Keeping up with the simulation and Spreadbetting?

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.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Keeping up with the simulation and Spreadbetting?

Post by Medius »

Got my perfect systems tested now running it live using spreadbets in the UK.

As recommended elsewhere here I use the simulation to Generate orders and carefully monitor for the unexpected.

Three of four times now I've had instruments hit stops out of US time zone yet positions stay in the simulation.

Also I've had a Eurodollar position constantly have its stop hit but not occur in the simulation. Losing over 1R in the process when the simulation is showing a modest profit.

Any thoughts on how best to handle out of US time zone moves?

As for spreadbetting have I got any chance of winning here or should I give up until I can open a "proper" futures account?
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

I dont spreadbet, but have experienced similar issues as you described based on bad/erratic quotes. My solution was to quit placing my stops in the market and only use market-on-open orders based on yesterdays closing prices. I had to re-tweak my system based on these new rules. The historical performance was not as good as when I had stops, but if the stops are being triggered erratically then the better historical performance is irrelevant.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Excellent thanks Levijean, sounds a scary but very practical solution.

Is it possible to simulate the exit you describe using TB? If so how?
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

I still use stop functions for risk and position sizing calculations, I just dont use any broker calls to place the stop orders into the market. For example, instead of

Code: Select all

broker.exitallunitsonstop(stopprice) 
i use

Code: Select all

if instrument.close < stopprice then broker.exitallunitsonopen

It may seem scary to not have the stops resting with your broker in the market, but if your stop is triggered, I think it is basically a coin flip as to whether tomorrow's execution price will be above or below the stop price. This is for a long-term system.
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

Also, this makes the workload of having to manage dynamic stops for your portfolio much easier. I used to have to modify stop loss orders on 40+ positions daily - what a chore and very error prone. Now I only have to act if there is an actual trade to make.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

After the hastle of maintaining a similar number of stops what your suggesting seems almost theraputic.

Thank you.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Levijean do you do the same thing with Entry orders? i.e. wait for the simulation and enter on the the following open?
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

Yup. I did that long before I implemented it with my exit stops.

Every trade in my system(s) is a market on open order. Every entry, exit, position adjustment, etc. If a system only tests well with stops, I don't want to mess with it. Call me a lazy trader.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Lazy may be...but it sounds like youve done a lot of work, thinking and testing up front. Gives a whole new meaning to LTTF though.

I'm using TB Turtle edition at the moment so unfortunately I cant make the code changes you suggest.

I'm wondering how I can approximately simulate the trade entry and exit on the next days open in performance terms perhaps using a large slippage % currently set at 5% any thoughts?

Do you recall the approximate difference in peformance?
Last edited by Medius on Tue Oct 06, 2009 9:50 am, edited 1 time in total.
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

Sorry, I cant say. All i can suggest is get Builder. I don't know why any serious trader would settle for less.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Originally my thinking was if the system made the Turtles so rich why bother with anything complicated.

I trialed the Builder version after reading WOTT, having not seen a trading system before it was a bit overwhelming.

Now I know a little more I see what you mean.
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

I dont think a system will make anyone rich. What makes you rich is continuing to follow a system even when it becomes unbearably painful. Only by spending hundreds of hours tinkering with a system and understanding every minute detail about it is what will give you the confidence to do that.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Further thoughts ok so the stop gets triggered by a spike but then the instrument resumes trend. Do you?

a) Exit?
b) Check your out orders list and see if the instrument is still on the list and if so just revise the trailing stop to that of the new order?
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

I'm confused. I dont use stops. There is nothing to trigger.
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

Amazon dot com sells a book by Charles LeBeau and David Lucas that discusses this topic.

Their answer to questions of the form: What should I do if XYZ happens? is always the same: Follow the system. If XYZ happens, follow the system.

They suggest that if you are very concerned about some event XYZ, you can put additional rules into your system. Rules which specifically deal with XYZ events. Then you can backtest your system and see how it handled examples of XYZ events, in the past (on historical price data). Once you're happy with your system and its rules, including the XYZ rule: follow the system, says LeBeau and Lucas.

Here is a small tidbit of their thinking, taken from pages 24-25. It's amusing that they use the same terminology as The Original Turtle Trading Rules by C. F. : "Failsafe Breakout"
Attachments
LeBeau and Lucas pp. 24-25
LeBeau and Lucas pp. 24-25
L_and_L_p24.png (221.86 KiB) Viewed 10901 times
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Thanks sluggo "follow the system" is has a proven tested outcome.

Levijean, following our discussion on not using stops in the market: I had an instrument go through an exit level during US time thus exiting me from the simulation. Therefore time to exit the broker position on the open. At the UK open the coin flip as you say was positive and the instrument looked to be trending again I sold in UK time locking in a higher return than the simulation yet "following the system" a good result. However as per sluggo's post re-entry becomes a temptation as the instrument keeps trending and the generate orders still suggests buying but at a new high level.

My thought was to short circuit the system using the Donchian trailing exit.

I think I'm just going to have to buy TB Builder version a bit sooner than planned....lol.... thanks for your thought provoking comments.
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

It sounds like your stops are much closer than mine. My system will rarely get stopped out one day and re-enter in the same direction the next.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

The system hadnt but I had!

I'm using a Donchian system with a fairly hefty offset for the exit breakout. In testing, exiting on the day of the signal, not the preferred open of the following day, things seem ok. CAGR just slightly less than the MAX TEDD.

Manually running the system as Levijean suggested exiting on the following open has so far proven positive. The unknown future is acknowledged of course.
Last edited by Medius on Thu Oct 15, 2009 12:25 pm, edited 1 time in total.
Medius
Roundtable Knight
Roundtable Knight
Posts: 107
Joined: Fri May 22, 2009 5:38 am
Location: Berkshire, UK

Post by Medius »

Any trader using a buy on open strategy have a rule for executing the trade in the market like using a limit order, waiting for the US open or just buying at local (say UK) market open immediately?

Since trying to follow the simulation the exits have been good but an entry today although better than the US open was not as good as the simulations entry. As much as .4R difference.
LeviF
Roundtable Knight
Roundtable Knight
Posts: 1436
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF »

Medius wrote:Manually running the system as Levijean suggested exiting on the following open has so far proven positive.
I really didnt suggest you do this. I stated this is what I do. If you are asking for suggestions, I suggest you buy a programmable copy of Blox and test it yourself.
Post Reply