanks

• Paolo

TP and SL of a trade are not modified once they are set. However the trade may get closed by the EA before either condition is met, bacause market conditions have changed in the meantime.

• Andrea Post author

Why not “shells/mirrors” for 2000 years? 😀

Just kidding, but we can’t as we don’t have all that data available. We used all the data we have for all pairs. And we’re working on adding more.

• anjelo

Why can’t you? You can get historical data from Alpari or Asirikuy, starting from year 2000. Basically, I like the strategy implemented in the EA. I also like the R:R ratio and how the EA manages the trade. That said, I still need to see how this strategy handles different types of market. The longer the backtest, the better we can see how the strategy can handle different market condition. I am sure, if the backtest result is good, it will lead more sale for you.

By the way, I have create a thread on PercivalFX in donnaforex –> http://www.donnaforex.com/forum/index.php?topic=9658.msg243299;topicseen#new. where you can drop in and answering questions from members there. Good luck and wish you a great sale :).

• Andrea Post author

Hi Anjelo,

what we do is not really for sales. We develop strategies and try to share with all the people following us. Trying to make some money together. We sell them simply because we spend a lot of time doing that. I’d say ALL the time doing that and time is money

Sincerely I’m not a great fan of backtests. Usually I use them only during the development phase to see what can be optimized to have better performances but I usually don’t touch the strategy since. Backtest tell you something but are not a guarantee of anything. So I personally trust more my “feelings” on a strategy than what backtest say. This is because is so many years that I develop and trade Forex.

PercivalFX is a little different as it “needs” backtest to find the right “recipe”. That’s why we used them in this case.

Said that I also believe that strategies need to be “updated” and adapted to the changing markets. I don’t really think that there’s anything able to “always” work. That’s the main reason for our “multi”-timframe/pairs/strategy approach. Doing a parameter optimization using 13 years of data (other that it’ll surely bring bad luck!!) in my opinion won’t take to good results. I personally prefer, in those cases, to do much shorter ones that make your strategy to be better fitted to what the Forex market is now (that is sooo much different from more than 10 years ago).

But I know that there are different thinking schools about that and I respect your view.

PercivalFX has no backtest limit and everyone can do its own optimization and backtest. If anyone is willing to do so, we’ll be more than happy if she/he will want to share her/his results with us. The Yahoo! Groups is made for that.

About Donnaforex forum thread… thanks for that. I’ll give a look at it and try to participate to the discussion.

Thanks again and I hope to have you in the Yahoo! Group as PercivalFX owner 😉

• anjelo

“Backtest tell you something but are not a guarantee of anything.”
I agree. I also don’t trust any strategy merely based on backtest. But backtest has been my pre-requisite whether any strategy deserve much deeper observation or not. It’s not the most important thing and it’s only a part of the story. But still it’s important thing for me.

“So I personally trust more my “feelings” on a strategy than what backtest say.”
I trust what statistical analysis say, especially when dealing with automated trading. I found it much powerful. But I guess I am a very bad manual trader. Hehehe.

“Doing a parameter optimization using 13 years of data (other that it’ll surely bring bad luck!!) in my opinion won’t take to good results.”
You’re absolutely right. So, could you pretty please do me a favor to provide 13 year backtest. You can use 9 year sample for the optimization, and have 4 years out of sample and see what we will have during those years. Honestly, I am really curious on how this strategy behave during those years. I think we can start with EURUSD on 4H. Hopefully, the backtest won’t take so long now that trading decisions are taken only at the opening of each new bar.

“I personally prefer, in those cases, to do much shorter ones that make your strategy to be better fitted to what the Forex market is now (that is sooo much different from more than 10 years ago).”
I absolutely agree. There’s a thing called Walk Forward optimization. It would be great if you also apply this feature to this EA.

“But I know that there are different thinking schools about that and I respect your view.”
Which different thinking school? I thought I agree with almost everything you say. Except “trusting the feeling” but it’s because I am a bad manual trader.

Anyway, you really seem to develop a good system here. Congratulations for that. :). I will surely be the owner of PercivalFX if it fulfills all of the long term profitability requirements.

Cheers,
anjelo

• peter

Hello a tutti,

I tested this expert now for nearly 2 month and the only thing what this expoert prodcuces are losses every week….

I work now more than 7 years with experts and i had tested several 100’s of them, most are useless.
This is nothing new, i know.
The idea behind percivall will be good but this expert is just not a finished product. I’v tested this expert on 2 different demo account and on both, percival just produce losses.

Percival is in my personal top 10 of the worst perfoming experts this year.

Now i also understand why you are not working with a fair 60 day money back on your products.

Thanks and best regards,
Peter

• Paolo

Hi Peter,
most EAs do fail a 60 days test even if they are great EAs. That’s because market cycles can be longer than a few weeks. The same day of your comment we received the request of another customer asking a second live license of PercivalFX. As you can understand the fact of easily granting refunds to customers is just a way to ruin a legit business due to a plethora of random variables that are out of any control: market behavior, client’s lack of discipline or knowledge, unexpected technical issues on the trading platform, etcetera. We try to give all the background of each product we release, so that each customer can decide to buy with the widest range of information available.
Please send us a private email for discussing further your issues with this EA, we always strive to make customers happy.

• Cliff Monzeglio

I sent this on a different thread, so it may have been answered, but I don’t know where to look:

My problem is with the US FIFO rule:

When PersivalFX opens multiple trades in a single currency, then attempts to close one of the later trades, the FIFO rule won’t allow the close. Then a message is sent informing of the failure to close.

This sequence happen repeatedly with increasing frequency causing the trading platform to freeze if I’m not sitting in front of the computer to manually close the earlier trades.

Is it possible to provide a work around that could be compatible with the US’s no hedging-FIFO rules? This could then be turned on in the Inputs when using a USA broker.

Thanx,

Cliff

• Laurence

Hello a tutti,

Question:

After I purchase PercivalFX EA , will it work as depicted in the video with settings “out-of-the-box”.
Also, what indicators are necessary to get these remarkable results…