how much live trading is relevant?

Alot of people say that backtesting/backtested results are pointless and that trading the sytem in realtime with real money is the only way?

So Ive been trading a new method LIVE using very small stakes through a spreadbetting firm for a few months now.

Its a short term trading method. It just uses price action, based on my experience of watching charts/trading for a few years now as a source of extra income. I track every trade in excel.

It makes about 15 trades per day, and ive been trading it whilst keeping resultsin excel for about 1 month, so that's about 300 trades so far?
Stops are always very tight (about 10 pips) and targets are always slightly bigger

So far its up about 550 pips as of today. Nothing magnificent but its results are very consistent.

How long would you want to see reults like this until you were confident enough to start trading 'real' stakes (E.G futures contractw rather than spreadbet trades which can be as low as the equivilant of a 10th of a contract)

thanks

 

Don't listen to what 'people' say, test everuthing yourself.

 

A lot of people are total idiots, obviously.

There are VERY good reasons to backtest anything but the most simple trading system - to make sure the system behaves as you think it should. Systems are programming, programming has errors.

So, make a 6 month backtest, go through every trade, ignore profit/losss.... but look on whether the trade was valid. Basically... did the system trade as it should? Or do you have a programming error.

This, at least, is 100% valid backtesting. And will save you some money from simple programming mistakes.

 

My view?

A history of at least 60 trades starts to give some "statistical comfort" that the behaviour you are seeing is a measure of the system's performance, rather than just a statistically insignificant sequence of random outcomes. I am sure that many traders will say 60 trades is not enough, and I would be interested to see what others think...

Another point on this .... again it is my view that important aspects of market behaviour have changed lots, and several times, over the last 18 - 24 months. As a result, it may be hard to find a system that has worked consistently well for the whole period. Therefore, I think it can be counterproductive sometimes to backtest over say the last 6 months, or 12 months, or 18 months, if it is your view that some important aspect of market behaviour related to your system might have changed totally over that time. Again, this is just another way to justify shorter backtesting periods rather than longer ones ...

And finally (I promise !), given that most systems will work for a period and then stop working, if you find it's working after 60 trades then in my view you should get cracking and put it into the market as soon as you can, because at some point possibly soon it's going to stop working again, and it would be a shame if all of the good run had been spent backtesting rather than making money!

There, that's three argument in favor of shorter backtests ...

I await the flaming ...

 

Not from me. Makes a lot of sense.

I epersonally find that all software on the market is terrible in regards to categorizing and analyzing the backtest results, sadly. Gets condensed WAY too much without analysis functions.

 

As the above posters explain things change.

If you can keep pace with the change with an adaptive system then I should be ordering up those Ferrari catalogues.

The changes are like a virus trying to survive annhilation by medics. I doubt you will find all it's twists and turns but ..........

PS if you do !! hint hint I'm here

 

thanks for that reply. That's more the kind of answer I was 'looking for'. I know some people say that you need a gazillion trades over decades before you can put faith in something,but because mt 'method' is very short term it naturally makes quite a few trades and I believe that only a few months consistency is likely statistically significant.



However, when you talk about how 'most systems will work for a period and then stop working so i should get cracking', I find that kind of bizarre.

Is one supposed to be endlessly designing new systems for when their current method stops working? that just seems wrong to me? You could be spending months and months working on something new and all that time not be earning any money whilst you hope to find your next working system??

Also, there are surely only so many 'ideas' one can come up with!?

Doesn't anyone here have a method that they have never had to change?

thanks again!

 

To quote Victor Niederhoffer (some might say it foolish to quote him given how many fortunes he's lost but anyway ...) in ‘The Education of a Speculator’ “The cycles change ... results that appeared significant in one period had a tendency to evaporate in subsequent periods. The reasons are legion. If a phenomenon truly exists, shrewd operators discover it and start anticipating it in following periods, thereby evening out the moves. It’s much more likely, however, that a peculiar constellation of factors is in effect at the time and is causing the phenomenon to occur”.

 

But isn't there such a thing as someone just 'being a good trader' (not saying that I am) and who just trades the market as they see it intraday using whatever method they use, who never has to change anything?

I welcome the views from traders here who dont believe that you have to jump from 'method to method' and who have been consistent for a long time.

I never see threads titles of 'my system has stopped working' and the like.

I imagine that the only time a system 'stops working' is if the system is like: "when the 45 ema crosses the 10 ema and the RSI is under 12 then wait for a volume spike and then look for the CCI to cross the zero line and get long" etc etc....?.....

 

It depends on your definition of backtesting. Backtesting is a tool, not a guarantee of success.

If you just ran various ideas for awhile and focused on the best performers, then you are cherry picking

If your backtest is a few months, then it is means little. It needs to be tested across multiple market conditions

If you separated your backtest data, found something that works with a large chunk of trades, and then then used the same concept on another chunk of trades and it still worked, that is interesting.

But any backtesting is useless unless you WALK FORWARD tested it for awhile, on "live" data, and you avoided any kind of data snooping or biasing. And it STILL worked. Without cherry picking, again.

Then, you <i>might</i> have something worth trading. but it is no guarantee. Many backtest, few make longterm lucrative money.

So, live trading is very relevant.