Statistical Gambling System (High Winrate, but NO EDGE)

Hi guys & girls, using my statistical knowledge i`ve put up the most optimal gambling trading system that can be used in forex.But first a little disclaimer:


RISK DISCLAIMER: BY USING THIS SYSTEM YOU AGREE TO NOT HOLD ME RESPONSIBLE FOR ANY FINANCIAL OR OTHER TYPE OF LOSSES OCCURRING BY USING THIS SYSTEM, NOR FOR ANY OTHER INFORMATION NOR FILE WHICH IS PRESENT IN THIS THREAD. IT IS A GAMBLING SYSTEM, SO IN THE LONG TERM THIS WILL LOSE YOU MONEY, BUT IN THE SHORT TERM IT CAN BE PROFITABLE.IT IS A VERY VERY HIGH RISK SYSTEM,IT COULD WIPE OUT YOUR ENTIRE ACCOUNT, SO MAKE SURE YOU UNDERSTAND THE RISKS OF IT, AND DON'T RISK MORE MONEY THAN YOU CAN AFFORD.




1)BASIC STATISTICS UPON WHICH I BUILT THE SYSTEM (i know its boring but its useful to read it):


Using all my statistical knowledge, i will show you how to maximize your gambling chances with this system.Most of you know about me that i dont like using indicators of any sort, but in this thread we will make an exception.The Bollinger Band and a Simple Moving average will guide us.Most of you dont know how to use the BB correctly, so i will show you how to use them correctly.


First of all, the Bollinger Band is not a range bouncing indicator or any of that sort, by using it as a range bound indicator to buy at low band sell at high band, is the biggest mistake one can make.The BB is a statistical tool, and measures the standard deviation of the distribution of the candlesticks.


For those of you who dont know anything about statistics, you can learn quick the basics here:
http://en.wikipedia.org/wiki/Standard_deviation
http://en.wikipedia.org/wiki/Probability_distribution


The system will use the "3 sigma rule" but we will use 7 sigma to ensure our odds of losing to be very very small.If the price is normally distributed, then we'd have a 99.999999999744% winrate.But PipMeUp showed that the price follows somewhat a Cauchy distribution which has a fat tail, so if we use even a 7 sigma distance for the STOPLOSS, our winrate will be less than that because the price has a bigger chance to deviate away from the median of a certain period, we use 5 days, but the winrate will be still very high.


http://www.forexfactory.com/showthre...58#post6646358





Here is a little spreadsheet to calculate it:


But the difference is that a Cauchy Distribution has an infinite range, whereas the forex market cannot have an infinite range because of liquidity, so in any way the price must reverse at some point (less chance during Quantitavie Easing when the Central Banks pump liquidity).But still have a fat tail, so we must use a very high deviation to counter that.


So because of this its hard to calculate exactly the winrate but its still higher than 95% presented in my system.I will rely on my backtest data to estimate the winrate.


==============================================================================================


2) PHILOSOPHY OF THE SYSTEM


The system will lose in the long term, and 1 loss can wipe out a significant portion of the account (OR CAN WIPE IT OUT COMPLETELY) , but the system has a so high winrate that it is very unlikely,nearly impossible that a loss can occur.So the longer you use it, the most likely the loss will occur (of course the outcome of each trade is statistically independent but still, the longer you try eventually you will take a loss at some point).


Because of this, since we have a very very high probability of succes, we must take advantage of it, be incrementing the LOT size acordingly to it.


So you cant use this system indefinitely, have to stop at some point or risk losing everything, but hopefully before a loss will come to wipe everything, we can already make much money.Its also good to withdraw cash regularly after all trades are closed, but dont close a trade prematurely.


We could use the lot formula which would increment the lot size accordingly to risk a certain % of the account balance.But why bother? Instead to maximize the efficiency of our advantage of the winrate, we will always increment the LOT size, but make sure if the SL will be very wide to not stop incrementing it. You dont want to get a margin call, and also use smaller leverage for this.


So the risk is very high, but also the winrate, and the potential reward too, so make sure you can afford the risk.1 loss could wipe out the entire account, but the chances of it is still remotely small.
==============================================================================================


3) THE SYSTEM


You will only need a Bollinger Band and a Simple Moving Average, the system can be automated, but it can be used manually too.


Since our system profits from the distribution of candlesticks, and since the distribution of the price has a small chance of occurring in the tail of its distribution graph, we will use a very high period BB.


Pair: Smallest spread pair ,the smaller the spread the better the chances, so EUR/USD
Timeframe: H1 (1 HOUR CHART)
Bollinger Band: Use it on H1 (1 hour chart) and set it to period 120 = 24*5 , so it will look at the distibutions of the past 120 hours, a.k.a last 5 days of trading.Since the price sentiment usually changes weekly, the BB will learn the distribution of sticks within a week.Set the deviation to 7.
Moving Average: Timeframe H1, period 120, price close, and simple moving average


The middle line of the BB is exactly the SMA if you wondered, but it will give you a better visualization if you set the MA to another color


BUY ENTRY: If the price closed or simply moved above the SMA, and previously it was below or equal to the SMA, you enter long
SELL ENTRY: If the price closed or simply moved below the SMA, and previously it was above or equal to the SMA, you enter short


You dont have to wait for the price to close, since a close of a stick on H1 is 1 hour, and you might miss your 4 pip trade chance, the quicker you enter after the price penetrated above/below the SMA, the better the chances.


TAKEPROFIT: STOPLEVEL + 1 or 2 pipette , the smaller the TP the better the chances, so the smallest TP is your STOPLEVEL, my broker gives me a 40 pipette STOPLEVEL on the E/U + 1 or 2 pipettes depending on the volatility is 42 pipette on 5 digit broker or 4.2 pips ~ 4 pips on 4 digit broker.If your STOPLEVEL is bigger or smaller on the E/U, you have to calculate it.


BUY STOPLOSS: Exactly at the lower band of the Bollinger Band, so the chance of losing is 7 sigma small
SELL STOPLOSS: Exactly at the upper band of the Bollinger Band, so the chance of losing is 7 sigma small


You might think the RR is too small, but its mathematically proven that the chances of hitting your stop is very small, since we use a 7 sigma deviation,its very very small.Besides way its better than a NO SL system, and than a fixed SL system.You SL is dynamically set according to the distribution of candlesticks in your chart.


LOT SIZE: Since there is no point in adjusting the LOT to the % of risk of capital since our SL is small, we want to make the most cash possible until a loss comes and wipes our account.So the system uses 1% of the account balance (just to not get a margin call, but you can use more if you have more capital & small leverage).


The formula to convert that into LOT is: LOT = ACCOUNTBALANCEINCASH * RISKPERCENT/10000
So for 100$ the LOT = 100* 1/10000 = 0.01 in MT4 lot units or 1000 LOT UNITS (1 MICROLOT)
And adjust the lot size as your account increases.


ALSO MAKE SURE: STOPLOSSSIZE * TICKSIZE * LOTUNITS < FREEMARGIN , to not get a margin call , use smaller lot size if necessary!
FREEMARGIN = (ACCOUNTBALANCE - MARGIN REQUIREMENT)


NOTICES: Make sure your leverage is small not more than preferably 1:50 or 1:25 not more, so dont get a margin call when your SL will be very wide, and if you dont have enough cash in the account use smaller than 10% of your balance.Try to not enter on news, make sure your slippage is small, and the spread too.Also never move your SL or TP, or you compromise the system, so your trades could be on for days, so if you might get swap bonuses overnight or pay the swap depending on the currency, but with the EUR/USD you will lose a bit money there because of negative swaps , but not much.


FACTORS WHICH CAN IMPROVE THE PROBABILITY OF SUCCES:

1. the smaller the Takeprofit the better, but a STOPLEVEL + a few pipette sized TP is good
2. the smaller the spread & slippage the better
3. the less volatility in the pair the better
4. the higher liquidity the better (assuming that the volatility is low)
5. the higher the deviation the better (7 sigma deviation is enough)
6. after the price moves below/above the simple moving average , the quicker you put in the BUY/SELL order, the better chance of winning it, dont let the price move too far away from the SMA because the odds of failure increase by every pip away from the SMA.
7. the longer the period of the BB the better, we use 5 days price distribution (24*5 hours =120 period) which is fairly enough, because the sentiment changes somewhat every week, so the BB will contain the entire weekly distribution of the price, so the price deviating away for than 7 sigma from the weekly median is very very small, so perfect option to put the STOPLOSS outside it.
8. stay away from all news, and high volatility events, but dont close the trades manually once they are on, and never move the TP or the SL

LATEST VERSION 1.5 CAN BE SEEN HERE (post #330):
http://www.forexfactory.com/showthre...48#post7248548

EXPERT ADVISOR FOR V1.5 (post #352):
http://www.forexfactory.com/showthre...53#post7256253

UPDATE: VERSION 1.3

1. Did a few backtests and optimizations, and i had to update the system
2. DEVIATION 7 is too much and unnecessary, if the data sample is increased, then DEVIATION 6 is enough
3. By using deviation 6 risk is reduced by atleast 15% even if we lose we lose less , and less chance for margin call
4. We will use now period 480 on 1 H chart, so we will check 20 days price fluctuation now
5. LOT SIZE= 2% of the account balance to not get margin called

Pair: Smallest spread pair ,the smaller the spread the better the chances, so EUR/USD
Timeframe: H1 (1 HOUR CHART)
Bollinger Band: Use it on H1 (1 hour chart) and set it to period 480 = 24*20 , so it will look at the distibutions of the past 480 hours, a.k.a last 20 days of trading.Set the deviation to 6.
Moving Average: Timeframe H1, period 480, price close, and simple moving average
BUY ENTRY: If the price closed or simply moved above the SMA, and previously it was below or equal to the SMA, you enter long
SELL ENTRY: If the price closed or simply moved below the SMA, and previously it was above or equal to the SMA, you enter short
TAKEPROFIT: STOPLEVEL + 1 or 2 pipette , the smaller the TP the better the chances, so the smallest TP is your STOPLEVEL, my broker gives me a 40 pipette STOPLEVEL on the E/U + 1 or 2 pipettes depending on the volatility is 42 pipette on 5 digit broker or 4.2 pips ~ 4 pips on 4 digit broker.If your STOPLEVEL is bigger or smaller on the E/U, you have to calculate it.
BUY STOPLOSS: Exactly at the lower band of the Bollinger Band, so the chance of losing is 4 sigma small
SELL STOPLOSS: Exactly at the upper band of the Bollinger Band, so the chance of losing is 4 sigma small
LOT SIZE: we will use now 2% of the account balance

LOT SIZE (in MT4) = ACCOUNTBALANCEINCASH * RISKPERCENT/10000

Only 1 order open at a time, dont move SL or TP, and dont close orders manually, wait for rebounce to your TP.

So your going for MIN TP then, as small as possible there reducing your exposure, but increasing the amount of trades you have to make and therefore exposing yourself to a run against you.

Toying with the idea of doing similar, based on trend following but with a 50 - 100tp, no SL, just a $50 micro account, something to feed my gambling while I trade my proper account proper like!!



Mainly something to deter the newbie's away from no SL trading of this sort, which is becoming annoyingly the next best ever grail around here.



Yes I will and can afford to lose the $50 it's sacrificial and there will be a TE


Basically I'll go the opposite way to you and then there is examples of both failing, although only 1 trade at a time the same!


I'll trade GA or EJ and I'll use the best edge I can think of, so therefore it's a semi serious attempt for a more real test.


p.s. explain Sigma to none scientists, not many will get that!

BACKTESTS:


Sorry for many editings and new fixes, but i had to make sure i fix every parameter.I`m happy to announce, that the V1.3 of the system can run 1 year without any loss.So its 100% winrate for year 2013-2014.

Here are some backtests:

1) Average daily spread set to 1 pip =10 pipette on 5 digit broker, initial capital 100 € , lot size 2% of the capital, TAKEPROFIT 41 pipette or 4 pip, year 2013-2014, EUR/USD , ultra-high quality tick based historical data (over 10 GB)



RETURN: 47.22% /year , no SL hit, yet there were a few minor losses when the TP closed at a negative profit because of gaps & slippage


2) Average daily spread set to 1 pip =10 pipette on 5 digit broker, initial capital 100 € , lot size 2% of the capital, TAKEPROFIT 41 pipette or 4 pip, year 2012-2013, EUR/USD , ultra-high quality tick based historical data (over 10 GB)



RETURN: -87.76% /year ,SL was hit 2 times,

Not at all, every outcome of any trade is statistically independent from eachother.The same % of win or lose for all of them, so its more like, more trades more profit and same chance of profit.

Putting NO SL is stupid, we already know that, but putting a very wide SL without any justification is also stupid, the BB is a great statistical tool, probably the only indicator i use, and helps you to analyse the distribution of sticks.

And because of aligning our SL with the BB indicator, we can vary the RR smart and the most optimal way to get the least losses possible.Because 6 sigma is a very unlikely event so its wise to put the SL there, because it will reduce the odds of a loser trades significantly.

I`m not stating that its a no loss system, such doesnt exits, but hopefully before a loss comes you already make a lot of money, and you can stop then.

The GBP/AUD and the EUR/JPY has both a high volatility , high daily range, and higher spread than the EUR/USD, so its not the best pair, with this strategy its not good to use, you will greatly lower your chances if you will trade on those pairs.

Just use EUR/USD and its fine.

The point is to show failure, while trying not to fail, like all the newbies will be doing, you can show failure via this method and I'll pretend to be smarter and show failure via another method, while being sensible.

I'll go no SL and ensure 700pips before MC, So I'll need a $75 account I guess!! 10cents per pip! $5 margin requirement I suspect.

Yep, the system just failed in 2012, but still survived 1 year, not bad for start. I didnt said its a no lose holy grail system, but it has still very high odds of succes.

I will reconfigure the parameters in v1.4, soon, and hopefully the system now will last more than 1 year

So i`ll keep u updated guys!

I`m also open for suggestion guys, so if anyone has any ideas how to improve the system, please share it with us

Yadix allows you put a take profit (and stop loss) as near to the current price as you like. If that helps?

I think the TP is good as it is, further decreasing it would greatly diminish our gains, and not improve our probability of winning by that much.

I think by optimizing the bollinger settings, and adjusting the SL according to that helps us to imporove the winrate by more than adjusting the TP.

Besides the last thing i want to do is to create a "1 pipper", because the slippage and spread will destroy that. 41 pipettes or 4 pips is a nice TAKEPROFIT for me atleast.

I thought you would probably think that but didn't know if you were aware of the option. Good luck with the system.

Thanks, i`m working on it for 4 hours continuously, expect a huge update which will decrease the chance of loss to almost 0, and triple the yearly return

And also now it works on 2012-2013 too, here is the 2012 return:

Can you share your EA, I could help you on the backtest among other markets with quality data.

Not yet, i`ve used my old EA and modified it to test this system, but the functions are not yet good for demo or forward test, but i will release that probably but not yet.

Until then here is a little more teaser of the 2013-2014 year backtest, more trades, more profit, higher winrate (99.999...%):



from 47.22% return/year to 178.31% , i excluded the month December to let the price reverse in our direction, and not have trades near the new year.

I like this kind of trading analysis, what settings did you change from post #2?

Reduced the period of the BB because the autoregression of the price is not that long and strong in short periods of time so why bother.Instead increased the deviation to a high level, but also experimenting with something very new.

Sorry for the interruption but whts the role of bb in taking buy/sell decision.

The BB is only used for SL at upper/lower bands.

Subscribed. Look really really interesting!

Look like another holy grail.

Ok, so just thinking out loud here but isn't this the type of thing you could use to increase your account massively or blow you out completely? If the strike rate is high enough then surely you could use it to double your account in a short space of time then just withdraw your profits and do it again? If it blows up then so what? As long as you double your account more often than you blow it then you're banking money. Just an idea.