Another bit of Martingale lunacy

Note: I no longer make my work available at FF. Instead, I have my own forum. You can find this by looking at my profile and finding the 'About SteveHopwood' section.


SERIOUS WARNING

1. This robot will kill your account if you trade it incorrectly. You must read the trading hours section.
2. Even if you take notice of trading hours, this ea will still kill your account if you over-trade.

I received this from mysticgenie earlier tonight:

"Hi Steve,
Here Mystic agan!
I need your help to make a EA on this martingale type system. I have been trading this manually with good results during peak hours.
These are the rules:
I need your help to make a EA out of these simple rules to be traded on EurUSD ( can be in any pair, but I prefer EUrUSD due to low spread).
The EA should be MT4 compatible with 5 digit broker.
These are the rules:
1. Open EurUsd chart 5 min frame.
2. Open a buy order randomly
3. TP 10 pips from the open price; SL 10 pips from the open price ( no need to include spread).
4. Open 1 lot initially ( this should be adjustible with lot change feature with micro and mini lots options).
5. If the position stops out then open sell order with 2 lots with TP and SL 10 pips.
for exmple, you open initially 1 lot buy eurusd at 1.4550. TP 1.4560; SL 1.4540.
current price at 1.4540; you open 2 lots sell at 1.4540 with TP 1.4530, SL 1.4550.
if this hits 1.4530 you will win.
Suppose, price does not hit TP, but moved back to 1.4550, then EA should open buy position with 4 lots.
so the sequence of lots is: 1, 2,4,8,16,32,64,128 (Martingale strategy)
the TP 10, SL 10
the EA should continue to open positions in opposite directions till the price reaches one of the TP levels.
This should not take long since it is only 10 pips.

6. Please make the trading hours adjustible ( that is I can specify the trading hours).
Please let me know if you need any clarification.

Thanks so much in advance,

Mystic "

I coded this just for fun - the only way to treat an ea such as this one.

Newbies, there are two things you need to know about Martingale;

 

1. it will perform fantastically well right up to the point that it blows your account.
2. it will blow you account sooner or later. Probably sooner.

One addition to mystic's strategy I have made is to add MaximumAllowableLot to the inputs. When the Martingale sequence exceeds this figure, the bot restarts with a Lot buy trade. The ridiculously high default effectively turns this filter off; it is there for those of you who want to play with limiting the sequence size.

Trading hours
Enter these in your own computer's local time, not your criminal's server time. They should represent the period of a few hours from the start of the European session, or maybe the London opening an hour later. You will have to work this out for yourself.

Copied from post 18: "Turn on during late Asia early London open, and late London and early US market overlap. these times market move significantly no matter what as long as they are open. Monitor your positions if possible. Start with 1 microlot every 1000 usd ( leverage 1:100 or higher). you will get profits and can make 5-10% per month without much sweat."

Initial trade with the trend
This is not part of mystic's strategy.

This ea should be successful so long as there is movement in the market, so I have added my favourite Rsi filter. Choose your own time frame and period, then:

 

1. Rsi >55 or < 45 means the market is trending either up or down, and so the ea will trade
2. In between these numbers, the market is ranging and the ea will not trade.

It would be a matter of moments to add code to make the initial trade in the sequence be with the identified trend. Let's see how things go as they are.

Measuring volatility with Bollinger Bands
This is not part of mystic's strategy.

What this strategy really needs is a moving market. Bollinger Bands expand with increasing volatility and shrink as the volatility decreases. I have added a BB filter to see if this helps. Where the bands are expanding the bot will trade; where they are shrinking, it will not trade.

Starting a new trading sequence from scratch
You should only leave this robot on your chart during the appropriate trading hours and should remove it at the end of each session.

At each restart, the bot will start a new trading sequence at your chosen initial lot size. You have the choice of deciding which direction the initial trade in each sequence will be with these inputs:

 

1. StartLongSequence: will send a buy as the initial trade
2. StartShortSequence: will send a sell as the initial trade

If you wish to start trading by picking up where you left off in a previous sequence, then use one of the attached market long/short scripts. Drag the script onto the chart and select your initial lot size. Once the trade opens, drag the robot onto the chart and it will pick up as though it has sent the open trade.


Semi-automating the initial trade direction
Over in the DIBS thread. Peter Crowns states that the trading bias for the day can be taken from the open price of the day. If the market is above this price the bias is long; below is short. The bot creates a line on your chart that the robot will consult before sending the initial trade in the sequence. Once the line is on the chart, adjust it to represent whatever is the price of your choice. PC states that the daily start should be 6.00 GMT for those of us trading the European and NY markets, so I am using this as the start of my day. You can use any opening price you want.

Turn AutoCalculateTradeDirection to true to use this feature.

Allowing only one successful trade per session
To stop the robot immediately taking another trade after one has hit its tp, set StopTradingAfterTpHit to true. The bot should send no further trades until you re-drag it onto the chart, or the next tsazb- session opens.

Pivot
mystic asked me to add a daily pivot so the bot can trade off this.

The inputs are:

 

1. UsePivot: turn this feature on. If this is turned off, then the yellow line is draws at Bid - 20 pips and you can adjust/ignore it to suit yourselves.
2. PivotTimeFrame: set this in minutes - the default is D1. This allows you to use a different tf pivot to the chart you are using for trading.
3. CriminalHasSundayCandle: set this to true if using the D1 pivot and your crim has a Sunday candle. It forces the bot to use the hi-lo of Friday, not Sunday night.

Advice to all,
Please trade carefully.
Try forward testing for atleast 6 months.
My appreciation to Steve's hard work,
Best wishes,

There is a mention of sell orders too

"5. If the position stops out then open sell order with 2 lots with TP and SL 10 pips.
for exmple, you open initially 1 lot buy eurusd at 1.4550. TP 1.4560; SL 1.4540.
current price at 1.4540; you open 2 lots sell at 1.4540 with TP 1.4530, SL 1.4550. if this hits 1.4530 you will win."

Oops. I seems I can't be assed to read my own first post properly.

For sure, I coded the wrong spec from mystic, and so the robot is useless - even more so than I first thought.

I am removing the bot from post 1 and closing the thread until I have time to look again.

Lol, well before you removed the EA I was able to download it and run a couple tests. Standard 10 pip TP/SL flopped, 25 pip TP/SL is doing well (so far) backtesting.

Edit:

Seems that a higher TP/SL is needed to avoid entering too many times too quickly blowing up the account. This means less profit but it would be safer. Since it currently only trades long what if we chose a pair with a good + swap rate for long positions since this EA is always in a trade. It may be a small lot size but hey it's still some extra pips daily.

Wouldn't it make more sense to only enter Buys OR Sells, not reverse? Here's my reasoning with a screenshot of a 25 Pip Renko for conceptual purposes.

For those who dont know, each renko "brick" is 25 Pips.



Now, assuming you "Bought" randomly at the close of the bar I marked A. This trade would go against you, hit your SL, and you would then sell at the close of B, which would do the same, and so on and so forth until you hit the G close, at which point you would have a string of winners.

Now the problem comes with the martingale doubling.
A = 1
b = 2
C = 4
D = 8
E = 16
F = 32
G = 64 which would be a blown account for most people.

Now assuming you only bought, you would loose, then win, then loose, then win, then loose, then win.. etc. You would only run into problems when you hit long runs like this.



As you can see between May 13th and May 15/16 there was a 225 pip run that would have had you in for 9 loosing trades using a buy only strategy, which once again would have blown your account using martingale.

Simply randomly entering for 10 pip TP with a 10 pip SL or 25 or 50 or whatever, and reversing OR only buying OR only selling WILL eventually blow your account using martingale. There has to be some kind of statistical edge to a system other then flipping a coin.

I personally find using a renko breakout strategy usefull, where you wait for those formations like in picture A, then when you see a 2 brick move away from one of those formations, your "more likely" to have a winning trade if you enter with that run.

Still, just tossing this out as a cautionary for those who may not know or have thought too much about it.

OK, sorry about the mixup. The bot now trades as required and I have re-opened the thread.

I have added an input to allow users to control the maximum size of the sequence - see post 1.

I will test it and see if everything is fine.
Again you are da man!

Steve, Can we open multiple positions instead of opening a double lot on the opposite side if we hit the stop loss.

Nope.

Pure, unlimited marty is a killer. But if there's an underlying strategy that's profitable in itself and has a reasonable profit/loss distribution pattern, a limited marty can actually (surprisingly) improve the expectancy. Let's forget the buy/sell/buy or buy/buy/buy coin-toss for a while (there's no edge). Take some KISS "strategy" like this:

Pick a pair and dig some unique property of that pair; some 'hard-coded' tiny edge like this one which means: eurusd / only trade first hours of day. Enhance the probability of selling high or buying low with some intraday trend indy.. rsi(14) on h1 or rsi(4) on h4.. classic values.

if rsi >= 55 buy,
if rsi <= 45 sell,
if 45 < rsi < 55 stay out,
timeframe doesn't matter, just trade the hours and the tick that rsi crosses the level.
keep 1:1R, TP/SL=50 pips.. or 40% of 3-day ATR..

and play every single chance with this (or another, or more - the more the better) slight edge(s). Not just the initial trade, but all trades. The market doesn't go up OR down. it goes up AND down. There's nothing worth in keeping direction or reversing direction just because the init trade was a buy or a sell. Yes, price finally will move enough one way, but it will make a stop-out whipsaw (or the opposite - a parabolic runaway) sequence along the way too often to make it unreliable in the long run. See the lenghty '100% Win Math Grid EA' thread here on FF; you can't beat the market with random entries and exits. Even RR's coin toss entry strategy relies on some edge with exits to be successful. Plain presence of (presumed) volatility is not enough. Volatility + directional bias is needed with any trading strategy.


2010.01.01 - 2011.06.04, constant 0.01 lots:
http://i53.tinypic.com/2uivh8o.gif
http://i51.tinypic.com/2i254m0.gif

And now the same with limited marty (2nd level max; 0.04 lots):
http://i52.tinypic.com/vz8q9u.gif
http://i52.tinypic.com/2hdnnk4.gif
Same trades, double profit but not double drawdown.

Precisely!!! Gotta have that edge .

This ea is quite entertaining, so I have added an update to post 1 - see the trend section at the bottom of post 1.

Hello Steve,

did you ever play around with an average rsi value, e.g. (rsiTF15 + rsiTF60) / 2 or (rsiTF5 + rsiTF15 + rsiTF60) / 3 ?
Would that be an improvement?

Depending on the rsiPeriod and the timeFrames, that could be used:
RSI > 80 : OB (sell signal)
RSI < 20 : OS (buy signal)
RSI >< 55/45 (same as you wrote above)

I forget where I read about, I assume it was a post by scooby-doo in that forum here (BOJ thread).

One indicator showing the average value is attached (I changed some existing indicator for that). Same would work for stochastics, too.

edit, simple function to determine average rsi trend:

I just had a bath. Read the result in the Bollinger Bands section in post 1.

This thing is very high on the Fun-O-Meter...........But I have ear plugs in....

whip

Wise precaution.

Interesting developments so far.
Like Steve's idea of BB filter.
Basically as long as market moves more than 10 pips upto 20 pips you will get your profit ( if euro is trending up). If euro is down, you will lose first position, but the next sell position will bring you the profit as long as it moves 20 pips.
You can just manually turn on or off this EA during market over lap hours.
Turn on during late Asia early lodon open, and late london and early US market overlap. these times market move significantly no matter what as long as they are open. Monitor your positions if possible.
start with 1 microlot every 1000 usd ( leverage 1:100 or higher). you will get profits and can make 5-10% per month without much sweat.

Best,

Could an adx-filter also be suitable?

For example, enter trades only if
(+di > 25 || -di > 25)
&& (adx > 25)

Latest update in post 1:

 

1. Removed the live trading restriction in light of mystic's comments about trading hours.
2. Added the ability to choose the direction of the initial trade in the sequence - details in post 1.
3. Added the ability for users to initiate the sequence manually, then drag the robot to take over running the rest of the sequence - details in post 1.

I was too late for the morning session here in the UK, but I caught the NY open and made a few pips on demo.