A statistically rare pattern/anomaly - reversed? | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

How to keep in faith with a statistically profitable system? 58 replies

Statistically Infallible 11 replies

Statistically testing every indicator with my Genetic Algoritm 132 replies

Are my 6-mth-old results statistically valid? 17 replies

What Constitutes a Statistically Significant Sample for Back Testing? 4 replies

A statistically rare pattern/anomaly - reversed?

  • #1
  • First Post: Edited 8:09pm Sep 8, 2014 5:23pm | Edited 8:09pm
Hey all,

I was thinking is it easy to produce a statiscally poor strategy (around <2%, <3%, <4%, <5%) and then reverse the concept to have >90% accuracy?

Just a thought....

Like to hear constructive opinions.

MM

EDIT: Maybe I've phrased this thread incorrectly, instead of "poor" the word I should have used is an "anomaly" or "rare" occasion.

Let me give a clearer example of what I actually mean...

Imagine you find a certain behaviour in price and measured it statiscally, and statistics showed it has a low probability to do X. Meaning you have an edge of what price won't do?
  • #2
  • Sep 8, 2014 7:19pm Sep 8, 2014 7:19pm
  •  FerruFx
  • Joined May 2007 | Status: MT4/MT5 EAs/Indicators/Alerts coder | 6,542 Posts | Online Now
Reversing a loser won't make a winner. That was proved 1,000's of times ... Similar threads are opened each month or so since many years ...
MT4/MT5 EAs/Indicators/Alerts coder
  • #3
  • Sep 8, 2014 7:22pm Sep 8, 2014 7:22pm
Quoting FerruFx
Reversing a loser won't make a winner. That was proved 1,000's of times ... Similar threads are opened each month or so since many years ...
If I had a nickel for every time I read about this strategy, I wouldn't need to find a profitable strategy :-)
  • #4
  • Sep 8, 2014 7:25pm Sep 8, 2014 7:25pm
If it were only so easy. With all the failed systems out there, it'd be a gold mine.
Ghost Rider - WWTBMD?
  • #5
  • Sep 8, 2014 7:27pm Sep 8, 2014 7:27pm
Quoting FerruFx
Reversing a loser won't make a winner. That was proved 1,000's of times ... Similar threads are opened each month or so since many years ...
Quoting srt
{quote} If I had a nickel for every time I read about this strategy, I wouldn't need to find a profitable strategy :-)
Ok, neither of you seem to understand. Maybe I'll explain a bit more later.

I'm simply not saying when you want to buy (sell) and vice versa.

I'm talking about statistics/probability.

MM
  • #6
  • Sep 8, 2014 7:37pm Sep 8, 2014 7:37pm
Quoting Madmoney
Hey all, I was thinking is it easy to produce a statiscally poor strategy (around <2%, <3%, <4%, <5%) and then reverse the concept to have >90% accuracy? Just a thought.... Like to hear constructive opinions. MM

At least you are thinking creatively! The key is the historicity with which you get that "<2%, <3%, <4%, <5%." If you get "<2%, <3%, <4%, <5%," across all market phases on a consistent basis, then you have just created a high accuracy mode through the back-door. However, that is typically not the case and the "<2%, <3%, <4%, <5%" that you witness will have periods where its consistency is not high enough to warrant Signal Reversal.

A better method is to develop a signal or system and then conduct MAE/MFE analysis. This tells you the full story about not only whether or not your targets are being struck, but the manner in which they are being struck with respect to all price action between Entry, MAE, MFE and Target. Remember, your actual Target could be above or below the average MFE. With that information, you can begin tweaking entries, exists and target expectations with a higher level of fidelity.

Every Indicator and every Signal in my system goes through MAE/MFE analysis independently. The entire system them undergoes MAE/MFE analysis. I call them Dynamic Performance Profiles and there is a database that contains their real-time output which is algorithmically looped back into the system. This DPP input channel is one of the ways in which the system "learns" about its own behavior, making adjustments on entry, targets and exit levels.

It takes a little more work than just reversing bad signals, but again, at least you are thinking creatively and trying to solve problems. Remember, as a systems designer, you are a problem solver - you get paid to solve problems - not create them.
F-X-C-M | A Classic Retail Bucket Shop Scam Artist Who Manipulates Prices
  • #7
  • Sep 8, 2014 7:42pm Sep 8, 2014 7:42pm
Quoting iDoubleStoch
{quote} At least you are thinking creatively! The key is the historicity with which you get that "<2%, <3%, <4%, <5%." If you get "<2%, <3%, <4%, <5%," across all market phases on a consistent basis, then you have just created a high accuracy mode through the back-door. However, that is typically not the case and the "<2%, <3%, <4%, <5%" that you witness will have periods where its consistency is not high enough to warrant Signal Reversal. A better method is to develop a signal or system and then conduct MAE/MFE analysis. This tells you the full...
Great, thanks.

I would love to know more about your MAE/MFE analysis of a signal/system, could you elaborate more?

Are there efficient ways to measure maximum adverse excursion and maximum favourable excursion?

MM
  • #8
  • Sep 8, 2014 8:00pm Sep 8, 2014 8:00pm
Quoting Madmoney
Hey all, I was thinking is it easy to produce a statiscally poor strategy (around <2%, <3%, <4%, <5%) and then reverse the concept to have >90% accuracy? Just a thought.... Like to hear constructive opinions. MM EDIT: Maybe I've phrased this thread incorrectly, instead of "poor" the word I should have used is an "anomaly" or "rare" occasion.
I did try that a few times but I could not get enough losing trades to come up with a reverse strategy.
  • #9
  • Sep 8, 2014 8:28pm Sep 8, 2014 8:28pm
Quoting Madmoney
{quote} Great, thanks. I would love to know more about your MAE/MFE analysis of a signal/system, could you elaborate more? Are there efficient ways to measure maximum adverse excursion and maximum favourable excursion? MM

Each indicator is a mini signal, producing its own mini trade profile (entry, target and exit levels). So, anything producing a trade profile can itself be evaluated for MAE and MFE. Signals within a trading system are implemented as compositions of one or more Indicators. Thus, each signal can therefore, be evaluated for MAE and MFE.

A google on MAE/MFE will provide sufficient explanation for the simple calculations necessary. What you need is a quick way to conduct the analysis on indicators, signals and trading systems and inside MT4 that is done best be linking code that paints the screen with MAE/MFE levels per trade execution and that writes that data to a .csv file on your HDD. You can then take that data and conduct your own analysis in Excel, of course. This runs as an EA of the trade logic used trigger the entry, target and exit.

There is EA code out there (google it) that you might be able to use as a starting point or guide. I believe this site might have some threads on MFE/MAE Analysis. Often times people place the Max Favorable in front of Max Adverse, so type if both ways. I typed it MAE/MFE because that's how my brain works - I'm always thinking about Risk First, then Profit Potential, while others think Profit first, then Risk.

A very short descriptor but enough to get started: http://trading-journal-spreadsheet.c...s/mfe-mae-bso/
F-X-C-M | A Classic Retail Bucket Shop Scam Artist Who Manipulates Prices
  • #10
  • Last Post: Sep 8, 2014 8:32pm Sep 8, 2014 8:32pm
Quoting iDoubleStoch
{quote} Each indicator is a mini signal, producing its own mini trade profile (entry, target and exit levels). So, anything producing a trade profile can itself be evaluated for MAE and MFE. Signals within a trading system are implemented as compositions of one or more Indicators. Thus, each signal can therefore, be evaluated for MAE and MFE. A google on MAE/MFE will provide sufficient explanation for the simple calculations necessary. What you need is a quick way to conduct the analysis on indicators, signals and trading systems and inside MT4...
Many thanks.

I will look into this.

MM
0 traders viewing now