Virtual Trading - martingale solution
Hi Minime. Seeing as you've opened an 'Improved Martingaling' thread, let me add my 0.00002 pesos.
I'm pretty sure that anyone who can program and has attempted to crack the martingaling-nut will have tried what you have suggested. But how about looking at it from another angle - most probably realise that probably 98% of every 'series' of trades a martingaler opens will close profitably (it's the other 2% that blow your account), meaning that the vast majority of trade 'series' that are opened are made up of just a few trades (2-3) - so why not try and use this to your advantage! (from my experience, it is far from easy...) How? well, instead of trading 'virtually' at the start, trade for real, and if things don't pan-out, switch to "virtual-mode" - the opposite to what you suggest - then try and figure out when to switch back to real-trading! As an example, it might be best to switch to virtual-mode if a news event causes your currency-pair to rocket away, rather than continue opening orders in the 'hope' (dirty word that) that you catch a reversal... Another thing I discovered almost by accident is how crucial lot-size calculation is (yes, back to boring MM) when martingaling - most EA's let you set an initial lot-size that you think will make you gobs of money, then the EA code blindly martingales this size, ultimately to oblivion - in other words, there is no account taken for slippage, down-time etc.
btw, after lot and lots of backtesting different pairs, I think there is no question that USDJPY is best where martingaling is concerned.
Hi omelette,
Thanks for the suggestion, your opinion is valid and we have tried it on the 5thelement expert, which is to limit the number or open trades, but it didn't work well " from my side ", as the 2 or 3 time the price move more than 200 pips - last year we saw 400 pips- on a 6 level limitation the account was blown, the draw down took all the profit made in two months, and a margin call was the last exit I have seen.
You are right, most of the profit comes from moving between the first 4 levels and the loss will happen at the stages I am suggesting to trade at, therefore if we apply what I am suggesting the expert will trade less, and we will have less money.
but personally I prefer less trades with less draw down than more profit with big draw down.
another attempt was hedging but it didn't work out as well , as the trades stays open for god knows how long.
I have an expert that does what I am suggesting but not virtual trades, instead it count the last 3 bars and if the move was x number of pips it open a trade, more or less like what I am suggesting but I am not using any martingle, I having about 1-2 trades on GBPUSD each week and the winning ratio is more than 80%, with small acceptable loss.
as you said it goes down to MM, and it all depend on the trader, and how he wants to handle his account.
Regards,
Alan
Hi Minime. Seeing as you've opened an 'Improved Martingaling' thread, let me add my 0.00002 pesos.
I'm pretty sure that anyone who can program and has attempted to crack the martingaling-nut will have tried what you have suggested. But how about looking at it from another angle - most probably realise that probably 98% of every 'series' of trades a martingaler opens will close profitably (it's the other 2% that blow your account), meaning that the vast majority of trade 'series' that are opened are made up of just a few trades (2-3) - so why not try and use this to your advantage! (from my experience, it is far from easy...) How? well, instead of trading 'virtually' at the start, trade for real, and if things don't pan-out, switch to "virtual-mode" - the opposite to what you suggest - then try and figure out when to switch back to real-trading! As an example, it might be best to switch to virtual-mode if a news event causes your currency-pair to rocket away, rather than continue opening orders in the 'hope' (dirty word that) that you catch a reversal... Another thing I discovered almost by accident is how crucial lot-size calculation is (yes, back to boring MM) when martingaling - most EA's let you set an initial lot-size that you think will make you gobs of money, then the EA code blindly martingales this size, ultimately to oblivion - in other words, there is no account taken for slippage, down-time etc.
btw, after lot and lots of backtesting different pairs, I think there is no question that USDJPY is best where martingaling is concerned.I tried coding an expert with this similar idea a while back.
Statistically speaking, eventually in a martingale progression the current trade will win. The question now becomes this - Statistically how many losing trades can happen in a row, before your current trade will win? It does add the odds more in your favor to have these "virtual" trades.
What I coded kept track of the "virtual" trades, it would wait until there were X number of losing "virtual" trades before entering the a live trade into the progression.
This way trades less, and stacks the odds more in your favor, but still seems to get into trouble that 2% of the time. It's a great idea, but it always seems to be the same old story with martingale EA's. There is always that one time that loses 20 times in a row!
More Martingales
MiniMe,
thanks for the thread, many times I have tried and walked away bruised, and if I had a dollar for every Martingale variant I have tried I probably could have retired rich by now.
Currently I am testing an ea i wrote after reading a thread at the factory. It is more an average down type ea but I use the daily range as my pip space and only if an indicator shows a signal.
I am increasing lots but each trade has a take profit so that any gain is banked and works toward the profit goal. Of course I am paying for spread each time but hey, I am prepared to try anything.
I will attach it for your interest.
I have a separate ea for long and short.
By using a large pip space I am hoping I can set up ea's in the opposing direction to offset drawdown.
..... Currently I am testing an ea i wrote after reading a thread at the factory. It is more an average down type ea but I use the daily range as my pip space and only if an indicator shows a signal.....
@Minime - A strict implemetation of what you are suggesting could well be better, however, my variants proved inferior to the other idea. Below is a version of the '10pointhree' EA I modified 6-9 months ago while checking out various ideas. One is a 'buffer-zone' around the current price, where no trading is done - a bit like your 'virtual-zone'. You can indeed find settings that allow successful run-throughs of all backtest data, but the overall profit is unacceptably low. Reducing the 'zone' will see profit increase, but so does the risk of blow-ups. btw, it's been awhile since I looked at it so I don't recall off-hand if there are any other half-baked ideas lurking in the code - so it's definitely not for real-accounts!
@waltini - I think I know the FF thread you are referring to - it's basically a 'helper' EA that you use when manual trading, that will semi-martingale your next trade if your last was a loser. I also coded this idea. However when I sat down and thought about the idea, I decided it wasn't for me!
PS: - I have just re-read your description. Not the same thread...
Hi mini, hi all
Besides some other stuff, I'm constantly working on 10points and 5th element EAs too.
For example one of the 10p3 with modified trailing stops gave me a meager 18%profit in eurusd for the last 2 months. But the expert survived the huge moves at least.
And what really worked out well for both of them is going with the trend only,
instead of countertrend trading or putting them blindly in the market.
But to the topic above,if we pick up omelettes proposal and instead of "virtual trading" the upper levels
(which is the same as -let losses run- as far as I understand it)
the EA could freeze them, by hedging the sum of all open lots in the opposite
direction. ( No real further martingale trades would be opened from now on )This requires not more lots than they would be needed for the next step of the original martingale. ( next martingale-step lotsize minus the initial lotsize we have started with : 0.1 + 0.2 + 0.4 + 0.8 = 1.5 = 1.6 - 0.1 )
By this the account equity dosn't move a single millimeter.
And when the sea calms down and our pair is going to move in our favorite direction, the EA closes the hedgetrade.
This should be repeated one level above if we are on the wrong side again.
I think the spread would probably eat up the profits and maybe we would get out with some (small) losses, but after that we still have an existing account.
What do you think?
Or do I have overlooked something?
hbud
The main problem (as Minime suggested) with hedging is that you can end up having trades upen for an inordinate amount of time. I have also tried hedging and couldn't get it to work. The problem I found is that you are already experiencing d/d when you open the hedge, your d/d is compounded (most likely) when you decide to close the hedge, and you still have zero guarantee that price will not just turn around and dig you a deeper hole! Not dismissing the hedging idea, just relaying my experiences with it...
btw, the infamous Aleccoh and his imaginary "trading Team" traded my account with a martingaler that hedged the primary pair EURUSD with the correlated USDCHF pair - with a 1-step-up, 3-steps-down final result...
PS. Mistake. The martingaler didn't hedge, it was another EA that hedged. Still lost money though...
Just a screenshot of backtest results of the "reverse-method", which has largely been borne out by forward testing - 'largely' 'cos the version I forward-test has different 'greedier' settings! The PF is unspectacular, as is the equity curve - roughly 80% p.a. return though - but at least it points in the right direction 
What is interesting is the drawdown however, especially for a martingaler! Moreover, 'restrictions' imposed make it virtually impossible to blow a reasonably sized account!! ($10,000) btw, this covers from Sept '06 -> present - the only period that is worth a damn as far as backtesting results are concerned (imo), and it only seem to produce viable results with USDJPY...
Minime, excuse this brief interruption in thread-flow but I've had time on my hands (scary market today 
) - back to your original idea!
Dear omelette ,
feel free to post the ideas, I am here to learn and if you have already tested the idea and it worked then great ... I will follow your steps.
a better approach than hedge is the virtual hedge using correlation trades but for that you need to consider the +ve sawp trades only " Like buy EURUSD and sell USDCHF only " inorder to close the trades when a small profit happen ...
this type of trade, has the same problem as the hedge as the trades could stay open for a while , but with this method of virtual hegde there is a better chance of closing the trades at profit.
Unfortunately, you are right omelette. I made a demo trade this evening the
described way.After the hedge opened I manually removed all TPs and SLs,
watchted the account figures and made my calculations ( which I should better have done before ), only to found out what you have said .
The good thing trading this way is, even if a margincall occurs you cant loose more than the frozen losses, no matter what the pair is doing.
The bad one ,of course nobody wants to wait for days or weeks for this happening. Untradeable .
due to the aleccoh story, have you already seen some of your money back?
I'm not updated on this. I think I have to go to the other thread for reading the latest posts.
hbud
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I opened this thread to propose a solution to the martingle expert problem
I have already suggested the setup in a private forum but since privacy and manners have different interpretation between people, I am posting it here so more people has the benefit or using the idea.
Most martingales are based on taking a couple of losses, while double the lot size, each time a new trade is open, then wait for a market retracement to recover from the loss.
So the main problem is account balance , and draw down, a huge move can wipe the account and we have seen that on all martingales without exceptions.
What I am proposing will make lots developers happy, as this will help them solve their draw down problems.
Some of the good experts I have seen on TSD were the
"FiFtHeLeMeNt Blessing EA" another one is the "V1andV2 Hedged EA” and maybe the “10point3
Several attempt has been made to lower the draw down, but none has succeeded.
The FiFtHeLeMeNt experts take a trade and if the market went against the trade the expert double the lot size and wait for a market retracement, to recover from the loss, same with the other two experts.
Most if not all martingales trade either EURUSD or GBPUSD as the daily range is known ,and normally we had to see a 30 pip retracement or less on each 100-200 pip move.
If the market went for a more than 200-300 pips without retracement the account is blown.
Now here is the idea, if we skip the first 3 level of martingale trades and open a trade on the 4th, not only we have lowered the draw down but also we have increased the winning chances.
So for example on the FiFtHeLeMeNt expert we had 6 levels,
My proposal is to mix virtual trades with real, for example
1- Level 1,2,3 virtual trade
2- Level 4,5,6 take a martingale trade based on 0.1 then 0.2 then 0.4
So we skip the first 3 levels but we keep track of the them in terms of open, close, TP and SL, and when the chance comes to open the 4th trade we open a real trade but with a 0.1 or 0.01 lot size
I hope you guys like the idea and I hope it will help solve some of the martingale draw down problem
Regards
Alan