Reversing an EA

Hello,

Many of us when we started trading forex lost regularly. And I'm sure a lot of people got that idea : If I could just reverse my trades, buy when I sell, take profit when i take my stop etc...

But it doesn't work because of human psychology.

Now with EA there is no emotions in play.

But EAs are mostly losing, many of them are losing with gusto and in a beautiful regular fashion.

I'm not talking of inverting the parameters in an EA, but simply automatically reversing a losing EA actions.

What do I mean?

The EA buys : you sell
The EA takes profit : you take your loss
The EA is stopped out : you take profit

All of that automatically.

I'm not a programmer but there is a solution : zulutrade gives you the option of "reversing" a signal provider. In the exact way I want it to be done.

From monday I'll do exactly that with this EA:

EURUSD

GBPUSD

GBPJPY


I'll start with a live account, trading 0.01lots and update the results in this thread every friday.
Of course I'll keep my regular trading in my main account, so the test will be 100% pure EA reversing.

Stay tuned

Take care
Julien

Hi,

I have a little programming experience, however I'm still a relatively new trader. I read your post and I must say I'm intrigued, and I feel that in theory this idea may be profitable. However, don't you think it would be a better idea to analyze the trades made by the EA and determine where it went wrong instead of blindly reversing the EA.
There also exists the possibility that the EA was right all along, but it is taking losses due to recent news events or market conditions.

That said, I wish you the best of luck and I'm looking forward to see how this idea pans out, I will also reverse some of my losing EAs and check see how they do, albeit on a demo account.

Cheers
Abi

ON EUR/GBP ABOUT 15000 trades
Now with a system that does everyting this EA does but in reverse.
buy=sell
sell=buy
stop=take profit
take profit= stop

What would happen?

Lot size is 0.1


Bars in test915885
Ticks modelled44232270
Modelling qualityn/a
Mismatched charts errors937
Initial deposit10000000.00

Total net profit-317929.49
Gross profit17477.12
Gross loss-335406.60
Profit factor0.05
Expected payoff-21.43
Absolute drawdown317932.83
Maximal drawdown317982.18 (3.18%)
Relative drawdown3.18% (317982.18)
Total trades14839
Short positions (won %)7422 (7.10%)
Long positions (won %)7417 (7.31%)
Profit trades (% of total)1069 (7.20%)
Loss trades (% of total)13770 (92.80%)

Largest
profit trade37.22
loss trade-419.21

Average
profit trade16.35
loss trade-24.36


Maximum
consecutive wins (profit in money)5 (124.64)
consecutive losses (loss in money)177 (-3906.18)
Maximal
consecutive profit (count of wins)130.03 (4)
consecutive loss (count of losses)-4103.36 (173)
Average
consecutive wins1
consecutive losses15

Tell me why it wouldn't work. The average loss size is certainly high enough to include the spread/slippage issues.

I've tried reversing EAs too and it doesn't work because of spreads: you can't open and close trades at same prices.

your EA is random and it has no edge. The equity curve is typical of losses due to spread.

When you do the opposite of random, its still random, and it still wont have an edge, and the equity curve will look just the same

its nothing to do with psychology, its just the lack of an edge

Actually I am following a loosing provider on zulutrade by reversing his positions. He does bring profits this way. But testing a theory with real money is very brave in my opinion. Way too brave for a trader.

A random EA would never give you after 15 000 trades.

Largest
profit trade 37.22
loss trade -419.21

A 1120% deffirence between largest winner an largest loser.

Average
profit trade 16.35
loss trade -24.36


A 49% difference between average winner and loser.


So you do it and it works but it is too brave to do it cause it might work?


Let's see how the live trading will go, first batch of results friday.

No, I am saying that it is too brave to test the theory with real money.

BTW with that reversing provider I had problems last Friday with ghost trades. Two trades ended in -17 pips loss each in his account, which should be translated in +17 pips in my account. Well they didn't and I closed them manually as soon as the market opened. Just another thing you should be careful of on zulutrade's reversal system.

0.01 lots is not real money.

KMidas you said that with the reverse in zulutrade your provider was -17 , and how much were yours if not +17?

In my account it didn't close because, according to zulutrade's support, the provider had changed his tp but the change didn't reach my account. So I closed -46 pips manually.

How is this working so far?

http://www.zulutrade.com/TradeHistor...aspx?pid=96597

All the stats you need are there.

If that 1120% difference is just a single outlier, its probably not a method youd want to trade, but if there's a nice fat tail of losers in the distribution, then its looking promising (I'd definately trade it !).

The stats for this EA are a bit puzzling, it has very high consecutive losers, so I assume its using a very tight stop relative to target, but on the other hand, its largest loss is huge, how does that happen ?

Well the average loser now that it is running live is around 22 pips, with the largest now being over 50 pips. We got one win out of 34 trades.


I'm not quite sure, the EA as just a few parameters. It seems to take loses according to some MACD values but it is unclear even loooking at it running.

Doesn't really matters it does the job!

Hello everybody,
This is the first time I am posting, I have been reading your threads but this is the first time I am posting or replying to a post.
Reversing an Ea or rather binary inversion, does it work or not?
Well I beleive yes but under certain conditions. I have spent a lot of time with this topic of inversion and I have concluded the following:
The formula with which I came up, is this, Rsp=(T/P-2*Sp)/(S/L+2*Sp) where Rsp = Spread Ratio, Sp=Broker Spread and T/P, S/L = take profit and stop loss.
In other words if the Ea gives a sell signal and you reverse it and open a buy trade instead, this trade will open at 3 pips higher (assuming that you are trading with a 3 pips spread Broker).
Now the S/L and T/P, let's say will be 10 pips. The sell trade if lost (if it hits the S/L) will close 7 pips up to make up for the broker commission.
So in this case your profit from the reversed buy position will be 4 pips. If on the other hand the sell trade takes profit it will go down to 10 pips plus another 3 for the spread that is 13 plus the initial -3 pips for the buy, a nice 16 pips loss. So to put in a form of an equation Rsp=(T/P-2*Sp)/(S/L+2*Sp) or Rsp=(10-2*3)/(10+2*3)=4/16.
I hope I am wrong if I am please do correct me, but I am afraid this is tue.
With a ratio of 4 pips win and 16 pips loss all the odds are against you. Ok you may trade with a Broker of say 2 pips fixed spread. In this case the ratio becomes 6/14, this is better.
Now we have another ratio let's call it Rwl=(No of trades won)/(No of trades lost). So if the Ea, for example, wins 30 trades and loses 50 we have a ratio of 30/50.
Therefore in reversal or binary inversion 30 won (lost in inversion) * 16 pips=-480 pips lost and 50 * 4 = 200 pips won, thus a profit loss of -280 pips.
If the case of a 2 pips Broker we would have Rsp=(10-2*4)/(10+4)=6/14 so for a Rwl ratio of 30/50 we would have 30*14=420 pips lost and 50*6=300 pips won, again 420-300=-120 pips lost.
So what can we do?
1. Produce an awfull Ea which has a terrible loss so that we change the Rwl ratio to say a Rwl ratio of 30/80, in this case inversion will give a better result.
2. come with an agreement with broker for lower the spread due to the large volume of monthly trading.
I will say it again binary inversion or Ea reversal is a very atractive idea, but....
I hope and I wish I am wrong with the above calculations, if I am please do correct me I would love to be wrong in this matter.
I am not a programmer I am a mechanichal engineer I know a bit about programming and I use software to make Eas.
I have spent a few years trading with Forex and I know there are long term solution, however we all want and we are attracted to short term.
To conclude if someone could produce an Ea which would simply detect the trades produced by another Ea on the MT4 platform, and open opposite trades but close them at the instant the first Ea would close the normal trades them maybe me o rsomeone else could produce a really awful losing Ea which would result in profit.
By this I mean if awfull Ea1 opens trades at 0.01 lotsize, Ea2 detects those trades and opens opposite at 0.1 lot size, at the instant Ea 1 closes the trade Ea 2 also closes the 0.1 opposite trade.
If Ea1 is a very bad loser then Ea2 will produce a profit, after all it is all mathematics.

Hello all,

This is my first time posting, like this gentleman above (5 years ago, almost). First of all I am kind of impressed by the absence of any answer for all this time...

Getting into the point, the claims above are wrong, because the gentleman above assumes that he pays the spread twice...: Let's make an example:

Assumptions:

1. Initial Margin Capital = $10,000
2. Broker's Spread = 3 pips (which is very high at this present moment of time -not like 5 years ago- yet I am trying to include average slippage and/or commissions)
3. Take profit 10 pips
4. Stop loss 10 pips

Now lets get to a case study:

Now assume that at the moment of the trade(s) are executed the BID/ASK prices for EUR/USD = BID 1.2350 / ASK 1.2353. The following cases are possible:

A. Initial Trade: Buy 0.1 lot EUR/USD @ 1.2353 (Open Price of the Trade), STOP LOSS Price = 1.2343, TAKE PROFIT Price = 1.2363

B. Reverse Trade: Sell 0.1 lot EUR/USD @ 1.2350 (Open Price of the Trade), STOP LOSS Price = 1.2363, TAKE PROFIT Price = 1.2343

C. Initial Trade: Sell 0.1 lot EUR/USD @ 1.2350 (Open Price of the Trade), STOP LOSS Price = 1.2360, TAKE PROFIT Price = 1.2340

D. Reverse Trade: Buy 0.1 lot EUR/USD @ 1.2353 (Open Price of the Trade), STOP LOSS Price = 1.2340, TAKE PROFIT Price = 1.2360


In the above example we duplicated exactly the numbers of the Initial Trade to the Reverse Trade, meaning the SL and TP values.

Now we have the following cases:

i) If the Initial Trade is a loser, then the Loss will be equal to $10 + $3(Com/Spread) = $13 so the Initial Margin Capital = $9,987. Hence the Reverse Trade is a winner, but the profit of this trade will not be equal to $13 but to $7 (and notto $4 as the dear Forexcos claimed in his post), and Initial Margin Capital = $10,007.

ii) If the Initial Trade is a winner, then the Profit will be equal to $10 so the Initial Margin Capital = $10,010. We easily understand that the Reverse Trade is a loser yet the loss will not be equal to $10 but to $13, so the Initial Margin Capital = $9,987.

What we can extract from the above simple paradigm is that when the Reverse Strategy wins it wins 7 pips BUT when it loses it loses 13 pips. (I kindly remind you that we assume that our account is denominated in US dollars so the value of one pip when we trade 0.1 lot is $1, so $7 = 7 pips, $13 = 13pips, etc).

Lets now proceed to what we are looking for: Is it possible and How can we be profitable from reversing a losing EA?

The answer to the above crucial question is two fold: YES, it is possible but IF we overcome the trading transaction costs (OR taking the broker's side)! The (simplest) mathematics of this are the following:

Our predetermined Reward to Risk ratio = Rw / Rs = 7/13 ~ 0.5384 (and NOT Rwl = 4/16 = 0.25 as Forexcos says), since if we adopt the Reverse Strategy as we saw in the above example, when we win we win 7 pips and when we lose we lose 13 pips. Of course there's a huge difference between a Rws ~ 0.54 against a Rws = 0.25. Because if we execute this Reverse Strategy 1000 times (which means taking 1000 trades with the exact SL, TP numbers as the above example) then if the Reverse Strategy has a 50% chance of winning then after 1000 trades we will lose our...pants (I know this doesn't sound so scientific but its the blunt true). The numbers are like: 500 wins = 500 x ($7) = $3,500 and 500 loses = 500 x (-$13) = -$6,500.

So, after 1000 trades our Initial Margin Capital = $10,000 + $3,500 - $6.500 = $7000, and we will have an already huge 30% draw-down. The conclusion is that if you are trading any strategy with equal SL and TP magnitudes (so a predefined Reward to Risk Ratio of 1:1) you need more than 50% probability of winning in order to overcome the transaction costs in the long term. Returning to our matter, if we Reverse a trading strategy keeping its original SL and TP prices then we will need a lot more than 54% chance of winners in order to make a single cent. But how much is this "winning chance"?

Let's assume that we want to win $3,000 instead of losing them. So we want our Initial Margin Capital after 1000 trades to have a positive return with a yield = 30%. The following first degree system of linear equations gives us the solution:

[(X number of winning trades) x ($7)] + [(Y number of losing trades x (-$13)] = $3,000 AND X + Y = 1000 trades

Using the adequate simple algebra we can derive the solution to the above system of equations which is [X = 800, Y = 200]. Thus, we need an Initial Trading Strategy with less than 20% winning ratio and with a predetermined 1:1 Reward to Risk Ratio in order to Reverse it to a 80% winning ratio strategy (with the EXACT numbers of the SL and TP orders of the Initial Strategy followed) in order to have a Return of 30% on our Initial Margin Capital after 1000 trades are taken.

EPILOGUE: I am sorry if I was over-analytic or if my post was too long, yet I have tried to contribute as I can to this matter.