I start this thread here to perhaps solicit discussions about basic statistics and how it could be integrated in trading systems etc. I will copy both post I have recently placed in another thread which has a topic of its own etc…

Perhaps there are some of you who have done similar experiments etc and wish to share it with the rest of the board.

Best wishes,

FX Sniper

__________________
"I do not need a compass to tell me which way the wind is shining."

Perhaps this will help you see things differently. I will attempt to write this in very laymen’s terms.

I attach a spreadsheet demonstrating basic statistics that if created for other pairs also, could be used to set proper profit targets etc. The spreadsheet I just threw together in a few minutes, it is not complete and only shows 1 statistic for long trades only. This should serve to provide an example of how to look at probabilities in the most basic form.

Why all this now? So you know how to enter a trade for this system, next question, how far to set the take proffit levels, you can suck this from your thumbs or do it the right way using basic statistics. The next question would be, how far to place stops, well, use the example and the message it contains to think a bit further and create those levels for yourself, in this post I outline a example of finding the odds to exit for example 98.7% of the time with 10 pips, at what cost (drawdown) you can find out easily.

The information in the spreadsheet or similar information gathered on your own, can be applied to any system. I use statistics a lot to help me determine exits. I trade the 5 min Euro only and know that statistically speaking, when my trade is correct, price should not retrace back into the previous bar’s range by more than 3 pips from the high of that previous bar. This simple information has helped me to reduce my trade risk to only 3 pips excluding my 1.5 pip spread and 1 pip slippage ( 1 pip dived by 2 = 0.5 pips both ways ). This has helped me to take on trades when the market is extremely volatile without having to increase risk to adapt to the volatility etc etc… bottom line, using common sense and some simple tools, you can optimize profit targets and stops etc. to a fine degree. Why risk 10 point stop loss, if I know that 92.3% of the time, if the price goes against me more than 3 pips, the trade is wrong, you see my point.

With regards to the attached spreadsheet: It contains the daily Euro/USD data from 19 March 1996 (I was to lazy to get more, but I recon 10 years of data might be enough for now). You will notice a MFE column. MFE is the Maximum Favorable Excursion. It measures that if today’s’ price took out yesterday’s high, what is the maximum amount of pips it went above yesterday’s high. In other words, today’s high – yesterday’s high gve you the maximum amount of pips you could of made if you entered on the break. Of course we NEVER know how much it will run and where the exact high for today will be, but that is why we calculate the statistics.

You will see a column called BINS (CELL J2). One uses bins when creating a frequency table. A frequency tables tells one how many times a certain event happens. In our case I have set the bin sizes to 10 pip intervals… The HITS column (CELL K2) tells us how many times the maximum pips we could of made falls into the bin on it’s left.

For example we can see that 226 times the maximum amount that today’s high exceeded yesterday’s high falls between 20 and 30 points. Another example is that 86 times, the maximum amount today’s high exceeded yesterday’s high fell between 70 and 80 points and so on an so forth. This is the value of setting up a frequency table, it gives better clear information about how many times something happens in a certain way…

In total in this example, we had 1345 times where the high broke the previous days high… the %HITS column just converts the HITS column to a %... so as an example, we see that 16.8% of the trades gave an opportunity to take between 20 and 30 pips profit etc etc..

The ODDS column is the fun part. This shows us that for example, if we took trades as the broke the previous day’s high, and set a take profit for say 20 pips, we would of taken profit 85.7% of the time, if we set take profit at 10 pips, we would of taken profit 98.7% of the time.

Ok, so now you know a way to answer your questions using some simple statistics.

But before you think this is the holy grail of all grails with such high accuracy rates, remember, we have just investigated a small part of the picture…sure we can get 10 pips 98.7% of the time perhaps, but before it reaches that 10 pips, what happens in the mean time, how far does it go against us first before it gets there. This is called MAE of maximum adverse excursion. You will notice such a column in the spreadsheet also…but also bear in mind we are looking at the daily data, and from it we can not judge what came first, the 10 pip profit or the adverse excursion, so one needs to drop down to intraday data etc. to do some more checking and so on…all of this is not impossible, and if you want this badly enough, you will put in the effort to achieve the desired results.

Also bear in mind, that statistics in this case is only modeling historic data, we can only make assumptions based on the results and it does not guarantee future consistency.

With regards to systems, we can see by simply trading every breakout of every daily high, we can achieve 98.7% probability of receiving 10 pips profit, without using any of this system’s rules etc… If a system and it’s rules has a purpose, it should be to decrease the MAE, the odds of getting 10 pips is not the problem here, the MAE is !

Best wishes

FX Sniper

Attached Images

STATS.gif (15.7 KB, 848 views)

Attached Files

EURUSD Daily Stats.zip (299.8 KB, 184 views)

__________________
"I do not need a compass to tell me which way the wind is shining."

Ok, you lot, here is something to tease you a bit with.

From the statistics in my prev. post we see that 98.7% of the time, if we set our take profit level at +10 pips, we would get it. That means that 1.3% of the time we would not.

From a total of 1345 trades it would mean that 1327 times ( 1345 trades x 98.7% ) we would receive 10 pips, which gives us a total of 13270 pips.

This also means that 17 times ( 1345 x 1.3% ) we would NOT have gotten our 10 pips.

Now, I have not checked this out, but it is easy to do, so let us make an unchecked assumption for now. Let’s assume that on ALL of the 17 days we did NOT get our 10 pips, we lost 60 pips. This would place our losing pip total at 17x60=1020. The net result is a positive profit of 12250 PIPS !!!

Now Euro’s average daily range is abouts 90 pips, so even if we exaggerate a little and make the 17 losing trades 130 pips each, we still end up with a net profit of 11060 pips ! An we are assuming that the worst case which is that if we did not get +10 we lost -130 which I am sure is not remotely as bad as we are assuming. We are also just talking about Long trades here and have not taken a look at short trades…

Anyhow, now you have enough info to keep you exited about Forex trading and what seems to be boring statistics

Best wishes,

FX Sniper

__________________
"I do not need a compass to tell me which way the wind is shining."

Ok, I had some time on hand before lunch and added some missing parts.

If anyone is interested, just indicate so and I will post the adjusted spreadsheet. In the mean time I just attach a screenshot of the figures.

From the previous posts I suggested taking a long trade on each and every break of the prev. day’s high. I showed that history suggests that if I placed a take profit target of 10 pips, I would get it 98.7% of the time. I also noted that I temporarily made an assumption that if I used a stop loss level of 130 points for the loosing 17 trades out of the total of 1345 possible trades, there would still be more than enough pip profit left.

I have now completed the stats for short trades also as well as created the frequency tables to replace my assumption with something a bit more concrete than just a guess.

In the screenshot the right hand tables are the MAE’s. From their ODDS columns it is clear that setting a stop loss of 130 points there would be only a 9.4% chance of it being. You can follow the tables to find other combinations etc.

I think that if a 130 pip stop is hit, the chances of it ending up positive if the take profit was not reached before then would be very close to 0.

Also note how things equal out between the total amount of high breeches and the amount of low breaches, only a difference of 3 trades !!! Also a high correlation between short and long trades, basically the price action with regards to high and low breaches are well balanced.

Please just a reminder, this is not a suggested system or rule set etc. I am merely demonstrating to those interested, how easy it is to answer some questions that sometimes sound impossible to answer, or at the very least have a more informed answer.

Best wishes,

FX Sniper

Attached Images

STATS2.gif (23.7 KB, 850 views)

__________________
"I do not need a compass to tell me which way the wind is shining."

nice spreadsheet the only thing that threw me was the formatting on the cells like 0.0150 is 0.015 or 0.0100 is 0.01. if anyone else wants all the cells to read the format of 0.0000, select the cells you want to change then go to "format > cells > format > type:" and type in 0.0000 in the example box it should show up as 0.0030 or whatever box you have selected.

something i noticed is under the "frequency statistics" "hits" the whole colum shows up as the numbers, as in your diagram, but the input box at the top (normally displays the formulas) shows "=Non-supported formula!" it works though so cant complain

Yeah those damn extra zeros are annoying when they are not aligning nicely, anyhow, fixed all that plus added another sheet displaying simmilar data for GBP/USD also...here is the updated spreadsheet...

Laters,

FX Sniper

PS: the FREQUENCY formula is entered as an ARRAY FORMULA, so you can not adjust them one by one, you need to adjust the whole array.

Attached Files

Daily Stats.zip (814.6 KB, 169 views)

__________________
"I do not need a compass to tell me which way the wind is shining."

Thanks fxsniper for the data, as always your input is appreciated.

Just for fun I checked the last thirty trading days on my SBFX demo and unfortunately five days a break of the previous high failed to reach 10 pips and likewise four days on the short side.

I like a simple strategy. I realise that the 130 pip stop loss was just a rough figure but with my rough maths that would need a 93% win ratio to break even. Is that about right?

im having problems loading the latest version, but im not using word so that may be it. see how everyone else gets on with it.

for a break even stop would it be calculated 13270pips / 17 lost trades = 780stop = breakeven over all.

__________________
There is no candle.

Dear, FX Sniper,

This Study is quite interesting and your apporoach of looking at the thing with different view is good. many times things are simple...But we need vision to look at it.

Keep it up...we are along with you and sharing our views on this subject...

Happy Trading
Ramdas

Hi

Your observations underline a very important factor that should be considered when one does these kinds of things - DATA QUALITY !!!

The data I used came from Alpari, although I get slightly differenct results with 2 different other brokers...on some it might reach 9 pips and turn back, on another it might go the 10 etc... This is and has always been my biggest problem concerning using stats for forex.

Pretend we did have a system in the lines of my discriptions, the error margin is very narrow and thus leaves extremely little room for data inconsistancies. Antoher thing to considder in my case at least, is that I use Oanda as broker, their prices also do not match up exactly to Alpari for example which complicates things a little more

If that was not the case, I would be a damn gazillionare already hehehehe, I tried contacting Oanda some time ago and asking them for some historic data for me to play with in Excel, but they did not want to give it to me I think that the best way to go, would be to do all the stat work based on the broker one uses, expecially in extrmeme cases like the one we are discussing where there is very little room for error.

On the other side though, this does not dismiss the value of statistics by no means, my other stats that I am actually applying to my real trading is based on 5min Alapari data where I have calculated a MAE for my trades to adjust my stops, the margin for data inconsistancies is a little larger there and so it seems to be working just peachy !!!

Your esitmate about the accuracy ration is quite close yes I think, remember we worked on a hit rate of 98.7%. If you do want to play a little with this, try changing the bin size increments to say 8 pips each, who knows, we might catch more 8 pips or 9 pips which could increase our error margin for data inconsistancies... just a though. I love playing with this stuff, keeps the mind sharp and objective I always tease myself hehehehehe

Take care,

FX Sniper

__________________
"I do not need a compass to tell me which way the wind is shining."