Portfolio Companion Thread

Well that's the question... If I have any hopes of trading professionally and raising the kind of money it takes to really make it worth while, a methodology has to survive for five years or more. We both know that is a lot easier said than done. When the model was initially developed I was well aware of it's limitations and internal biases so I had addressed those through different filtering processes. It 'APPEARS' it held up over time, at least in back-testing from the point we were forced to halt trading. That's worth about a four pack of Charmin TP. The real test will be when I get money back into the market and am back in the water swimming with the sharks.

With that said the models algorithm will be updated; I have more quality data this time around and I think (hope) the testing process has improved. There is a large section of data I've left untouched so I can initiate a step forward test; I'm always concerned that in the large back-testing batches I've somehow injected a forward looking bias that skews the results to the positive. I plan on stepping through about a years worth of trades, then dumping that same data through the back test process and see how close they are. If there are differences I'll start hunting and will rework the back-test (again - for the 1000th time). It's amazing how easy it is to inject unwanted bias into the testing process. Sometimes I look back and wonder how I ever made money the first time around LOL. Hopefully it speaks to the filtering process that takes the model off-line at specific intervals... Algorithm efficiency has been a huge part of my research and work.

i've never come across a methodology that works for more than 1-2 years consistently without tweaking it as time goes by.

Usually the system slowly degrades in terms of profitability, until you lose your edge.

i see that you make some very advanced tests compared to me, so i'm wondering - for a system to work consistently for 5 years in a row, you're probably looking for anomalies that happen less than 1% of the time?

as to the anomalies you're after, i hope you don't mind me asking but can you give an example for one? i would just love to know how a professional trader assesses the price movement.

I can't give you an honest clear example without giving up a lot of my hard earned knowledge! So I won't bore you with a dance around the bush. Sorry...

Please keep in mind it's purely speculation that the model worked for five years; I had to back test up to the old results. Who knows what would have happened in real time...

If you find you have to tweak your system after one or two years, you may have over-optimized. That's not a bad thing if you're just trading for your self. If you're trading for others you have a certain responsibility and need to be consistent. Changing parameters can affect their return correlation, and that can impact your funding.

To be consistent sometimes we have to dial back the optimization, use larger data-sets and accept lower, but hopefully more consistent, returns over the long haul. It depends on what you'r goals are:

Do you want to generated 20% a year on a $100,000 portfolio ($20,000); or
Do you want to collect a 20% performance fee from a 9% return on a $50m portfolio ($900,000).

Of course in the last option there is a lot that goes into that. Research, Licensing , Compliance, Admin, Operations... blah blah blah. So that the end of the day maybe you only keep 65% or so of your fees, but you can see the up side.

It has always been my position, if you are honestly good at trading (in the very broad sense of what it mean) it's more advantageous to manage funds than try and generate unrealistic individual returns.

fxanalytics, i totally understand. you know i had to try <img src='' alt='wink'/>

i'm sure many here will disagree with your last statement - it's not always better to trade OPM. you forget about the added pressure to your trading, and i'm talking from experience; there's a huge mental change when you trade for others than yourself.

now of course it all depends on the money you can raise - if that's million, then yes i would say it's worth it.

anyway if your system from several years ago is still working, than congrats - i didn't think it was possible to achieve this.

what kind of trading style can one expect with your system? scalping? short/long term trades?

people who can argue the last statement are those who come to trading with significant portfolio's... Typically it's money made elsewhere. On average, people are not going turn small portfolio's into large ones in just a few years, forget about just a few months.

If you want to trade for a living you need a LOT of cash up front. Not every model is going to make money every year... You need to either be working and making income elsewhere or you need to have enough cash to survive the down times. It's very difficult. The great thing about trading is, anyone can choose the path they want... The downside is, rarely does it work.

In my approach trades last two days to a week just depending on how things shape up; down time between trades can be one to three weeks. It's worse in the summer when things to loose there steam. If I were trading for myself I would just dial into a shorter time series but I don't want to fiddle with the dynamics of the model and potentially upset someones allocation.

On a different note and for reasons I'm not at liberty to discuss, this will be my last post. I didn't really expect anyone to take notice of my thread but I've enjoyed the discussion! Good luck with all your efforts biz!