Does the trader effect affect robust trading systems?

[Sicilian Trader]

I have recently finished reading Way of the Turtle by Curtis Faith, a pretty good read. However, its probably a little out of my league given there is a good chance - based on the position sizing he states in his book - that i am well and truly under capitalised if i were to emulate some of the strategies adopted in the book. So i will put it down to an entertaining read, for now.

There were a quite a few of interesting things I read about in his book and one of which has sparked some additional interest from me. And I thought I would share this topic with some of you. Its on the topic of the “trader effect”. I have tried to search for this trading phenomenon on the net and on this forum , as well as others, but couldn’t really find anything of great depth. The closest i came was on the thread below. I have attached 2 of the more interesting posts on this topic. One from Nick Radge who gives a thorough run down of what it is and the other from Markrmau, who gives a good analogy for some of the beginners who may not fully grasp all the aspects of trading just yet (me included).

Curtis Faith describes the “trader effect” in his book as :

“....The fact that a particular method has made a lot of money in the recent past increases the likelihood that other traders would have noticed it and will start using similar ideas, increasing the chances that the method will not work as well as it did initially...”

He goes on to say “....An observer effect is a concept from physics in which the act of measuring a phenomenon affects the phenomenon; the observer affects the experiment by the act of observing. A similar thing happens in trading: The act of the trading can change the underlying market conditions on which the success of the trade is predicated. I call the this the trader effect. Anything that repeats with enough consistency is likely to be noticed by several market participants. Similarly, a strategy that has worked particularly well in the past is likely to be noticed by many traders, however if too many traders start to try to take advantage of that particular strategy, that strategy will cease working as it did previously....”


If this holds true with all trading systems, indicators, well known methods etc, then this gives me some concern.

I have seen a real time example of a trading system being "arbitraged out". Roger Montgomery released a Hang Seng trading system several years ago. Whilst the system's integrity was flawed to start with (for another thread) the system quickly deteriorated due its users.

Basically the system traded an illiquid market which in itself was a poor choice. Then when people started getting entry/exit slippage they decided to buy or sell ahead of the official entry price so they could get set at a better price. This slowly built upon itself until these "front runners" were essentially triggering the systems signals rather than allowing the market itself do it.

I used a Hang Seng System that tended to trade in the same direction as Montgomery's, however we did share the same exit. Because I knew that all his minions were going to exiting at my level I would initiate my exit ahead of theirs, to their detriment and my gain.

Other popular patterns/systems that have now broken down is the Larry Williams Oops! pattern (which in fact was first revealed back in the 50's before Williams was even trading). That pattern simply does not work on major markets anymore. If you do historical testing up until his book "Long Term Secrets to Short Term Trading" was released, you can see that it deteriorated from that point on. People basically front run to beat the crowd.

Another system that has deteriorated is the simple channel breakout created by Donchian back in the early 70's. We know this system now as the Turtle System but such basic breakout systems do not work anymore.

I guess I am using the term arbitrage quite loosley here. [ie. A real example of arbitrage might be buying a news paper for $1.50 because it has a cupon to recieve a free meal at maccas worth $5]

As Nick and others implied, I am using 'arbitrage' to describe the situation where everyone tries to use a publically known system - and this tends to raise the entry and drop the exit price (assuming a 'long' system) until there is no net gain.

If the newspaper and maccas markets were sufficiently liquid, what you would find is that all the newspapers would be sold out (increasing thier price on the 'black market'), and the queues at maccas would increase. The original price differential would be nullified by the increased cost of the paper, and your '$ time value' wasted in the maccas queue. So there would be no net benefit from using the system. (ie it has been arbitraged out)

If gives concern because of the following:

Hypothetically speaking , lets say I have designed, back tested, in sample / out sample date, forward tested, MC analysed, and optimised what i think is a very robust and dynamics system. It has all the hallmarks of a great system and I it will be very hard for me to gain any more confidence before going live it it. Because it a positive expectancy, a good pfactor, CAR%, MaxDD, win% , etc. Its a long system designed for up trending markets etc, and my rules are simply ‘to follow the rules’, on entry, exit, position size, money management etc.

The problem arise here: say i hit a period of significant drawdown. (fine most would say ‘just keep going if you have a positive expectancy’) only now you have exceeded your maxDD (20% based on 10 years backtested data using MC analysis) and you are now reaching 30%. You’re consecutive losses are now greater than the max in your testing (12), you are currently at 15. Max drawdown so far is $28,000 (backtest showed a max of 20K)

WHAT WOULD YOU DO????

I read many posing this question on threads with varying answers. And the majority I have observed would say keep going (which is where my thinking would have been). But i after giving more thought to the trader effect, i am now not as confident that could continue trading during a new maximiumDD. What if the market has truly cotton to my macd crossover coupled with rsi and Bollinger bands (or whatever it is). Doesn’t that mean that all those popular indicators are now arguably obsolete and new traders like me should use them because they provide a false sense of security?

What about people writing books, selling trading packages, and running courses on established trading systems? By the very existence of this phenomenon, arent they are diluting their own profits by selling their system? or maybe their system used to work but now the profits aren’t as great so they realise they can make more money selling a ‘proven system’?

So my question is what are peoples thoughts on the 'trader effect', does it really exist? How long does each indicator or system last before people cotton on destroy?

In the above example would you continue trading in that maxDD phase of your trading??

look forward to hearing yoru thoughts

 

[tech/a]

The problem arise here: say i hit a period of significant drawdown. (fine most would say ‘just keep going if you have a positive expectancy’) only now you have exceeded your maxDD (20% based on 10 years backtested data using MC analysis) and you are now reaching 30%. You’re consecutive losses are now greater than the max in your testing (12), you are currently at 15. Max drawdown so far is $28,000 (backtest showed a max of 20K)

WHAT WOULD YOU DO????

My answer would be the complete opposite.
Your tested results give a blueprint which allows you to measure your forward trading to that tested.
If it falls significantly outside of its blueprint (as mine did (Tech Trader))
You would as I did stop trading it.

My test period was 8 yrs and to find market conditions which encompass all market conditions which could be faced in a systems lifetime in 8 yrs just isnt going to happen.

The trader effect does occur and I see it regularly during trades where you'll see massive slippage on stops held in common technical points of Support or Resistance or pattern failure.

Although as pointed out system failure can and does in some instances occur its more likely that the system (well your or my system) is likely to degrade or fail due to market conditions not seen in any testing.

In my opinion.

 

[Sicilian Trader]

thanks tech

is there anyway of testing for trader effect on any of the testing platforms available?

ie. can one test the mangitude of a profits on say the the 5ema and 10ema crossover, for example, and see if in time its profits reduce due to trader effect? or whether there was a proven indicator that worked very well , say 10 years ago, but now there risk reward ratio is such that (1:1) you wouldnt take the trade?

i know its a bit of a long shot but just keen to know if anyone has been able to quantify this effect over time in their testing??

tech, i completely understand your point about affects of unforseen market conditions (not seen in testing) outweighing the trader effect phenomenon

thanks for your input

 

[beerwm]

wouldnt such a collective use of a system have a positive effect,

for example,

if more people were to use fibonacci retracements then you could predict the market to a better extent.

same with trend following,

 

[Sicilian Trader]

wouldnt such a collective use of a system have a positive effect,

for example,

if more people were to use fibonacci retracements then you could predict the market to a better extent.

same with trend following,

thats what i initially thought, though after reading some of the examples regarding slippage and front running (which also make sense), its now made me a little confused about the trader effect, hense the reason for me posting the questions

anyone else have views on the trader effect?

 

[Nick Radge]

wouldnt such a collective use of a system have a positive effect,

No because people will tend to 'front run' the signals rather than take them at the required level, or people will front run the stops.

A great example of this theory is Larry Williams "oops!" pattern in the S&P 500. It worked a treat until he published it then it went pear shaped.

 

[grid]

People like Home Trader would dispute this.

However I believe Nick is right. The HT people have a less experience to know this stuff.

 

[Sicilian Trader]

No because people will tend to 'front run' the signals rather than take them at the required level, or people will front run the stops.

A great example of this theory is Larry Williams "oops!" pattern in the S&P 500. It worked a treat until he published it then it went pear shaped.

I read up on it and i see what you're saying

So with the exception of trading educators like yourself, and others who educate on money management and risk, etc , why would anyone sell or divulge their proven profitable systems when there is so much proof (larry william oops) saying that once everyone catches on to a profitable signal then that signal ceases to exist (or to at least remain as profitable if taken)?

Does that mean that what everyone on this forum should be focusing on is to develop their own indicators, systems (abandon the macd's, the boll bands, and every other conventional and well knownl oscilalltor) and just stick to their methids and TELL NO ONE because the act of doing so will invariably and indiectly dilute those profits that they have been enjoying for all this time. Would this be the inevitabe outcome or sharing profitbale systems with people?

 

[aarbee]

People like Home Trader would dispute this.

However I believe Nick is right. The HT people have a less experience to know this stuff.

I think the 'Trader Effect" certainly exists but it's effect is overstated. The subsequent failure of Larry William's system or the Darvas system would perhaps be attributed to the fact that these are really well known around the world.

Recently, some standard HomeTrader systems which backtested and traded well were backtested on the Toronto Stock Exchange and the results were very impressive and in line with those shown on the ASX. Talk of really OOS backtesting. This would indicate to me that the systems are indeed robust. The Trader effect does not show perhaps due to relatively small membership.

Just my 2c
Cheers

 

[Sicilian Trader]

The Trader effect does not show perhaps due to relatively small membership.

this instills a bit more confidence, i would be inclined to agree. I guess another consideration would be who the authority is who is saying 'here is the method, it works' and how well known the methodology is. I would dare say that if i published my winning formula then it may not have as much of a following as larry williams - just a hunch.

 

[James Austin]

this instills a bit more confidence, i would be inclined to agree. I guess another consideration would be who the authority is who is saying 'here is the method, it works' and how well known the methodology is. I would dare say that if i published my winning formula then it may not have as much of a following as larry williams - just a hunch.

The guy that headed up the turtle traders is reputed to have said something along the lines of, "even if my strategy were printed on the front page of the NYTimes, it will have little impact. Traders have an inability to follow rules".

So unless the strategy is computerised/mechanical or very simple (ie., crystal clear, right across the board), I suspect the "trader effect" is going to struggle to take hold.

 

[James Austin]

wouldnt such a collective use of a system have a positive effect,

for example,

if more people were to use fibonacci retracements then you could predict the market to a better extent.

same with trend following,

No because people will tend to 'front run' the signals rather than take them at the required level, or people will front run the stops.

A great example of this theory is Larry Williams "oops!" pattern in the S&P 500. It worked a treat until he published it then it went pear shaped.

hmmm,
so many know about fibs, and they still seem to show some validity

https://www.aussiestockforums.com/forums/showthread.php?t=14459

 

[aarbee]

Although as pointed out system failure can and does in some instances occur its more likely that the system (well your or my system) is likely to degrade or fail due to market conditions not seen in any testing.

In my opinion.

This is where Van Tharp's way of defining the different phases of the market and then working out your system's "System Quality Number (SQN)" for these phases of the market to really understand the behaviour of your system would be handy. It's a more objective way of deciding whether the system is broken or the market is out of whack with what the system was backtested on.

Cheers

 

[Cartman]

So with the exception of trading educators like yourself, and others who educate on money management and risk, etc , why would anyone sell or divulge their proven profitable systems when there is so much proof (larry william oops) saying that once everyone catches on to a profitable signal then that signal ceases to exist (or to at least remain as profitable if taken)?

howdy ST ----- if u had a ball tearer of a system, the best thing u could do is sell it to a squillion people ----

1) U make a motza selling it
2) U make a motza trading it cause all yr 'buddies' are moving the price around exactly where and when u planned it
3) when u eventually sell half a squillion more than youd planned so the system gets a bit too predictable ---- u still make a motza cause u then become the "smart money" front running yr own system

 

[Sicilian Trader]

This is where Van Tharp's way of defining the different phases of the market and then working out your system's "System Quality Number (SQN)" for these phases of the market to really understand the behaviour of your system would be handy. It's a more objective way of deciding whether the system is broken or the market is out of whack with what the system was backtested on.

Cheers

hi rb, thanks for the systems, will look into it

is that like the t-test in stats , where they look at the results happening because of your inputs rather than just by chance alone?

i had a look at sqn and its formula reads very much like a t-test

sq root(N) x (average trade/ standard deviation of trades)

is this the test you are referring to?

 

[Cartman]

hi rb, thanks for the systems, will look into it

is that like the t-test in stats , where they look at the results happening because of your inputs rather than just by chance alone?

i had a look at sqn and its formula reads very much like a t-test

sq root(N) x (average trade/ standard deviation of trades)

is this the test you are referring to?

ST ---U R obviously well versed in mathematics --- i assume u r trying to develop an auto system to apply to FX trading ?? --- have u tested yr system parameters either live or on a demo a/c to assess the possibilities? -- and over what time frame?

 

[aarbee]

hi rb, thanks for the systems, will look into it

is that like the t-test in stats , where they look at the results happening because of your inputs rather than just by chance alone?

i had a look at sqn and its formula reads very much like a t-test

sq root(N) x (average trade/ standard deviation of trades)

is this the test you are referring to?

Spot on ST. It is indeed a T-score.
The idea is that it is pointless trying to make the same system work in every market phase. Define the market phases objectively and backtest the performance of the system in the different market phases so one has a far better idea of system's performance or lack thereof at any given time than just trying to second guess it.

Tech is right in saying that the system is not working when the performance is significantly different from backtested results. The problem is to define that "significant" deviation objectively. When in a deep drawdown, every losing day appears like a 'significant' deviation from backtest results. I had a rule that I will stop trading the system when the DD exceeds twice the historical DD. I stuck with the system but boy was it painful. Taught me that my pain threshold was not as high as I thought so I went back to the drawing board to redesign the system.

At the moment, I'm trying to incorporate VT's way in my system design.

Apropos Trader Effect, Howard Bandy is of the view that over time systems will fail. The systems make money by exploiting inefficiencies in the market and over time when a particular inefficiency ceases to exist, system performance degrades.

James Austin is quite correct in inferring that it is the trader who is likely to fail rather than the system.

Cheers,

 

[Sicilian Trader]

ST ---U R obviously well versed in mathematics --- i assume u r trying to develop an auto system to apply to FX trading ?? --- have u tested yr system parameters either live or on a demo a/c to assess the possibilities? -- and over what time frame?

hi cartier,

i am relatively new to trading and am in the process of putting some trading parameters into amibroker, i have been live trading for about 8 months (profitably) but have come to realise, with self-education, that i dont even know what the keys stats are to my current trading system, its a little ad hoc.

With all the info there is on backtesting, pos sizing, money management etc, i think i will stop trading for a couple of months (because i dont know what my 'risk of ruin' is - could be bankrupt with my style of maverick trading - who knows??). So once i have got a better grip on my system, i will then be able to answer your question a little more accuracy. At the moment , its in its infancy stage.

btw, i am by no means well versed in mathematics , just remember some basic stuff from quant methods 4th yr uni, long time ago now. Not trying to automate a system, even though that is what it appears to be, still want to be in control and have some discretion over trading, (ie. not completely ignoring the fundamentals - i personally think that would be a little dangerous IMHO)

good chatting to you cartier

nick radge, on a side note, bought your book on adaptive analysis this arvo and havent put it down, nearly finished. Defo one of the better ones i have come across. Really like the 'no nonsense / straight to the point' approach.

Highly recoommended

On another side note, does any use the kelly criterion when deciding upon position sizing / leverage (in aid fo maximizing returns with the account you currently), would be very keen to know what peoples thoughts on this?

ST

 

[Cartman]

hi cartier,

i am relatively new to trading and am in the process of putting some trading parameters into amibroker, i have been live trading for about 8 months (profitably) but have come to realise, with self-education, that i dont even know what the keys stats are to my current trading system, its a little ad hoc.

With all the info there is on backtesting, pos sizing, money management etc, i think i will stop trading for a couple of months (because i dont know what my 'risk of ruin' is - could be bankrupt with my style of maverick trading - who knows??). So once i have got a better grip on my system, i will then be able to answer your question a little more accuracy. At the moment , its in its infancy stage.

ST

ST ---- if u have 8 months profitable trading in the current market, i'd be forgetting the back testing issues ---- just keep doing what u r doing ---- sure, adjust yr position sizing and MM to suit yr style, but dont bog yourself down with superfluous information from the past ------ unless its from a comparative trading era, which may be of some benefit ------- trade the now. and the future --- not what happened 10 years ago ----- all just my humble opinion of course ----- Cartier lol

ps i still suspect yr maths is ok

 

[Sicilian Trader]

fair point cartier, but with all my reading and learning (mainly thanks to this forum), i realise that leveraging up (average 50:1) on blue chips when the xjo is near its lows (who knows where that truly is) and hoping for the best is probably not the most sound strategy. Probably akin to spray and pray, and hoping that my obscene leverage will be my friend when i need her most.

Not sure i can really hang my hat on that and call myself a good, consistent trader as yet, far rom it ,because based on my current strategy i could blow up my account when the xjo hits its new lows again (if i am still long). That said, i cant complain about my returns as i hav never really seen that amount of money before in such a short period of time, so what i might do is ease up abit, and learn in the background, as i know that sooner or later i will invariably be giving some back. I thnk its been beginners luck (analagous to flipping 7 tails consecutively out of my first 10 coin flips - there has to be a series of heads in there somewhere)

what kind of position sizing strategy do you adopt cartier?
:luigi: