Do your trading systems have to work 100%..

[ragzROI]

of the time?

Let me clarify. Im not talking win-rate. But when developing a trading system, does it have to show positive results for all equities you test it for?

Ive been developing a system, coding a compiler, and have tweaked a small figure that I found was a bit of a bug. Im testing all equities for a period of 5 years, although I've dabbled with 10-15-and 20 yr backtest, but it really bogs down the system with the complexity of the program. Before across the 16 or so equities I tested it for, it showed stellar results for all 16.

Now, it shows great results for all but 3 (with a slight loss in equity after 5 years in those 3)

Is this bad? Would you chuck it?

My question is when backtesting/developing a system, when backtested does it have to be (in my case) 16 for 16 across all equities it tested for? What would you do in my case?

 

[sainthake]

From what I know and learned personally is that there is no system that works 100%. All systems and strategies have their flaws, this is why I would never consider fully automated trading. Rather I take the signals that meet the required conditions in order for it to be a 'good' trade, but even so it can still be wrong. From then on it becomes discretionary from experience with 'said' system.

Doesn't necessarily have to work with all pairs. Find the one it works best on and milk it dry.

 

[barjon]

Many individual equities have individual characteristics in the way they move which may, or may not, be conducive to the method you are using. To complicate matters, they do not necessarily display those characteristics for all time. To complicate matters even further, those characteristics are often overshadowed by how the overall market is progressing.

I trade equities from the FTSE100 and I have a stable of 15 or so favourites that suit my method best. It is not a static stable though. I also have a blacklist of those I am extremely reluctant to trade since they have rarely displayed the characteristics that suit me.

Hope this helps answer your question.

 

[the hare]

If your going to diversify I can't see the problem. If your going to trade the system on one instrument, that might be problematic

However, I'm not really the best person to advise on this as Ive come to realize, I know nothing !

 

[numbertea]

I have recently been noticing that my system ALWAYS has the greatest probability of highest win loss ratios when I am trading indexes or futures markets vs trading individual stocks, BUT I also have the biggest winning trades with individual stocks. I like consistency so I stick to the indexes now. I guess it is a question of the roller coaster ride. Lately my neck has been hurting looking up so high at the current charts so I think I will go for slow and steady.

Cheers

 

[big tymer]

do the other stocks show a terrible loss? if it is just a small loss or near breakeven then keep it. not everything will work 100%. in trading there are losses, sure as shi*, and as soon as you start trading it with real money everything that worked 100% of the time isn't working anymore and the ones the that tested lousy all of a sudden work. that's the markets at times.

 

[Splitlink]

However, I'm not really the best person to advise on this as Ive come to realize, I know nothing !

At last! Yippee! Join the club!

 

[NVP]

of the time?

Let me clarify. Im not talking win-rate. But when developing a trading system, does it have to show positive results for all equities you test it for?

Ive been developing a system, coding a compiler, and have tweaked a small figure that I found was a bit of a bug. Im testing all equities for a period of 5 years, although I've dabbled with 10-15-and 20 yr backtest, but it really bogs down the system with the complexity of the program. Before across the 16 or so equities I tested it for, it showed stellar results for all 16.

Now, it shows great results for all but 3 (with a slight loss in equity after 5 years in those 3)

Is this bad? Would you chuck it?

My question is when backtesting/developing a system, when backtested does it have to be (in my case) 16 for 16 across all equities it tested for? What would you do in my case?

Depends how correlated the Equities are you are testing......funnily enough the ones that were losers wont correlate to the winning ones

how does it perform on the stock indexes ?

N

 

[Jack Francisco]

I think, there is no trading system that can generate 100% profit or safe.
And if anyone says that his trading strategy is 100% safe then he do not know about that and he can say it for the advertisement.

 

[graydrake]

of the time?

Let me clarify. Im not talking win-rate. But when developing a trading system, does it have to show positive results for all equities you test it for?

Ive been developing a system, coding a compiler, and have tweaked a small figure that I found was a bit of a bug. Im testing all equities for a period of 5 years, although I've dabbled with 10-15-and 20 yr backtest, but it really bogs down the system with the complexity of the program. Before across the 16 or so equities I tested it for, it showed stellar results for all 16.

Now, it shows great results for all but 3 (with a slight loss in equity after 5 years in those 3)

Is this bad? Would you chuck it?

My question is when backtesting/developing a system, when backtested does it have to be (in my case) 16 for 16 across all equities it tested for? What would you do in my case?

Tradomg

 

[perox]

I don't think that any trading system in the world wok 100% perfectly. Even you can earn a good income with 50% or less winning system by managing risk correctly.

 

[graydrake]

I do not know why my name is listed above, I do not find a response that I have drafted.

A trader that has entered 1000 trades under a defined system that has delivered a profit that meets his/her expectations, moves forward with the absolute confidence that the next series of trases the same rules will deliver the same approximate corresponding profit. I did not say the next trade, but the next series of trades. If these 1000 trades returned 42% annually on the portfolio, the trader has confidence his average future trade using the same set of rules will provide the same sort of return.

If you have no significant history of trade performance againsts a rigid set of rules, you can not prejudge your future level of success - of one trade or 100 trades.

The first and most important rule of trading is to maintain a record of past trades that have met your entry/exit rulles - EXACTLY. If you have no such data then you hould have the same level of expectation as yoi have at the casino.

Drake