...there is a difference between

1. success of a trader
2. quality of a trader
3. what dukascopy expects from a succesful quality trader in this contest

1. and 2. can be measured by classical methods.

in fact 1. gets already measured by performance and pip counting

3. is a matter of definition

and at this point it start getting tricky, because the automated dukascopy expert was implemented to measure3., but it ends up in an interpretation of an unclear definition of dukascopys expectations regarding the trading methods and strategies of the participating traders.

additionally, before we find formulas and ways to measure "quality" we should define what quality means.

in real life trading beside risk and money management the most important quality factor is keeping draw downs as small as possible. keep your draw downs small and you are a good trader. equity protection is usually the most important target.

perhaps this is the best way to calculate the quality of a trader automatically. avoiding big draw downs needs a skillfull combination of all factors. risk and money management, the right position sizes, position sizes calculated according to stop loss targets, an adequate win ratio according to a good Risk reward ratio, etc.

dukascopy already has the needed data. just look at the performance graphs. some traders hold their highs or lose only a little ground after achieving a new high and some graphs look like rollercoasters. these rolllercoaster graphs are a clear indication of really bad trading. at least a fund manager or an institutional trader would be fired by his investors if he would perform like that. you dont need to look at any other parameters. jojo graphs show bad trading quality at a glance.

the way dukascopy calculates the "quality" of a trader right now is somehow inapropriate or at least inefficient.

what exactly is the purpose of the automated dukascopy ranking?

right now there is a contradiction. in the calculation method of the automated dukascopy ranking points.

until now dukascopy used for example R/R as a parameter in the final "manual" evaluation. but in the automated formula dukascopy uses right now, a higher R/R is not rewarded. in fact it is better to have a low R/R as long as it is above 1 and as long the trader uses equal FAR AWAY targets for SL and TP.

a trader who uses a classical pattern like for example 1:3 or 1:2 rr of 2 or 3 is getting less points than a trader who uses an rr of 1:1 or 1. (regarding the same values for TP)

this is not fair and not the right way to measure trading quality. i understand the idea of this automated quality measurment as a real improvement. it was me who proposed this from the beginning. but the formula you use should generate objective quality measurememt and not prefer some trading methods the dukascopy team evaluated lower in a manual evaluation because there the higher R/R was the better R/R!!!

so where do we go from here? we can sit and develop dozens of formulas but i guess we wont find the wisdom in winning pips and losing pips. as long as we do not combine it to draw downs and win ratios.

the part of calculating the percentage of opened and closed by limit orders is legitimate if you say dukascopy defines the way they want the people to trade. no market opening or closing appreciated. i still dont understand why this is bad trading, but okay i can live with that and deal with that.

but forcing me to let the trades open to the bitter end without being able to adapt SL or TP levels and ADDITIONALLY forcing me to use a far away 1:1 R/R makes no sense to me.

i have here a comparison with data from sunday evening. the data in this ranking was fixed and has an equal status:

parameters: opened by limit orders/ closed by limit orders/ average winning pips / average losing pips / net winning pips per average trade / DUKASCOPY points / RR

skytrader: 74 / 46 /29.98 / -21.3 / 9.15 = 57 RR: 3.329 (graph steady, holding the high)
managedfx: 29 / 48 / 52.47 / -34.44 /11.91 = 66 RR: 2.84 (graph rollercoaster)
Gusman: 12 / 71 / 92.83 / -78.89 / 12.23 = 76 RR: 2.416 (graph steady)

i have chosen these three examples to illustrate that 2 traders who have worse stats than skytrader regarding their opening and closing stats get 10 or even 20 dukascopy points more than skytrader! although their real RR or lets say the net average pip outcome is nearly the same.

so where comes the difference in the evaluation from?

it is the absolute size not only of the TP but also the SL and this in fact is inapropriate in my eyes and by no means any sign for a better trading or even a trading style that dukascopy should favor. if these guys would at least show a much higher R/R i would somehow understand that better, but not with an almost same real R/R.

and the way these numbers are generated right now, my dukascopy points would be even worse if for exmaple my average lost pips would be lower! lets say 5.6 instead of 9.15. that is really strange!

i guess this idea is not finally developed and i am pretty sure that there will hopefully be some improvement, that makes more sense and is more fair.

thanx for your attention.

kind regards

skytrader

Thanks for the detailed feedback, skytrader.

You are right, there will definitely be improvements and amendments. As you correctly told, it looks like finding an ideal formula will be an endless process. Moreover, it seems that it cannot be fixed at a certain time because we have to remain flexible and to adjust to the changing tendencies of the contest.

One of the coming amendments: starting from May we plan to substitute the "Share of positions opened by conditional orders" with the RR ratio and, thus, to bring it back.

That would be awesome !

its great to see you working so tough on this, but it is also necessary, because automatic evaluation will be the only way to cope with the future, when you will have 2000 or more participiants. but, before bringing back R/R like you have it right now, this R/R feature needs some improvement. as i have posted in a post at the general discussions the R/R how it is right now can be "cheated" easily by placing many orders with the minimum SL and for example 150 pips TP. the only downside of this cheating: one ruins his stats that count right now for the dukascopy points. if you now take out again "Share of positions opened by conditional orders" and still calculate only the closed by limit orders for the R/R like you have it right now this would result in even more and effective cheating or lets name it cosmetics to the stats.

i guess what you started with, the net average winning pips and average losing pips is already sometghing like a real R/R. perhaps all oyu have to do is to compare the winning pips of a trade with the SL it had and the losing pips of a gtrade with the TP that was entered with the order. than you would have an exact and true R/R based on the real values. then you would only have to find a treatment for the moved SL or TPs that you pkan to enable, becasue this is soemhow a game changer then again. it will always be. so what a trade rplans, when he enetrs a trade should have no relevance at all. the real numbers he produces should only be counted. so your appraoch right now is only a simplified attempt to achiev exactly this. becasue you dont have a method right mow you measure only the conditional closes and opeings to determine R/R. perhaps my idea is worth 2 cents, and you dig into it deeper.

with the above described method it would be easy to calculate an exact and real R/R for every single trade and out of these then an average. a real R/R is important. and to determine the R/R ALL the trades must be taken into account. if you in the next step combine this to a win ratio in a way that makes sense you get much more qualified data!

one little thing to your first sentence:

i dont think right now not the "Share of positions opened by conditional orders" are the problem in the inapropriate claculation. as i said, this is a circumstance you can easily define and everybody who knows it must deal with it. the real problem is the calculation of the third addend or multiplicator! adding of winning and losing pips dived through 2 is the problem!

here is why:

the methodology:

The total Dukascopy that a trader receives is assigned basing on the following parameters:

Share of positions opened by conditional orders (the more - the better) - 25%
Share of positions closed by SL/TP (the more - the better) - 25%
Average amount of pip won/lost per position (all figures taken by modulus) - 25% *

* Detailed explanation of the 3rd parameter's calculation:

The average amount of pip points lost per position is calculated for negative trades
The average amount of pip points won per positions is calculated for positive trades
If the average negative pips (by modulus) are higher than positive ones - the participant receives a Zero for that category
If the average positive pips exceed the average negative ones - the following formula it used to define the standing of the participant: "(Average positive pips + Average negative pips) / 2". All figures are taken by modulus.
Participants are ranked and receive their scores on the basis of the derived figure.

the problem is the formula:

perhaps i understood your addition wrong? do you add the real negative value of the average pips of the losing trades?

example1: average winning pips 60, average losing pips -20.

do you then calculate (60 + 20) /2 = 40

or do you calculate

example 2: (60 + -20)/2? = 20

what i want to show is this: if a trader for example has a strategy with typical TP 60 pips and SL 20 pips (trader 1) with the calculation method in example 1 he wont get such a high addend or multiplicator like a trader who has SL 60 and TP 60 (trader 2) although he has a much better R/R ratio.

calculated like in example 1: it would be for trader 1 (60 + 20 )/2 = 40 and for trader 2 (60+60)/2= 60

calculated like in example 2: trader 1: (60 -20)/2= 20 and trader 2 (60-60)/2=0 which does not make any sense, because you already set the value to zero when the negative pips are higher than the positive pips. so it must be calculated like in example 1.

and if the addend or multiplicator gets calculated like in example 1 this would mean, although trader 2 has a worse REAL R/R ratio (1:1 or 1.0 ) than trader 1 (1:3 or 3.0) he gets a higher multiplicator or addend, just because the absolute values of TP or SL are higher.

this would mean, that you favor clearly traders who set far away targets and "punish" traders who use for eample only 60 instead of 150 or 100 pips as TP, although they might have a much better or higher R/R? or even less logical, when trader 1 and trader 2 set the same TP far away, lets say 150 pips, then trader 2 gets a much better ranking value because his average loss of pips is 150 pips and trader 1 perhaps has only 50 becasue he always stes there his SL.

i really dont understand the logic behind that. and i guess there also is none.

did i understand you correct on this?

and if so, how does a much higher absolute value for SL and TP bring more quality in the trading of a trader? and if it is not about quality and you say it is about the way dukascopy wants the traders to trade, pleas explain, why you want to evaluate the traders with the farest TP and SL values best?

i am pretty sure that most of the participating traders are not aware of the fact that they must use equal TPs AND SLs of 100 or higher if they want to have a chance to get some substantial automated dukascopy expert points!

with this claculation method you cripple the creativity of different trading styles although this should be your highest interest, to introduce different trading methods and strategies. a lot of meanigful and succesful strategies simply dont work with the values one needs if he wants to get 80 or more DUKASCOPY points!

for example just because they might only have a win ratio of 40%. if the trader has to set his SL to an equal value like his TP he must hit more than 50% to make money. with an R/R of 3.0 he needs to hit only 1out of 2.9 trades to make money.

you see the problem? you exclude some of the most classic win/loss ratios from a good evaluation.

so please change this and use the only real quality measurement like drawdowns or for example the existing graphs. some traders graphs look like rollercoasters. this is a sign for really bad trading, although some of them get automatically an evaluation of more than 80! the good traders you can easily identify. if they achieve a high in their graph they dont lose much ground...they go sideways or retrace a little bit to climb further. the key word is equity protection and draw down. these are the keys to a succesful high quality trading and not choosing 150pips TP and 150 pips SL and closing all orders by limit orders without manual market control.

dukascopy should be interested in a big variety of trading styles, techniques and methods while still being able to determine objectivley succes and quality through measurement methods that already get used in the finance world like equity curves or draw-downs.

thanx for your interest, your attention, your efforts and your time.

kind regards

skytrader

the calculation occurs at every single trade. no trade is left out of the calculation!

the R/R is calculated like this:

in case of a winning trade, the system compares the winning pips to the entered SL

in case of a losing trade, the system compares the losing pips to the entered TP.

out of these single trade R/Rs the system calculates automatically an average value. that is the final true R/R with which an automatic evaluation can occur.

like this you get a true R/R out of every trade, and it does not matter if closed manually or conditionally. no cheating anymore. traders who enter 50 pips SL and 150 pips TP and then close after 7 pips win get an awful R/R. thats the way it should be. and it should be quite easy to code. this true R/R is then a substantial parameter, that especially can be taken into account for calculating fair DUKASCOPY points!

i believe this contest will have a great future and will fullfill its purpose for dukascopy. it is worth to invest a lot of brainpower, what of course all of you in the dukascopy team already do!

but sometimes ideas that come from outside the developers team can also be charming! i have already developed myself web based applications. i know that every new function can be somehow like opening pandoras box.

right now, although proceeded already 6 months it is still somehow at the beginning of the process. to develop a really incorrupt automatic evaluation that is also capable of showing real quality in the trading you will have to add more parameters step by step.

A TRUE R/R is the first one. next should be a combination of the true R/R with the true win ratio. after that subjects like equity curve, draw downs will also make sense.

the evaluation section could and should one day also be a perfect stat center for a trader who wants to analyze his trading or different strategies. you can do so much with that! such a well thought out stat center would give a huge added value and make the community and the contest even more interesting, also for experienced traders who want to analyze their trading.

if you are interested in a creative consultant, just contact me.

I have a hard time taking seriously someone who suggests that "the performance graph be the way to show quality trading" when all but one of your month's performance charts, directed at you Skytrader, shows serious draw downs and sub 100k endings. The only reason you have achieved the high rankings that you have is based off the questionable popularity score Dukscopy seems to favor in an attempt to draw people to their site. Take out the popularity score... and the subjective Dukascopy assesment.... and keep the only two true quantifiable rubrics for trading success, pips and profits. I think I speak for the majority of the participants of this contest when I make that suggestion.

From what you said, skytrader, actually you are proposing the risk-adjusted return, in fewer words.
Risk-adjusted return is in fact the ratio between profit vs drawdown(standard deviation of risk). It would be fair in general, however every trader have a different trading style and despite the final formula it will always make someone unhappy. For example you agree with pending orders score but I find it useless in determing the value of a trader, because it forces me to use them all the time otherwise I will lose points.
Regardless of ranking's calculation it has to be a simple formula for everyone to understand. If it would be too complex maybe you will understand it, but the majority of traders will need to spend time to figure out the mathematical formula instead of enjoying the contest.

In conclusion, even if I do agree with you (skytrader) to make the contest more fair in terms of keeping the score, I can't agree to adapt the calculation of rankings to fit your trading style or another. As a competitor I have to thank you for your interesting and original ideas in order to make to contest better.

Until now, I am more than happy to see the amazing improvements of the contest, month after month.

especially the automatic calculation of the ffactor or addend used right now to determine the dukascopy points is right now inefficient and unlogical. it is not a complicated formula but simply wrong the way it does it. the true R/R that i suggest is fair, because it makes cheating impossible. all you have to do is to compare the winning pips /the real TP) of every single trade with the intitial SL it had and in case of a losing trade all
you have to do is to compare the losing pips of a trade (the real SL) with the initial TP. what you then get is an an incorruptable real and effective R/R! the R/R right now is a joke. they know it themselves, otherwise they would not have put it out of the claculation. it takes only into regard the trades opened and closed by conditional orders. that is not usable because it is not cheatable.

thanx for your support on some of the ideas OneGoodTrade!

dear anonymous879, posting such abusive content anonymously says everything about your character and your guts.

i appreciate your suggestion regarding the popularity. by the way, i think i was one of the first ones, to propose this. i did this already 8 weeks ago. i proposed to reduce the importance of popularity. do the research before you start your ill minded, meaninglessenvyful rants. we are not talking about my performance, we are talking about improving the contest evaluation methods. igathered my performance when i reached 1400 unique IPs while winning this contest. and with 100 points in popularity i proposed to reduce this parameters importance.

your childish post made me smile. i dont udnerstand why dukascopy allows such abusive content in such a dsicussion. lets see if they show this one. perhaps you are already fed up with constructiv, creative critique dukascopy?

if so, just say it. i dont need to post here. although my idea of the realtiem dukaskopy evaluation was implemented. in a strange way but nevertheless implemented.

the platform already generates enough data. all it needs is an intelligent and meaningful combination and application of the already existing data. look for example how you calculate the pip addend in the formula. if you understand what i have explained it is totally wrong using the data that way. you could use it much more efficient to generate really comparable values that also stand an objectivity test.

i dont agree at all with the pending orders! the markets in our days move in extreme volatility. i dont think this parameters have any useful meaning! i just wanted to show in that example that one could find that legitimate, because it is the definition of the contest rules and DUKASCOPY perhaps wants this trading style. i just wanted to compare it to the total useless addend they are calculating right now. look at my example. before i got a higher value of average losing pips than winning pips my average losing pips were at about 9. my average winning at about 20. with the average losing pips rising becasue i made some bad trades in a row, my dukascopy points actually got higher before the addend became per defitnion zero!

the formula is not complicated, in fact it is too simple and it makes no sense.

but i definetely do agree with you that using pending orders has nothing to do with trading quality. but i also understand dukascopy that they dont want the 3 to 7 pip scalpers acting totally without any pending orders. so rewarding pending orders is okay, but punishing it too strong in an evaluation is not okay.

and if DUKASCOPY uses the true R/R calculation like i have proposed it such a problem would not exist, becasue the "quality" would be measured obejctivle! if someone enters an initial SL of 100 pips to then close manually after 7 pips. the true R/R like i proposed it would show that clearly! he would have least have to enter 10 pips SL if he wants to scalp without ruining his R/R totally and if he then closes manually before he reaches 10 pips TP, his R/R would also go under 1.0.

the method to determine the real R/R by comparing true closing values with their initially entered counterparts is plain, simple and VERY EFFECTIVE!

i cant wait to see it in the contest. i am pretty sure dukascopy will understand how good this measurement method would work.

[quote="Anonymous879"and keep the only two true quantifiable rubrics for trading success, pips and profits. I think I speak for the majority of the participants of this contest when I make that suggestion.[/quote]

anonymous, besides the low profile approach of your anonymous post the content of your words is totally meaningless, because measuring only the profits and pips would make this contest to a gamblers bonanza.

a lot of developers at dukascopy and the inventors of this contest think about means to avoid exactly this.

thanx for your comment, because it helped to determine quickly how much you are into this complex subject and additionally it was quite amusing.

in my opinion pips should be more important that balance. much more. and balance more important than popularity. important besides the pip performance should be objective quality measurement.

ranking calculation should be like this:

PIPS 35%
DUKASCOPY 35% (if dukascopy gets to objective formulas and parameters like true and incorruptable R/R)
BALANCE 20%
POPULARITY 10%

It sounds fair enough. But I do not know how many traders would agree. skytrader if you care so much indeed and Dukascopy agrees with you maybe you should try to make a poll.

Since you want someone to point a finger at... I am the previous poster. I find keeping a low profile classy, but if you need a more personal approach that is fine with me. I do not think it is appropriate for the contest to provide a quarter of the rating on popularity, which has nothing to do with actual trading. You are benefiting strongly from this facet of the contest, no denial.

Your opinion that the dukascopy assessment is appropriate, is based on the assumption that they know how to really quantify "quality" trading correctly... Just like any panel of judges they are going to be subjective, and their decisions will be based on opinion.

Why not cut the opinion out of it when we have two accurate metrics, pips and profits, for measuring a traders skill?

One major flaw in the Dukascopy assessment revolves around the whole concept of risk/reward. Anyone who understands statistics and knows the term "expected value". I'm not going to go into the topic, you can google if your interested, but expected value trumps risk/reward in terms of a successful trading approach. The difficulty is that it would be impossible for Dukascopy to evulate based on expected reward without having to do an extensive tecchnical analysis on each trader's system. My problem is that I trade based on expected value, not risk/reward, and even though I am using a more technically sound fundamental, I get punished for it. This proves the point that the evaluation is subjective, based on opinion, and not truly a metric for quality.

Sure, you are entitled to your opinion, which is that pips are more important than profits. I obviously disagree. I can show you at least a dozen traders in the contest this month who have sub 100k balances, but have positive pips. In real world trading, its money management that makes you a good trader... You can easily have positive pips every month, yet end up with less money than you started, I can gladly point out a few traders in this months contest that are doing just that. Yet if the opposite is true, and your trading with real money, who really cares if you have negative pips, but sitting on a profit?

Hence, profit>pips in my mind... I do not see a logical argument for the other case.

Also, the organizers of this contest have added parameters to limit "gambling", which makes it possible to use only pips and performance as the only two quantifiers needed

"thanx for your comment, because it helped to determine quickly how much you are into this complex subject and additionally it was quite amusing." - Grow up.