Example Trading System

Mmmh, wasn`t clear enough about the "more dynamic" MM, what I mean is to adjust the Size according to the exact stop width of the single trade, the real actual risk.

An exact 10% Fixed Ratio would be to do so:
Example:
10000,- Account
If the System tells you the actual stop would be 400,- $, 2 contracts are allowed.
1000,-(10% Acc.) / 400,- = 2 (0.5) contracts.
If the Range is smaller and the stop would be 150,- you are allowed to purchase 6 contracts.

I think a 10% would be little bit too much in this way.

Cause I`m not really an Easy Language Crack, I `m doing my simple calculation with the largest loss. I tried the real stuff but TS dosn`t like my code...

Michael

 

Moneymanagement the next Step…..

I just thought about not only adjust the betsize to the risk and accountsize, but also consider the risk/reward ratio of the single tradecondition. For sure it can easily occur some curvefitting, but if you find some very constant conditions with higher risk/reward, why not use higher betsize? Just to trade such conditions alone is not the way to good netprofit cause the more rare trades give you a better profitfactor but less money….

So, this curious condition I found, could be used in this way…

Condition8 =(VolatilityStdDev(5)of DATA2) > xaverage((VolatilityStdDev(5)of DATA2),test);

Or try any other things with Vola StdDev.

DATA2 = Daily YM


Starting point is a very basic Open Range BO Strategy with 59,8 $ Avgtrade, 12140,- $ Netprofit.

Every single # for “test” shows a better avgtrade/profitfactor than the unfiltered Strategy, for a reason I don`t know till now, I have no clue why this is working this way. But we don`t need reasons, we just need facts. So why not use a , lets say, 50 -80% higher betsize for such conditions.

First Pic shows the # of “test” every # has better avgtrade/profitfactor than the original Strat. with 59.8 $

Pic Left Bottom Stats of the Basic System.

Now a little sharp step up in betsize cause one more contract means 100% more, if condition8 = true trade 1 contract more:EL; if Condition8 then NCon = NCon + 1;
In this case I took a broader series of “test” between 5 and 10 to not rely on one single number.
Pic Right Bottom Stats with doubled betsize on the higher risk/reward condition.

So Moneymanagement could be more than any Risk% thing.

Does anybody have any clue how to compute the actual stopsize as risk in a MM Strategy in EL Tradestation 2000 I ? If I try I get a division/0 Error.

I use this simple stuff:

Input:StopSize(400), MM_Switch(1);
Vars:Equity(0), Risk(0), Ncon(0);

Equity = Startequity + NetProfit + OpenPositionProfit;
Risk = Equity/100 * PercRisk;
if MM_Switch = 0 then NCon = 1;
if MM_Switch = 1 then NCon = Risk/StopSize;
NCon = MaxList(NCon,1);

If StopSize is a fix input (400,-$ for ex.), no problem, if I compute first the actual Stop from the Openrange and take this number for Stopsize >>> Division/0 Error. I`m a somewhat stupid EL Hacker, any idea`s around? Is this possible at all?

 

Final Thoughts to the current situation of trading Openrange Breakouts in the YM

After actual testings I found following facts about the YM:

The earlier mentioned parameters like Open-Gaps, Entry-Distance from yesterday`s high/low and the size of the openrange make not much effect nowadays. In earlier tests especially on the ES the influence was remarkable. The answer is the very low Vola, as you can see in one of the Pics “Death of Volatility”, we have a complete absence of 3% days in the S&P, there are also not even 2% days and not very much 1% days. Of course there are consequently not too big Openranges, nor large Gaps and the distance from last day`s high/low is the same thing. In a coming higher Vola market such parameter should checked again, esp. in the ES.

In current market I observed following:

Shorter Openranges 60-75 Min are not very profitabel, but you can trade it if you wait 120 Minutes after open before your Stops getting active, that does not mean to jump in market if the price is already flying. Prices have to be under(for long) your Stop.


Condition1 = time > 1130 and close[1] < ChannelHigh + Threshold;{for long}
Condition2 = time > 1130 and close[1] > ChannelLow - Threshold; {for short}

Better are the first 105-120 Minutes.

It`s far better not to trade Breakouts in the Lunchhour, a Breakout here is often just a Fake.
It`s far better not to trade anymore after a Breakout in the Lunchhour.
If still flat take Trades not very later than 0230 pm(last 105 Min.)
Don`t trade on Mondays.

That`s the way the YM behaves at the time…….

Good trading

 

Any updates to current results on the website any time soon?

Thanks.

 

just posted this to an et index futures thread. relevant here as well though, I think.

--------------

if one develops a sustainable edge in orb trading, i would think it makes sense to test/deploy it across other markets. have you backtested your stragegy on electronic futures in other sectors? i.e.

fi - bunds
energy - brent
fx - eurusd
softs - london cocoa
metals - gold

that said, any theories why ES and NQ perform poorly? would the same reasoning preclude trading futures in other sectors, as suggested above? i unfortunately don't have the data to find out.

AM

 

AM,
I would also be interested in some tests in other markets esp. the bonds, but also don`t have data to do. Would be of only academic use the next time, cause i have not the accountsize to trade a portfolio of markets, that`s the reason i`m reduced on just the YM. If you go through the literature you find Openrange Breakout a very old principle used in Commodities in the first place, so i have no doubt to work in those markets you mentioned.

The ES is simply the most noisiest market to choose with a lot of chop and retracements, this in every timeframe. I did some tests about the "trendiness" of common indices like Dow, German DAX, and S&P over some years of sample and the worst is the S&P. If you ever tried to develop a daily trendfollower on indices you get the same results, S&P does not work, on the other hand, i don`t like countertrend systems, but the best results you get in the S&P with so called Overbought/sold systems. The less noise, the more "trendy" a market behave, the better i also expect an Openrangebreakoutsystem to work.

The NQ behave`s more trendy on daily base, daily trendfollowing works, but on intraday timeframes may be also to much chop and noise, i did not look to deep into the reasons why exactly...I`m quite satisfied with the knowledge that the YM and ER are the better choice, but in the case of changing marketbehavior i like to see at least a weak positive performance in the other indices too.

So, a good idea is some kind of analysis about the noiselevel or trendiness of a chosen market over a larger sample to define if this market is a good or bad choice for the princible of the used system.

You can use an afternoon breakout instead or you trade a break of the Openrange only in the afternoon, i just found the most netprofit ( of my YM system )was made in the timeframe between 01.30 and 02.40 for entries, the other times plays a more marginal rule, the reason may be the higher activity of "professionell money" in the afternoon, you find trends more consistent after Lunch, but that me be not the same in other markets like FX, Bonds etc.

I think some "intelligent" kind of breakoutsystem will behave most robust over most markets over time, i`m completely off those countertrend stuff, that`s something for quick scalpers or the guys with the very deep pockets.

Uuh, if anybody has data from ZN,ZB for me I would enjoy to public some tests...

Michael

 

Has anyone looked at holding a chunk of a winning trade overnight? i.e. 50% closed out MOC, 50% closed out on the next day's open. Might be interesting to find out. Might need some extra filters/rules though.

AM

 

Hi AM,
did some work on the stuff, could not find a working rule for holding till next open, erratic results, but i got more ideas meanwhile and keep on testing...
here some may be interesting work about the 135 min breakout stuff i put on my server.
http://trader.gmxhome.de/breakout/breakouttest.htm

 

...to confuse you guys completely, the shape of the openrange should not ignored....
The testparameter could be:

Where is the openprice in the range? upper, lower, middle...

How often did the price crosses over the openprice?

Where did the price close at end of the 135 min? Near high/low, upper, lower, middle ?

In Case you get a Long Signal for ex., i guess there is quite a difference in the statistical outcome of the trade in those 4 examples of openranges in the picture below.

(I didn`t test till now, coming soon...)

first we see a very neutral/choppy range, i guess it`s ok to take any side.

second we see a nice trend to the upper side, closing at the high of the range, i guess a immidiately long could be a mistake, i guess some exhaustion to occur, may be wait for a second break of the range later.

third, looks very bullish, open in the lower range, uptrend, close in the upper range, don`t know, may be there will be a afternoon reversal at least, just the tests will tell...

fourth, looks bearish, open in the upper range, some kind of downtrend, a long after lunch could be nice, cause you can expect some shortsqeeze, just my opinion...

Just taking the break of the openrange regardless of the marketaction is far to simple, just my 2 cents...

 

Smart thinking.

I think these sort of insights could constitute part of a scoring system that generates daily bet sizes.

Building on your thoughts, another lens to look at the OR with comes to mind. Where is the OR in relation to yesterday's data? Above the high, above yesterday's close, below yesterday's close, below yesterday's low.

Has anyone looked at this sort of information? (Sorry I'm only come up with suggestions - data issues dictate this.)

Cheers,
AM