Expectancy Management | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

What is Expectancy? 13 replies

System Expectancy - Monte Carlo Simulation 5 replies

Expectancy question for experienced traders 2 replies

Random Market vs Set Expectancy 8 replies

Expectancy is always zero 41 replies

Attachments: Expectancy Management
Exit Attachments

Expectancy Management

  • #1
  • First Post: Feb 23, 2013 11:50am Feb 23, 2013 11:50am
Hello everybody!

Lately I searched FF for "risk management". It returned 99 pages of results. I read one by one the 250 first posts. 248 of them were instead about money management. I found only two of them were really about risk management. One is made of a single post of never answered questions. The second is one page long and the last post is about position sizing; understand money managment.

Either risk management is so trivial it goes without saying or it's the most overlooked element of trading in FF.

I'm sure that risk management is much more than just setting a SL the other side of the previous swing. Because risk management is mistaken with money management I chose another title to avoid any confusion. I wrote a PDF where I examine, on a purely probabilistic point of view, different trade managements to see how they perform in general. Don't hesitate to verify the equations! After a dozen proof readings I may have missed something...

I think this is only scratching the surface. I wish this will start an educated discussion. Unless it dies before the second page...
Attached File(s)
File Type: pdf Expectancy management.pdf   948 KB | 3,205 downloads
No greed. No fear. Just maths.
  • #2
  • Feb 23, 2013 11:50am Feb 23, 2013 11:50am
OANDA and FXCM have released stats about their retail traders. 70-75% of them are losing money. That's better than the 95% urban legend but still...
FXCM made stats on the positions taken by their losing traders. Surprisingly it appeared that more than 60% of the positions were taken in the correct direction! So when I read that you just need beeing right 51% of the time to be a winner...
(just copy paste "right 51% of the time" in FF search box to see the number of results returned!)
No greed. No fear. Just maths.
  • #3
  • Feb 23, 2013 1:13pm Feb 23, 2013 1:13pm
Mr Chris Capre has some interesting statistics on risk models
tradewith60
  • #4
  • Feb 23, 2013 6:23pm Feb 23, 2013 6:23pm
Quoting PipMeUp
OANDA and FXCM have released stats about their retail traders. 70-75% of them are losing money. That's better than the 95% urban legend but still...
FXCM made stats on the positions taken by their losing traders. Surprisingly it appeared that more than 60% of the positions were taken in the correct direction! So when I read that you just need beeing right 51% of the time to be a winner...
(just copy paste "right 51% of the time" in FF search box to see the number of results returned!)
Do you have a link where it is possible to see these stats?
My searching skills may not be up to much, but I have had no luck locating these reports.
They may be interesting to read.

Thanks
Please Do Not PM Me With Coding Enquiries
  • #5
  • Feb 23, 2013 10:14pm Feb 23, 2013 10:14pm
Nice document Pip, you make some very valuable points and I am glad some of your conclusions is what they are.

Will gladly try and help to continue this discussion.
  • #6
  • Feb 23, 2013 11:10pm Feb 23, 2013 11:10pm
Quoting PipMeUp
... 70-75% of them are losing money. ...
FXCM made stats on the positions taken by their losing traders. Surprisingly it appeared that more than 60% of the positions were taken in the correct direction!
The real problem for the losers, isn't to be right or wrong at the entry ... it is how they will manage the opened position.
MT4/MT5 EAs/Indicators/Alerts coder
  • #7
  • Feb 24, 2013 1:33am Feb 24, 2013 1:33am
Quoting Gumrai
Do you have a link where it is possible to see these stats?
My searching skills may not be up to much, but I have had no luck locating these reports.
They may be interesting to read.
Thanks
Those are links to stats from 2010 for IB and FXCM. I saw more recent but didn't bookmarked. The figures don't change too much.
FXCM is doing this kind of stats quite often. I think Jason Rogers was involved in the stats but not sure. You might PM him...
http://trading-gurus.com/fxcm-releas...ty-statistics/

The same for IB. I had found a table with more brokers. The results are quite similar.
http://trading-gurus.com/interbank-f...rs-lose-money/

Here about the direction being more often correct than not
http://www.dailyfx.com/forex/educati...ose_Money.html
No greed. No fear. Just maths.
  • #8
  • Feb 24, 2013 2:16am Feb 24, 2013 2:16am
Quoting 60minuteman
Mr Chris Capre has some interesting statistics on risk models
Thanks for the pointer. I think you are refering to this:
http://2ndskiesforex.com/strategies-...u-should-know/
http://2ndskiesforex.com/strategies-...-risk-of-ruin/

IHMO, and that's why I opened the thread, it's not only about MM or risk of ruin. Under-capitalization is part of the risk but the pair you trade as well. For instance I find E/U is very speculative and perhaps much more dangerous than USD/SEK no one wants to trade. It over-reacts to the news. Traders are lured by its smallest spread. It makes this pair the most traded thus perhaps the most efficient. The TF you trade is part of the risk too, see http://www.forexfactory.com/showthread.php?t=336715. I'm sure I'm missing others.

Ferru is right the management of the position is as important as timing the entry. I'm very clumsy at this.
No greed. No fear. Just maths.
  • #9
  • Feb 24, 2013 3:22am Feb 24, 2013 3:22am
Quoting PipMeUp
Those are links to stats from 2010 for IB and FXCM. I saw more recent but didn't bookmarked. The figures don't change too much.
FXCM is doing this kind of stats quite often. I think Jason Rogers was involved in the stats but not sure. You might PM him...
http://trading-gurus.com/fxcm-releas...ty-statistics/

The same for IB. I had found a table with more brokers. The results are quite similar.
http://trading-gurus.com/interbank-f...rs-lose-money/

Here about the direction being...
Thanks PipMeUP,
Although those statistics are a bit vague, they do seem to support FerruFx's post ( just before yours).
What do we do when we back test a strategy? Well often we will see that we can earn 100 pips with just a 20 pip SL. Yes, often, but not that often. What we see is that the optimal return may be 50 pips for our 20 pip risk and this happens time and time again.
So it may be that we Have an EXPECTANCY that we will profit 40% of the time from trades risking 20 pips with a TP of 50.
That is actually very good. Out of 10 trades, you will have 4 winners that give you 200 pips, the 6 losers out of the 10 trades will be minus 120 pips. So a net profit of 80 pips.
So you are confident, you start to trade with a live account. You have a great strategy. You only need 40% winners to give you a really good return. You start to trade, but in the back of your mind is the much repeated mantra "Nobody went broke by taking a profit"
So what do you do? When price moves 25 pips towards your profit target, you move your stop to BE +5 even though this was not a part of the strategy when you tested it.
Price moves back and hits the stop, you are relieved that at least you took a profit from the trade. 5 pips are better than nothing, after all?
Although that may be true, that is not the strategy that you intended to trade. The EXPECTANCY was 50 pips from 40% of trades.
You are now trading a totally different way than when testing the strategy.
I wish that I could say that I don't make these mistakes, I do.
Please Do Not PM Me With Coding Enquiries
  • #10
  • Feb 24, 2013 8:01am Feb 24, 2013 8:01am
Quoting PipMeUp
Hello everybody!

Lately I searched FF for "risk management". It returned 99 pages of results. I read one by one the 250 first posts. 248 of them were instead about money management. I found only two of them were really about risk management. One is made of a single post of never answered questions. The second is one page long and the last post is about position sizing; understand money managment.

Either risk management is so trivial it goes without saying or it's the most overlooked element of trading in FF.

I'm sure that risk...
Since you seem to be rather good at maths, I think you are looking for information in the wrong crowd. There is so much written on risk management by banks and universities who really try to base their results on some solid ground. I think most people here won't be able to understand what you are presenting. By the way, the big 4 also have extensive writings on risk management, not particularly for trading, but risk management as a whole for companies. Of course many things are also applicable to trading.

And let's be honest, most here are probably still in their puberty or in their twenties, they won't be able to contribute anything to your search. The CFA institute is also sometimes bringing forward some rather interesting articles on risk management.

cheers
  • #11
  • Feb 24, 2013 9:40am Feb 24, 2013 9:40am
Quoting Custos
Since you seem to be rather good at maths, I think you are looking for information in the wrong crowd. There is so much written on risk management by banks and universities who really try to base their results on some solid ground. I think most people here won't be able to understand what you are presenting. By the way, the big 4 also have extensive writings on risk management, not particularly for trading, but risk management as a whole for companies. Of course many things are also applicable to trading.

And let's be honest, most here are probably...
anything intellectual is beyond 'most here'... or most anywhere. it's not for them. that's their problem, not those who get it. that's life. :nerd:
Congratulations ScalaFX for coming out of the closet!
  • #12
  • Feb 24, 2013 12:34pm Feb 24, 2013 12:34pm
Quoting PipMeUp
Thanks for the pointer. I think you are refering to this:
http://2ndskiesforex.com/strategies-...u-should-know/
http://2ndskiesforex.com/strategies-...-risk-of-ruin/

IHMO, and that's why I opened the thread, it's not only about MM or risk of ruin. Under-capitalization is part of the risk but the pair you trade as well. For instance I find E/U is very speculative and perhaps much more dangerous than USD/SEK no...
As far as I can see everything moves in a structured way that can be followed, I have personally known one person who can read it very well and almost never lose, with stops!... but you can only ever tell what the next most probable move is, after you pull the trigger you need to then re-calculate that as more information (price) becomes available.

Now if you know this, and you can trade it (because i have seen many taught it who just never get it!) then your expectancy will be way different to some of the ego hogs on this forum who think the ability to turn a regular small profit somehow makes them "intelligent" or "gifted" ...

I think it's always interesting to find out peoples motives behind the answers they seek on these forums... for instance there's a thread recently about OP's of strategy threads not posting their results, i'm pretty sure the poster was failing to be honest with himself about his own results.. What are your motives?

The problem with the ideas on the pdf is that it presumes every entry is in the same part of the cycle, but an entry at a D1 top will have a very different expectancy than an entry at H4 top, even though they adhere to the same rules...
tradewith60
  • #13
  • Edited 7:40pm Feb 24, 2013 6:59pm | Edited 7:40pm
Hey PipMeUp, nice thread!

I've made it through the first section in the PDF regarding taking partial profits. At the end of page 4 you present a series of equations where Delta = EPartial - E0, and make the assertion that if Delta is positive, taking partial profits improves the system, but if Delta is not positive then the partial profit exit does nothing to improve system profitability.

The equation is simplified into:
Attached Image


where R is the final TP amount, p is the final TP win % and pb is the SL %.

While I agree with the commentary that effectively states that R should be maximized to let profits run, and p should also be as near to 1 as possible, we can't always pick our system outcomes. Some systems have a high reward but a low hit rate. And for some systems of this type, improvement may be found by including a partial profit exit.

For instance, a system with a win rate, p of 10%, Reward:1 R of 15 becomes:

((1-15)*0.1)/2 + 0.9 = 0.2

Thus it is possible to show small improvement in some select cases where R is large, and p is very small by adding a partial profit exit. However, in most cases as you point out, the result is worse by adding a partial profit exit. I had no idea that a closed form solution could be made for take profits and stops. Thanks for posting this pdf - brilliant!

--------------------------------------------
Edit:
Actually it appears that in this case, (with the assumed partial profit exit taking off 1/2 at 1) it is systems with an R less than two that show positive delta such as:

p = 0.5, R = 1.5

((1-1.5)*0.5)/2 + 0.5 = 0.375

I've never been much in favor of taking partial profits but unless my calculations (or interpretation) are off (very possible), maybe in some cases it does improve profitability?
  • #14
  • Feb 25, 2013 4:14am Feb 25, 2013 4:14am
Quoting FXEZ
...where R is the final TP amount, p is the final TP win % and pb is the SL %....
R is the TP when SL is normalized.
p is the probability of winning.
Pb is NOT the probability of losing but the probability of hitting the stop after the partial profit. The proba of losing is Pa+Pb, the complement of p.

So if p=0.1, Pb won't be 0.9 but less. Otherwise Pa would be 0 and closing at PP would be Holy Grail ;-)

But you are right, you can find cases where delta is indeed positive in the different equations. But quite often if you substitute p and R in the equation for E you get a negative expectancy: you improve a losing system and usually to get a slighly less losing one. Also chosing p and R doesn't give you the associated system.

In this PDF I was mainly trying to understand why do people use these managements. They are even advised in trading books. Yet they are suboptimal. I expected that with the time and the number of traders they should have converged toward an optimal solution.
No greed. No fear. Just maths.
  • #15
  • Feb 25, 2013 4:43am Feb 25, 2013 4:43am
You are forgetting about sharpe ratio.

New system with taking partial profits for example has smoother equity curve if positive expectancy, therefore you can increase size a bit to get same level drawdowns.

However, taking partial profits still adds no value. As you can split trade in 2. One leave for TP1 and another TP2. If trade with TP1 is overall loser or lesser winner, you better off trading smaller size and using TP2 only.

It is more psychological thing as trading is difficult on psyche.


You just have to trust your system and play optimal moves.
  • #16
  • Feb 25, 2013 5:58am Feb 25, 2013 5:58am
I don't forget about Sharpe or its variant Sortino. Simply the equations are complicated enough.

You are right that the more the equity curve is a straight line the better. I cannot find a better metric to the optimization problem. But lowering the expectancy to increase the revenue using a higher position size doesn't sound the best solution (just a feeling). Money management comes after the optimization of a system. One cannot improve a system solely based on MM => http://www.forexfactory.com/showthre...87#post6379487

You can smooth the curve by trading several non correlated instruments (portfolio diversification) and/or by trading the same instrument with several non correlated systems => http://www.forexfactory.com/showthread.php?t=11926
In the thread Systematic trading the OP trades 20 systems (EAs) in paralell to reach a very smooth (and impressive!) curve (walk forward backtest ; http://en.wikipedia.org/wiki/Walk_forward_optimization).
No greed. No fear. Just maths.
  • #17
  • Feb 25, 2013 8:55pm Feb 25, 2013 8:55pm
Quoting PipMeUp
R is the TP when SL is normalized.
p is the probability of winning.
Pb is NOT the probability of losing but the probability of hitting the stop after the partial profit. The proba of losing is Pa+Pb, the complement of p.
I clearly wasn't being as precise as I should have been. Thanks for the clarification.

Quote
So if p=0.1, Pb won't be 0.9 but less. Otherwise Pa would be 0 and closing at PP would be Holy Grail ;-)

But you are right, you can find cases where delta is indeed positive in the different equations. But quite often if you substitute p and R in the equation for E you get a negative expectancy: you improve a losing system and usually to get a slighly less losing one. Also chosing p and R doesn't give you the associated system.

Yes I noticed that too. For p and R where E is negative, typically you do get improvement. However, less of a losing system is still a losing system. But there is a minority case where you can see improvement in profitable systems. It's hard to visualize exactly where it is. There are too many variables in the equations to plot all outcomes but I've looked at several and maybe I'll post some screen shots to illustrate what I'm seeing for the different cases presented in the pdf.

Quote
In this PDF I was mainly trying to understand why do people use these managements. They are even advised in trading books. Yet they are suboptimal. I expected that with the time and the number of traders they should have converged toward an optimal solution.

Yes it's a bit of a mystery why the heuristic world of technical analysis continues to promote ideas that in the main have nothing to do with profitability. I think it is a lack of testing that is at the root of much of this type of "noise" when it comes to specific trading advice, whether it comes from a forum, or from individuals who have "improved" their systems by unwittingly lowering the expectancy by adding a traling stop for instance. In my testing (before looking at these formulae) I noticed that adding a breakeven or trailing stop typically made things worse. In fact I never found a case where adding a trailing stop to an existing system actually improved results. And it wasn't for lack of trying.
  • #18
  • Feb 25, 2013 9:02pm Feb 25, 2013 9:02pm
Quoting FXEZ
I clearly wasn't being as precise as I should have been. Thanks for the clarification.

Yes I noticed that too. For p and R where E is negative, typically you do get improvement. However, less of a losing system is still a losing system. But there is a minority case where you can see improvement in profitable systems. It's hard to visualize exactly where it is. There are too many variables in the equations to plot all outcomes but I've looked at several and maybe I'll post some screen shots to illustrate what I'm seeing for the different cases...
you must be talking about specific edge cases. the pdf is obviously only an abstraction meant to present the general case.

Quote
Yes it's a bit of a mystery why the heuristic world of technical analysis continues to promote ideas that in the main have nothing to do with profitability. I think it is a lack of testing that is at the root of much of this type of "noise" when it comes to specific trading advice, whether it comes from a forum, or from individuals who have "improved" their systems by unwittingly lowering the expectancy by adding a traling stop for instance. In my testing (before looking at these formulae) I noticed that adding a breakeven or trailing stop typically...
some of these guys can barely operate a typewriter, let alone a computer, let alone put together a math model. "it seems to work therefore it is right". or they use broken logic to prove something to be true when it isn't. it's not exactly intuitive to go from looking at charts to 'seeing' how certain things are when looked at from an abstracted perspective, such as this thread.

but, then again, if you can target edge cases then the game changes and is less abstract.
Congratulations ScalaFX for coming out of the closet!
  • #19
  • Edited 11:02pm Feb 25, 2013 10:10pm | Edited 11:02pm
Quoting PipMeUp
OANDA and FXCM have released stats about their retail traders. 70-75% of them are losing money. That's better than the 95% urban legend but still...
The figure might actually be higher than that. If I'm not mistaken, these numbers are posted quarterly. That only means 25-30% didn't lose in that quarter.
  • #20
  • Feb 26, 2013 1:43am Feb 26, 2013 1:43am
Quoting 2+2=4ex
The figure might actually be higher than that. If I'm not mistaken, these numbers are posted quarterly. That only means 25-30% didn't lose in that quarter.
Well what amazes me the most isn't the 70% or 95% or whatever % losers but that 59% (so more than the half) of 8 millions trades were taken in the direction of the trend and the overall result is negative. With several millions trades the law of large number should apply.

This hints me that, like ferru said, the trade management is more important than the entry. Now the pdf shows in a very general way --trying to be (or pretending to be) market and strategy agnostic-- that it's not by playing around with the stop or scaling in/out... It's not the MM either. What remains? The exit? A (proper) re-entry after a loss? Adding in a trend? Something I've no clue about?

And what is the optimal way of doing it? By answering these questions each dollar (euro for me) we engaged in a position becomes smart money. Well... OK... smart pennies ^^
It's not about becoming gazillionaire next week but putting money at risk in the most efficient way.
No greed. No fear. Just maths.
0 traders viewing now