Compounding=Bad, now what?

New comment
 

So I've read many discussions that all conclude that compounding your wins is bad, due to the risk/reward equity ratio etc. etc.

But my question is, now what??

Obviously if someone wins more they'll want to put that back in their winning system, but how do they do that without compounding?

I am confused, what do you guys do?

 

Don't mind qjol.

Is there anyone with a higher IQ who could help me?

Thanks!

 

In my opinion moneymanagement needs to be customized to each strategy. It strongly depends how your win/losses are distributed. there was a nice article here on mql, a while ago.

if you expect your system to win 5-10 trades in a row after a loosing period, then you maybe could increase lotsize in the beginning of each "winnig cycle" more then at the end. i think most "enemys" of compounding are refering to strictly recalculate the lots after each trade...

Maybe you can also give Kelliy's Criterion a shot?


//z

 

Thanks Z for the info, that helps.

Because isn't risking 2% of your capital a form of compounding anyway?? I never got why ppl said not to compound when that is exactly what risking 2% does..........?

 
Who told you that "compounding is bad"? Such a general statement cannot be made without exactly defining the context and the mechanism by which these bad things are supposed to happen or even defining what exactly they actually mean when they refer to the term "compounding", otherwise it is completely meaningless. There is nothing bad with the concept of compounding per se.
 

Is that what you use, just 2% of your free equity for each order?

Like:

double risk=CapitalRisk*AccountEquity();
lots = (risk / StopLoss) / MarketInfo(Symbol(),MODE_TICKVALUE);

 
something like that. There might be a more complicated algorithm, involving the expected winning rate or other properties of the trade or a sequence of trades but the essence is always the same: The more you have the bigger you can (and will) trade. I mean its obvious: Today I bet the amount x each time and once I have more money to trade I will obviously make bigger bets. Its my intention to make more money and not just play forex for fun.
 
aku11a:

Is that what you use, just 2% of your free equity for each order?

Like:

double risk=CapitalRisk*AccountEquity();
lots = (risk / StopLoss) / MarketInfo(Symbol(),MODE_TICKVALUE);


You may find the code I uploaded here to be helpful when using your "risk capital" (aka "equity at risk") to determine the position size to open based on stoploss price and the specific currency pair you are trading (it matters).

https://www.mql5.com/en/forum/127798/page2#356736

I actually have a simpler to use include file that I can upload but its not on this computer and it could take me a day to get to it.

As for your thread title, the "conventional wisdom" is that you do NOT want compounding when doing backtesting as that convolutes the statistics you probably want to perform as part of the backtesting itself but you DO want compounding when doing forward/live trading.

The saying goes "backtesting for risk and profit measurement, forward testing for risk and profit management". Measurement requires robust and stringent application of relevant statistical analyses procedures. Management requires robust and stringent application of relevant money management procedures.

The two are not the same, although they are often designed with each other in mind.
 

RE: The saying goes "backtesting for risk and profit measurement, forward testing for risk and profit management". Measurement requires robust and stringent application of relevant statistical analyses procedures. Management requires robust and stringent application of relevant money management procedures.
.........

Thanks 1005, I know this thread isn't really titled as such, but now you have me interested in what you are saying, can you tell me more about how you implement Measurement VS Managment?? Like I have backtested my EAs over the 9-11 period to see if they would have failed totally or weathered the time, is this what you mean by "stringent application of relevant statistical analyses procedures"??

New comment
Reason: