;-)

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

;-)

Post by Bernd » Mon May 05, 2003 8:41 am

:wink:
Last edited by Bernd on Thu Apr 17, 2008 11:23 pm, edited 1 time in total.

rs

Post by rs » Mon May 05, 2003 12:40 pm

Bernd,

This is a particularly interesting topic for me as I trade in a long term trend following style in a discretionary manner. Although, I have rules about setups, exits and money management/ position sizing which are systematic, the ultimate decision to take the trade is somewhat discretionary and based on pattern recognition among other things.

I have traded this way for just over a year now and have achieved reasonable success. I have made approximately 45 trades over this time period and thus have tested in real time the discretionary 'system'. I have gained a fairly good idea of no of losses in a row and drawdowns but of course one year is probably not a long enough time period to be sure.

What I do know, however, having gone through up to 30 years worth of charts of all different markets and looking at them and the patterns that are generated, is that the method that I use to pick trades is based on sound trend-following principles and setups that occur repeatedly within all markets. So the theoretical basis for my trading is sound.

It may mean that I miss some trades and trends but I also take less losses than a purely systematic trader would take, I imagine.

Psychologically, it is not that difficult to stick to for me since I am fully convinced that my methodology is sound. I don't think it is any harder than having to trade a mechanical system.

Thanks

rs

Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

Post by Bernd » Tue May 06, 2003 4:50 am

:wink:
Last edited by Bernd on Thu Apr 17, 2008 11:25 pm, edited 2 times in total.

rs

Post by rs » Tue May 06, 2003 7:05 am

Bernd,

I did not do any paper trading as such but started in right away with trading. Prior to starting trading I did, however spend a lot of time and energy trying to identify what factors were present at the beginning of large moves and noting them. I also was looking for a method which would indicate a clear place to put a stop which made sense.

It took me a long time to come up with a method and I have made slight alterations as I have traded, so the system has evolved as time has gone on. Not in a major way, but small alterations.

It is important to note, as you have already, that the purely systematic approach also has a lot of positive factors. I am in fact considering shifting to a purely systematic approach or at least running one alongside my discretionary account. It is much easier to backtest large amounts of data in a meaningful way under a mechanical system framework. It is also possible, if not likely that I could find a much better system by using a backtesting engine.

rs

TradingCoach
Roundtable Knight
Roundtable Knight
Posts: 176
Joined: Thu Apr 17, 2003 9:52 am
Location: Sacramento, CA
Contact:

techies and the market

Post by TradingCoach » Tue May 06, 2003 11:55 am

I hope no one will take this in a wrong way. Your initial post is very interesting and thought provoking. I have some fin. Technical background and have traded for at least a decade as a discretionary trader before I went long term trend following in about 1999-2000.
My observation is that technical people who try to quantify everything 100 percent either fall into two categories (indicator types and mechanical systems types who discovered money management.) Obviously the later is a better situation but my feel is that you always have to watch the markets and override err. tweak when necessary.
Tape reading is something we all need to learn regardless if we want to or not. Just my .02 cents
Last edited by TradingCoach on Wed May 07, 2003 11:21 pm, edited 1 time in total.

Howard Brazzil
Roundtable Fellow
Roundtable Fellow
Posts: 54
Joined: Wed Apr 16, 2003 12:45 pm
Location: Houston, TX USA

Post by Howard Brazzil » Tue May 06, 2003 12:14 pm

It was my wife who said: "You can't spend all your time testing systems. You have to start trading
once and collect some experience." She said also: "You are always trying to be safe. With your testing
of systems you are trying to avoid risk and to be on the safe side". She made me think...
Dear Bernd,

It sound as if your wife is not only supportive of your trading—which
is great—but her comments are very insightful, as well. I think you are
a very lucky man !

- Howard

Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

Post by Bernd » Sat May 10, 2003 4:17 am

:wink:
Last edited by Bernd on Thu Apr 17, 2008 11:26 pm, edited 1 time in total.

Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

Post by Bernd » Mon Aug 25, 2003 4:37 am

:wink:
Last edited by Bernd on Thu Apr 17, 2008 11:26 pm, edited 1 time in total.

Luke
Contributing Member
Contributing Member
Posts: 5
Joined: Wed Jun 11, 2003 9:12 pm
Location: Australia

Post by Luke » Mon Aug 25, 2003 6:58 am

I always thought I could use my experience/intution/judgement to improve the performance of my mechanical systems. I spent hours a day reading the news and gauging sentiment, and then only took a system generated signal if my analysis was in agreement. The theory was that I could filter out the trades that made no sense, or stay out altogether in uncertain times (Iraq war, big economic numbers etc).

As it turned out, my interfering added no value whatsoever. I missed a lot of trades, and their average return was about the same as those that I took. The end result was a lot less $ in the bank and lots of time wasted reading the news.

I am not saying that discretionary trading is not useful, only that it has not been beneficial for me. I have concluded that I have no real ability as a discretionary trader, my real strength lies in designing robust systems which is where I focus my time on now.

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Mon Aug 25, 2003 7:31 am

Luke,

When you use discretionary overrides on a system (I dont know your system so may be wrong) you are likely to be making anti-trend decisions more often than not. So the odds move against you. This at least was my experience of making discretionary mods to a trendfollowing system which as you said took a lot of time and added little value.

By comparison when I trade in a discretionary manner my trades are either with the trend or with some other edge in my favour. So now I trade mechanical approaches with absolutely no interference (for max probability of success) and discretionary approaches separately (to assist in my overal portfolio of systems).

So you may well have chosen the most difficult kind of discretionary trading to test yourself on!

John

MCT
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Fri May 16, 2003 7:27 pm

Post by MCT » Wed Aug 27, 2003 10:21 pm

16 - Best parameters for money management can be determined thorugh simulation when using a mechanical approach (M+)

17 - Best parameters for money management need experience and a good hand when using a discretionary system. (D-)
I personally employ a nondiversified hybrid approach, where risk management is purely mechanical, and everything else is executed using a discretionary rule set. I trade most commodity markets. I agree with 16 and it is straight forward. I haven’t found#17 to be true. For a discretionary trader, ‘best parameters for money management’ do not necessarily require extensive experience as you put it. Simulations, similar to a mechanical approach, can be employed. The difference is setups, exits and entries would be executed in a discretionary manner using pure price action [chart patterns]. For example, if you find through historical observation/simulation that you have a 50% historical win rate, and that you have a 5:1 payoff, that is enough info to get you started working on what would be the optimal bet fraction. You can then decide how close to the cliff you would like to get.

The values for expectancy and win/loss ratios are long term averages; as time goes by the optimal bet fraction changes/lags, for this reason the approach would never be an exact science as is the kelly formula employed in black jack or the transmission of data over telephone lines. After all prices reflect human minds that succumb to emotion. To compensate for this lag/risk, some discretion is required. I employ chart patterns to setup, enter and exit positions along with a risk management model similar to Gallacher’s, for controlling exposure, allowing me to stay closer to what is happening NOW. This is by no means a complete description of my method i.e unlike most trendfollowers, I buy low and sell high. 8)

So far I’m able to maintain high expectancy along with a high win ratio. This is one aspect I seem to enjoy about discretionary methods, at least for now. :)
I mentioned this point to Seykota, and his response was, “keep on dreaming.â€

bagherra
Senior Member
Senior Member
Posts: 33
Joined: Fri Apr 25, 2003 9:03 am
Location: Charlotte, NC

Discretionary Approach

Post by bagherra » Fri Aug 29, 2003 12:11 am

Discretionary traders should benefit from a higher win ratio and lower drawdowns. Criticism of discretionary trading suggests that the discretionary trader will miss a significant number of winners.
I take it this is not your experience?

I share your scepticism about backtesting. It seems that the hurdles of getting clean data, accounting accurately for slippage, and getting the rollovers right are formidable. Might it not be better to just surf a little? At least you're living in the now.

Most discretionary traders I've read about were short term players. Do you hold long term positions?

MCT
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Fri May 16, 2003 7:27 pm

Post by MCT » Fri Aug 29, 2003 12:09 pm

The human brain is very good at recognizing patterns visually. A trader I respect ones said, “If a computer must be programmed to recognize shapes in the sky, it would be difficult to teach it the difference between a cloud, a bird and an airplane. Once you have it programmed of course, a blimp floats by and the computer is in trouble.â€

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Fri Aug 29, 2003 4:26 pm

Good post MT,

I would add that there is another reason that many short term players are discretionary. The promise of higher returns thru more frequent trades. I was on the Bund last night and between opening and 2:30pm when I went to bed there were about 6 good trades of which I successfully took 4 and plus 1 that failed. My winners averaged a little smaller than my looser (because I scale out half of my position at a target and trail the rest and most didnt trend).

Now lets say it was 1 to 1 and I risked not 2% of my equity but 1% on each trade. In that case I would be up 3% in one evening. Assuming it was repeatable thats (1.03)^200 (got to take a holiday) = 369*(Inital Capital) earned each year. My long term systems, by contrast, will satisfy me if they reach 50% and make me very happy over 100%.

Imagine if you could have taken a couple of good trends on ESTX as well :-)

John

MCT
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Fri May 16, 2003 7:27 pm

Post by MCT » Fri Aug 29, 2003 7:22 pm

Kiwi
My winners averaged a little smaller than my looser (because I scale out half of my position at a target and trail the rest and most didnt trend).
Scaling out of your exits, as I understand it goes against the golden rule of cut your losses short and let your winners run. It tends to produce large losses and meager profits. It feels right, I used to utilize such an exit until I realized the futility of what I was trying to do. In TYWTFF Van.Tharp, makes this clear,â€
Last edited by MCT on Sun Sep 21, 2003 12:17 am, edited 1 time in total.

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Fri Aug 29, 2003 8:04 pm

You are correct in Tharpian long term terms. Although I think that its a little like the pyramiding arguments and some of Vans assumptions are like the mine - wrong or lacking experience needed to have a different view ... it depends. I've seen the high frequency vs high win argument many times before and I think that there are some interesting outcomes that can be counterintuitive - see the challenge below.

There is a point to why people take profits at close targets. Mark Douglas raised it in both his books and some have said that discretionary traders who do it outperform those who don't. If they are right it is all about the (primarily psychological) issues of short term trading where clear market perception, following your strategy and good reactions are everything.

Many take this approach just because they find it a more enjoyable way to get an income. Ref the dacharts website and the discussions there. If you trade for that reason then its nice to have no losing weeks (and few losing days amazingly). Just cruise along with very few losing streaks :-)


John


PS. For those who download that Monte Carlo tool there is another interesting reason - you may just find that you have a lower risk of ruin with a high win rate system than with a low win rate system with the same expectancy ... try it and let us know (use 35% win rate for a long term single entry trend follower and a 75% win rate for a system like the one I described perhaps) :)

PPS. Thomas Stridsman advised looking for low standard deviations in trade outcomes which might also give a reason for not always following the trend to the end.

MCT
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Fri May 16, 2003 7:27 pm

Post by MCT » Fri Aug 29, 2003 11:02 pm

Volatility — far from easily being compensated for — is at the very heart of what is involved in price action. What’s critical is getting into the right trade--the very best trade--at precisely the right time, and then selling whenever the market and the rules tell you it’s time to sell. The time between entry and exit could either be long or short (a few weeks). And I don't differetiate my system either long-term or short-term, even though it seems I more of a longer term trader. I normally let the market tell me which one it is. Braod diversification, almost always, only serves to water down your returns. The best results are usually achieved through concentrated compounding over the long run.

MT

Post Reply