Martingale with "low" margin hit

[garyfritz]

I'm also wondering what you're expecting from the strategy in general? According to my testing, there are sets of parameters that can work on almost any particular pair. But the params are different for every pair, and they are not very robust. Plus the pips of profit in the best cases will almost certainly get eaten up by slippage and spreads. The whole thing strikes me as pretty fragile and very unlikely to trade profitably in forward testing over the long run. Maybe you could pick separate params for each pair, and hope they continue to work, but that's asking for curve-fitting.

I've experimented a fair amount with Max Positions, SL/TP size, Multiplier, etc, and this thing doesn't look like anything I'd want to risk money on. If you could figure out what causes it to go into a losing streak so you could filter out those trades, then it **might** become usable. But I think there are other approaches that are more worthy of time & effort.

[DragosDanescu]

I mean, it's ok, I have managed to teach the Ea to adapt to any kind of lotsizes, based on various variables calculations, closing at whatever profit/lotsize you want...but, if someone can give me a success rate bigger than 70%, lets say, I can implement that trigger and make the EA to work with that profitably. At least, that's what I believe. The most important thing to watch for is the losing streak, how many times in a row we lose. That's all.

[blahn]

Perhaps I could suggest a slightly different strategy. How about changing the grid to a unidirectional grid instead of bouncing uncontrollably back and forth. I've been following a signal provider at zulutrade who has laid out a 25 pip spaced uni grid on usd/chf going long since it had that nice pivot in early august. you can look at all his trades and open orders here to analyze his current grid.

http://www.zulutrade.com/TradeHistoryIn ... ?pid=27850

He's made over 6000 pips between the pivot and now with that grid. at the start of the grid he makes a buy and puts one buy PO above(at +25 pip) and about 3 buy POs below(-25,-50,-75). almost every trade has a 20 pip TP but he does let them run up on big swings. If the top PO gets tagged he'll create another one +25 pips above and every time a PO below gets tagged he opens another PO below the last one(-75) so the grid can follow the price down until it hits whatever support level he has targeted. I believe he's using the daily chart as his major trend and support indicator since he's never bothered with 10 to 15 level drawdown, although it freaks out alot of his followers I've manually placed similar grids on eu and it does work. should also help control that DD and Martingale(should you choose to use it) to a user specified level of risk.

[spyderman]

What do you expect from the EA to do, exactly, cause it is still there...?
I was playing with it last days, but just let me know what you expect it to do, triggers, maxlots, what closing rules, etc, etc.

Hey DD,

Thanks for the response. I did try to see if I could utilize you EuroGrid EA but just couldn't seem to figure out all the parameters. As someone who's not a coder, it's hard for me to look at something designed for another purpose and see how I can make it work. To be honest I couldn't even see how to make it initiate a trade.

More to the point as to what I'd like to see in this EA:


To the extent possible the user could select parameters:
EA opens a trade (User defined long/short for immediate open or at particular price)
(S/L and T/P set by user)
If S/L is hit EA places a market order in opposite direction at a multiplier selected by user.
(EA has to round this "product" number up to nearest tradeable size. For example if you
put in an order for .015 lots, Empty4 will round it down to .01 by default. EA would need to
round up to .02)
S/L and T/P would be same distance as initial order.
MM settings:
Initial Exit: User could define number of cycles before EA would look for a BE exit as opposed to
waiting for hitting of T/P.
(For example with 1.3333 multiplier lot progression would be.01, .02, .03,
.04, .06, .08, .11. Users sets initial exit strategy at 7 cycles. EA continues
it's thing but looks to exit at the next BE of all cycles. This is one part I don't
know if it's possible to code in. Can an EA keep track of the losses on the
previous S/L's in order to know when the net has hit BE.
Emer. Exit: User defines number of cycles before EA places no further orders and market takes
to S/L or B/E as set earlier. (As Gary and I have discussed this would eliminate
the Death Trade of a classic Martingale.)

Would be nice for EA to display status of trading for current pair on chart.


That's basically the nuts and bolts that I would like to see. I'll probably think of some others but that's it.

[spyderman]

I'm also wondering what you're expecting from the strategy in general? According to my testing, there are sets of parameters that can work on almost any particular pair. But the params are different for every pair, and they are not very robust. Plus the pips of profit in the best cases will almost certainly get eaten up by slippage and spreads. The whole thing strikes me as pretty fragile and very unlikely to trade profitably in forward testing over the long run. Maybe you could pick separate params for each pair, and hope they continue to work, but that's asking for curve-fitting.

I've experimented a fair amount with Max Positions, SL/TP size, Multiplier, etc, and this thing doesn't look like anything I'd want to risk money on. If you could figure out what causes it to go into a losing streak so you could filter out those trades, then it **might** become usable. But I think there are other approaches that are more worthy of time & effort.

Gary,

Again I appreciate all the number crunching you've done. I can't intelligently comment on the results as I've not run them. That's why I would like an EA. As I mentioned before, I've done some visual scans of charts for 2011 and that coupled with a certain amount of logic (albeit biased logic ) has led me to think that the occurrence of a ranging PA in "selected" pairs that would trigger 10 or more 25 pip swings without hitting a 75 pip T/P, would be rare.

If you could scan your numbers and point to some examples that they've shown of market action that would disprove that then that would be beneficial.

Thanks again

[spyderman]

Perhaps I could suggest a slightly different strategy. How about changing the grid to a unidirectional grid instead of bouncing uncontrollably back and forth. I've been following a signal provider at zulutrade who has laid out a 25 pip spaced uni grid on usd/chf going long since it had that nice pivot in early august. you can look at all his trades and open orders here to analyze his current grid.

http://www.zulutrade.com/TradeHistoryIn ... ?pid=27850

He's made over 6000 pips between the pivot and now with that grid. at the start of the grid he makes a buy and puts one buy PO above(at +25 pip) and about 3 buy POs below(-25,-50,-75). almost every trade has a 20 pip TP but he does let them run up on big swings. If the top PO gets tagged he'll create another one +25 pips above and every time a PO below gets tagged he opens another PO below the last one(-75) so the grid can follow the price down until it hits whatever support level he has targeted. I believe he's using the daily chart as his major trend and support indicator since he's never bothered with 10 to 15 level drawdown, although it freaks out alot of his followers I've manually placed similar grids on eu and it does work. should also help control that DD and Martingale(should you choose to use it) to a user specified level of risk.

Blahn,

Thanks for your comments and trade idea, but that is obviously quite a different strategy and not what I started this particular thread for.

For now, I'm going to stick with trying this idea out until I can see the results (obviously I'm trading other strategy's as well)

[garyfritz]

If you could scan your numbers and point to some examples that they've shown of market action that would disprove that then that would be beneficial.

Depends a lot on the pair and the params you feed it -- which BTW is an indication of how unstable this thing is.

But as an example, trading EURUSD with TP=70 SL=-100 Mult=1.5 MaxPositions=7...

MartEQ.gif

It made over 4500 pips between 1/1/10 and 4/26/11, with a very nice steady growth. Then it lost 1111 pips in the next trap. It looked great between 5/4/11 and 8/8/11, but it has lost almost 3000 pips since then, with big losses around 8/10, 9/20, 10/25, and 11/17. For example, here is 11/17:

MartLoss.gif

You'll see different results unless you use exactly the same parameters I used, but that gives you an example.

The important point is, this thing has blowups just like any Martingale. By exiting after N positions you can limit the size of those blowups, but that just means they happen more often. Sometimes it works. Sometimes it doesn't.

[spyderman]

It made over 4500 pips between 1/1/10 and 4/26/11, with a very nice steady growth. Then it lost 1111 pips in the next trap. It looked great between 5/4/11 and 8/8/11, but it has lost almost 3000 pips since then, with big losses around 8/10, 9/20, 10/25, and 11/17. For example, here is 11/17:

Gary,

Thanks for the hard examples. You've definitely found one of the thornier ones here. I've marked up the chart below based on actual trades that would be taken. As can be scene in this scenario we'd go through 17 cycles before actually hitting T/P.

Even though I would want out long before that if you started at .01 lots you'd only be up to 2.0 lots with the low multiplier. A lot better then 655 lots you'd get with a standard doubling up Martingale.

But with the exit strategy I want to employ you could have easily exited at BE at either of the circles. I wouldn't of had it set at the first one with only 2 cycles but definitely on the 2nd with 8 cycles.

[spyderman]

Thanks for the hard examples. You've definitely found one of the thornier ones here. I've marked up the chart below based on actual trades that would be taken. As can be scene in this scenario we'd go through 17 cycles before actually hitting T/P.

I tried to put some $ signs to the numbers. If my figures are right, with a $3000 account, if you took it all the way to 17 cycles you would have lost $1,500 to that point and then got $1,537 when you hit T/P.
Again, I would have been long gone, but interesting to see how it would work out. Not quite a death trade either.

[spyderman]

But as an example, trading EURUSD with TP=70 SL=-100 Mult=1.5 MaxPositions=7...

Gary,

I see one area where we are getting a variance.
I have definitely shifted my thought process to closing out the loosing entries as they hit the 25 pip s/l. This enables us to use a lower multiplier of 1.3333 with the same T/P outcome. That should make a big difference in your calcs.

It would be interesting to test a smaller T/P which would necessitate a larger multiplier.