Automated trading systems cannot work. Discuss!

[Operandi]

What are your thoughts on having an automated system that trades as a sort of framework from within which you trade discretionary? My system often trades well but sometimes there are points when I think, ah that trade just shouldn't be taken or should be closed but I can't meddle with it for fear of the strategy getting out of align with where it should be.

Excellent point, DangerBoy.

Having the strategy enter trades, but then invoke the ATM pop up is one of the best solutions as you can intervene in the trade, but yet benefit from the strategy being vigilent in not missing ideal set ups.

We have developed quite a few semi automated systems like this for our customers.

[ctrlbrk]

Excellent point, DangerBoy.

Having the strategy enter trades, but then invoke the ATM pop up is one of the best solutions as you can intervene in the trade, but yet benefit from the strategy being vigilent in not missing ideal set ups.

We have developed quite a few semi automated systems like this for our customers.

Great observation Operandi,

I do find this sort of thing useful. Having a smarter ATM so I can enter a discretionary trade, but it has smarter elements for the trailing stop or when to exit the trade.

Mike

[ctrlbrk]

In my search for adapting systems I found Clayburg, you may find him of use.

Check this video out by Clayburg. It talks about selective optimization vs. curve fitting using out of sample data.

http://www.clayburg.com/UniversalSystem/selective_opt_only/selective_opt_only.html

Also here is a free universal clone, if you are unlucky enough to have tradestation. lol!

http://eminiedges.com/Emini_Edges/Emini_Edges.html

Mike

[DangerBoy]

Thanks for those links ctrlbrk, definitely interesting stuff, I think I need to do a bit more research down this avenue before I can attest what is and what isn't possible, I guess I've held back until now as I am in early days and don't want to start off on the wrong foot by meddling in live systems as you end up invalidating your own research prior. Operandi thanks for pointing me down the ATM ally, will have a look at that in my searches.

On a side note the Wait until flat option is something I have spent quite a bit of time with funnily enough as my current strategy contrary to the normal way of things is an always in the market strategy trading back and forward between reversal signals.

I absolutely have to use the wait until flat option also as it contains various points in which it uses market orders to stop and reverse, if this option isn't turned on well there is a big problem right off the bat should it have determined that a SAR happened half an hour prior to switch on

I basically use Wait until Flat and then every morning at 6:00am I reset my strategy so that all parameters are blank, null etc. and my strat actually is flat, then it starts trading from 7:00am and shuts down (doesn't take any new signals from 16:00pm) but will only trade manage anything outstanding until the next day at which point it resets again etc etc.

The issue I still have at present though is trying to start it later than 7:00am for whatever reason but still working on this.

anyway, enough jabber, ctrlbrk, I hope you decide to come back to the dark side at some point... keep the dream alive... Lol

[tortexal]

I only trade automated systems based on mathematics not technical indicators. And i trade the same 4 funds repeatedly.

Stop after your targets are hit
back testing is important so be sure to do it
dont rely on backtested results for P&L, use it to make sure the strategy code executes properly
i build models based on current (rolling 20 day) market conditions and make modifications to calculations whenever the current math will work against me
current calculations are fed into a monte carlo simulator so that i know the probability of gains and losses based on 100,000 random scenarios as generated from the rolling 20 day market conditions.
once probability and P&L is established, expectations are also calculated as it is one thing to "average" 80% success but if the losses are larger than the gains the math is wrong and needs to be fixed.
The other thing is that i never interrupt the strategy no matter what. If i do, it will no longer adhere to the forecasted performance results.

hopefully that will give you some ideas

[Operandi]

Tortexal, that is awesome.

I bet yours is one of the most successful automated systems on this board.

[DangerBoy]

"i build models based on current (rolling 20 day) market conditions and make modifications to calculations whenever the current math will work against me"

Thanks for posting, I recently have started thinking about this particular element in my strats and am coming to the conclusion that trying to find a general rule that fits all is not something that can be done, so was starting to look down the path of active market analysis on a rolling basis and having the strat trade accordingly. Your comments have confirmed my thoughts that it's a worth investigating. Thanks

Cheers
Ross

[ctrlbrk]

tortexal,

Regarding the monte carlo simulator, did you custom build your own app for this? Or how are you accomplishing this?

Very interested in that idea.

Mike

[tortexal]

danger-
as far as rolling period, you have to keep the math current as market conditions will never stay the same. every quant fund that i know of constantly recalculates to evaluate conditions.

operandi-
idk about successful as in profitable, im sure someone out there has something better. but it's probably one of the most predictable since no event/gain/loss comes as a surprise and is factored into the math. any losses are actually welcomed bc it will statistically bring me closer to a gain.

ctrlbrk-
you can use crystal ball for excel but the # of events will be limited due to excel
best bet is matlab it is extremely fast and powerful for all types of analysis for many industries.

before i discovered NT, i was running the strategy out of excel and trading manually based off of the sheet (it had a realtime data feed). NT was the only platform that i saw that worked w/ IB and was totally flexible code wise bc its c# and you can do whatever you want. So i basically automated my excel sheet.

[NinjaTrader_Dierk]

Guys, in case you would consider investing time in a Monte Carlo Simulator: NT7 will come with a Monte Carlo Simulator. I suggest checking it out before investing time in a custom development. We will provide details as we roll first public beta.

[ctrlbrk]

Guys, in case you would consider investing time in a Monte Carlo Simulator: NT7 will come with a Monte Carlo Simulator. I suggest checking it out before investing time in a custom development. We will provide details as we roll first public beta.

You just made my day!

Mike

[Elliott Wave]

Yes, very nice news indeed.

Likewise all of these NT7 details as of late are giving me hope in the June 30th deadline.

[Trader.Jon]

Dierk,

That is good news. Looking forward to the beta release!

TJ

[DangerBoy]

Tortexal,

If your still on this thread i meant to ask you this last time and forgot about it, I've asked this before on the forum but didn't really get an answer, your very first point of "Stop once your targets are hit"

Would you mind to ellaborate on the thinking behind this?

I have been mulling over this for some time and cannot think how this can be of benefit to a strategy. My thoughts are that the idea is to try and reduce the number of scenarios that your strat may hit over the long run and so delay the probability of a sequence of events that would cause a max draw down that would knock you out the market... is this the thinking?

Thanks
Ross

[tortexal]

i trade leveraged etfs and seek to capture the majority of the movement either long or short. because they are an index, once the main movement for the day has taken place, it is very unlikely for it to reverse. The reason is that all the money that went into (or out of) whatever stocks it tracks would have to literally leave completely, and in roughly the same stocks, inorder for the direction to change and go back to where it came from.

the algorythms look at current price vs where price "should be" in order for probability of the projected end price to be high enough to enter a position. Once in, price is still monitored to ensure that price vs where price "should be" is still providing ideal probability, it not it exits/stops out and will wait again

aside from that, the position is just exited in 3 stages to cover comishions and take profits 3rd stage it the final target. it then stops completely bc the chances of a counter movement with any type of signifigant magnitude is minimal and not worth chasing after.

thats the idea behind the main primary strat that i run long/short. I run another that is similar but specifically looks for "black swan" events that seem to happen a few times a month. that strategy does what you said as far as limits how often it goes in since its just looking for the probability of a styystical anomoly to be high