Win 15-18 pips 80% of the time los 34 pips 20% of the time

[Advocate]

Hi -
I did a study of 5 currency pairs over 4 months Feb through May 2003. It was a volitility/ range of pip price movement for each of the following time periods:1, 2, 3, 4, 5, 6, 7 and 8 hours. I learned a number of things such as:

It showed that each currency pair has its own signature in terms of volitility.

Probability of expected pip price movement over diferent time frames.

Here are some abreviated results on the EUR/USD

What is shown is the percentage of time an expected pip movement was achieved during the time frame. For example - during a one hour time frame I should expect the pip price movement to achieve 15 pips or more 80% of the time and I should only see a pip movement of 50 pips or more 2.8 % of the time.

1Hr TimeFrame|2 Hr TimeFrame | 3 Hr Time Frame
15 pips - 80% | 15 pips - 95.5% | 15 pips 99.3%
20 pips - 58% | 20 pips - 86.0% | 20 pips - 98%
23 pips - 45% | 23 pips - 77.0% | 23 pips - 93.4%
25 pips - 39% | 25 pips - 73.0% | 25 pips - 90%
28 pips - 30% | 28 pips - 66.3% | 28 pips 84.2%
32 pips - 22% | 32 pips - 52.8% | 32 pips - 74.3%
40 pips - 9.5%| 40 pips - 33.1% | 40 pips - 57%
50 pips - 2.8%| 50 pips - 12.4% | 50 pips - 32.2%
------------------------------------| 61 pips - 15.1%
------------------------------------| 71 pips - 7.2%
------------------------------------| 99 pips - 1.3%

? What to do with this information?
How about this:
Set your win loss ratio not at 1.5: 1 or 2: 1 or 3: 1 but instead

For a 1 hour trade styal
Set it at - Win 15 pips : to loose 35 pips - 15 pips 80% chance and 35 pips 20 % chance (win/loss ratio = 0.43:1)

For a 2 hour styal of trading
Set it at - Win 20 pips : to loose 50 pips - 20 pips 86% chance and 50 pips 12.4% chance (win/loss ratio = 0.4:1)

For a 3 hour styal of trading
Set it at - Win 25 pips : to loose 66 pips - 25 pips 90.1 chance and 66 pips 10% chance. (win/loss ratio = 0.38:1)

This makes me think of turning the traditional Win: Loss ratio upside down on its head.
What the heck if your average time in a trade is say 2 hours and you could win (86 out of 100 trades for 20 pips = 1720 pips) - (and loose 12.4 out of 100 trades for 50 pips = 620 pips) = 1100 pips.
Only one catch - which direction to pick. Hech trail your stops and the big winners will propel the numbers ahead of the times you went the wrong direction.

Anyway the study is interesting but I was so stuck in traditonal thinking about win/Loss ratios I didnt see these relationships in my study untill last week.

Is there any chance any of you have any experience with this in real trading practice? Any thoughts. If so could you share your thoughts.
Thanks.

[Eriksson]

This goes directly to the much debated win-loss ratio..

Some say it is to little to have 50% wins.. but if you use sl=20 and tp =70 ..50% win ratio is really great! Beacuse you actually are doing a LOT better than coin flip.

The underlying thing here is the following...

You are watching the cable, and you place a random limit 80 pips up (or down). Coin flip, and you have 50% chance of hitting the limit. This is simply a fact!

However, placing a 20 pip stopper in the other direction changes the picture. Flip the coin, and you will hit the stopper 2-4 times more often than the 80pip limit. That is hit the stopper before you hit the limit.

Perhaps some less confusing terms should be used, like gain per trade ? In the end, thats what counts.

However... I have done a lot of backtesting on this subject, and the marked has taken care of this. I have a personal theory about this.

The likelyhood of hitting a given target is ~ inversly proportional to the distance... therefore using 30 limit, and 90 stopper will end up even. 3:1 ratio . This is IMHO valid within the typical trading range for one day.

Still I may be wrong

Best,
Eriksson

[mishak]

Hi Eriksson

I can agree with you only partially
Yes, the likelyhood of hitting a given target is inversed to the distance. But there is no linear proportion.

Look at the figures provided above by Beliver.
Compare 20 pips and 40 pips movement expectations for three time frames.
They are not linear inversed.

They begin to resemble the linear reversed proportion only for lenghty time frames ! (3H frame looks better than shorter ones).

It is the interesting problem:
What is random - Forex market prices or Expectations of a Novice
Please take time to contribute to another thread of this Forum here

[Eriksson]

Well, of course I did put [~] mark before my statement
And yes, finding a basic linear relationship there seem remote.. but..

I am not quite sure how this experiment were done, but still I am going to try interpit it

Looking at these numbers that Beliver vas kind enough to share with us, you can see timeframe dependency. And yes there is no linear relationship there, looks a lot more like exponential function, and I belive it is one.

but.. compare the numbers in the timeframes and taka a close look how the function changes in larger timeframe, IMHO the direction points to linear relationship using long (big) enough timeframe..(more realistic)

If you take one hour period, 100 pip movement is of course very rare, but 10 pip quite frequent, and 500 pip movement occurs about never I guess. Time period of one day would be quite different. This has only to to with the growth rate.. i.e rate of change of the cable, like in pips/min! So .. what I am trying to say: the possible rate of change is playing the leading role in short timeframe.


Lets assume there were non linear relationship going on here -exponential for example, it is quite clear someone would have found it, and taken advantage of it, in similar fashion as Beliver mention in his good post. Brokers with close spread would be dropping like flies.

I have tested this over and over using 4 pip spread. And never have I seen a glimpse of hope using closer limit than stopper, the spread has always eaten up the little profit gained (and indeed there was none). There is no hole there. qed- ~ linear relationship.

Still: beliver is right, if you are only goint to trade for one hour, 90 pip stopper will hardly be triggered, but will this fact help at all?

Side note:

If I am right, we can easy define new more realistic equation for relative win rate, to see how good TA works.

Sl=20 pip, target =80 coin flip ratio to beat 2:8 or 1:4 This would indicate anything over 25% win rate, and TA is beating the coin flip test.

33% win rate, "common" for trend following systems is 1:3, and using the above stoppers 0,33/0,24=1.32 hit rate..

Then max hit rate using 20 stop and 80 limit would be 1/0,25=4

Now we have a brand new scale, coin flip equals 1

100% win rate using stop 50 limit 50 equals 1/0.5=2
100% win rate using stop 30 limit 90 ..... = 3
100% win rate using stop 20 limit 80...... = 5


What do you think!?

best,
Eriksson

[mishak]

You are making a good point.
You may be interested to read the discussion on setting risk:reward ratio in 1,2,3 treads of 6 starting here

In respect to your figures above, please let me ask a question:
You say (based on your data) that
the percentage of time a 20pips movement was achieved by Euro during the 3H time frame is 98%.

Does it mean that during these 3 hours Euro did never hit 30 pips in the opposite direction ?
Have you compared start and end prices OR examined the full path of Euro during these 3 hours (this can be done by examining OHLC figures) ?

Appreciate your interesting input

[Advocate]

Eriksson and mishak
Thank you for responding to my posting - It was a nice surprise to wake up and see a response to my 1st posting.
Its 7am and were getting the kids out to school and then I have a big deal dental apointment. I will get back to you with more info on my study after I recover from the dental apointment. I think we operate on different time zones.
Thanks- Beliver

[Urantian]

I have actually been trading this way for 10 days--first let me explain I was very frustrated at getting stopped on 20 pip or less stops while going for 100 or more pips--I woke up one day with the clear idea that if this was a proven loser why not turn it around--flip a coin, use a close limit of 20 pips and a 200 pip stop loss. So far I have won all 6 trades and one is till on at 21 pips out from breakeven. I haven't decided to keep it up however since it ties up capital for so long, and when I do get the loser it will be distressing. I have jokingly thought this fits my temperment of always cutting profits short (never thought of using a trailing stop with it, makes sense) and letting my losers run. Thanks for all the statistical analysis--but I just figured, what are the odds your stop is going to get hit before your limit?? I didn't really anticipate it taking so long, however!!

Another thing is, I figured if I got 3 consecutive losers (highly improbable) to establish my trade size from there so if it DID happen I'd only be down 50% of my account.

Thanks for the oppotunity to get this out--have wanted to ask how this would backtest--but what counts is how it works for me, I guess. Maybe I could sell it--call it the noise capitalization system or the coin toss maximizer!!

All the best

[skater68]

You put together the following stats...and I really like what you are trying to do...it showa you are thinking...

however one question....

when i look at each column the percentages for each row should add up to 100%

your chance of 15 pips is 80% and 20 pips is 58% that is 138%...i am confused......

if you analyzed the 1hour time frame and you noted pips obtained and the number of times they were obtained then you should divide each result by the total...this way your percents add to 100%....

1Hr TimeFrame|2 Hr TimeFrame | 3 Hr Time Frame
15 pips - 80% | 15 pips - 95.5% | 15 pips 99.3%
20 pips - 58% | 20 pips - 86.0% | 20 pips - 98%
23 pips - 45% | 23 pips - 77.0% | 23 pips - 93.4%
25 pips - 39% | 25 pips - 73.0% | 25 pips - 90%
28 pips - 30% | 28 pips - 66.3% | 28 pips 84.2%
32 pips - 22% | 32 pips - 52.8% | 32 pips - 74.3%
40 pips - 9.5%| 40 pips - 33.1% | 40 pips - 57%
50 pips - 2.8%| 50 pips - 12.4% | 50 pips - 32.2%
------------------------------------| 61 pips - 15.1%
------------------------------------| 71 pips - 7.2%
------------------------------------| 99 pips - 1.3%



like i said i like what you are trying to analyze i just think you need to check your math so we can further analyze the results

thanks..and great work

dave