Win/Loss continued

[pkfryer]

Just came accross this really handy little app!

It simulates winning and losing trades according to the parameters provided. The trade success is randomly produced acording to the average win/loss ratio and the probability of winning.

Playing around with the figures produces some really interesting results:
If win/loss = 2 (when you win you on average twice as much profit as you lose)
and if the probability of winning is 40%
the equity curve uniformly goes up... but sometimes with some hefty draw downs

However, if you change the probability to 30% you are guaranteed to lose in the long run with a win/loss of 2. Tweak it to win/loss of 3 and you are pretty much certain of it going up!

Its incredible what the slightest change to the parameters can have on your account!

Check it out and have a play... its really interesting!

http://www.financialpicks.com/simulator.html

 

[frugi]

Thanks for posting that pkfryer it's useful and fun.

Here's one of my systems!

Av win/av loss 2.39
Chance of win 37%

 

[theknifemac]

Just came accross this really handy little app!

However, if you change the probability to 30% you are guaranteed to lose in the long run with a win/loss of 2. Tweak it to win/loss of 3 and you are pretty much certain of it going up!

Its incredible what the slightest change to the parameters can have on your account!

Check it out and have a play... its really interesting!

http://www.financialpicks.com/simulator.html

Very interesting pkfryer ! The example you give shows what happens when the expectancy of the system becomes negative - that one figure shows you what you can expect from the equity curve.

 

[pkfryer]

Anyone else care to try their systems performance out? Would be interesting to see what the simulator makes of it based on your paper trading/back testing statistics?

Has anyone heard of the kelly value? The simulator seems to suggest up to 0.25 for good systems... is that 25% of your account per trade??

 

[TheBramble]

Has anyone heard of the kelly value?

pk - have a look at the pdf attached to this post by GrowlTiger.

http://www.trade2win.co.uk/boards/showthread.php?p=97671&highlight=kelly#post97671

 

[juanbyte]

Kelly's formula is simply

(2*p)-1 where P = win probability. So if you have a a system that wins 66% of the time the
calculation is thus

(2*0.66)-1 = 1.32-1 = .32

Check it out

 

[pkfryer]

The Kelly formula calculates how much to risk of your capital per trade based on past performance?

So in the above example .32 = 32%

I just got confused as the simulator suggested a good system would be most efficient trading with .25. I presume this is a theoretical optimal obviously without consideration of risk control. I noticed with some combination of parameters the draw down for some random sequences looked like it would have wiped you out even though in the long run you would have done good!

Risk management would never allow you to risk .25 or more of your capital. Or perhaps risk management should take into consideration of the performance of the system, but any performance of the system is based on back testing which only holds true the period it was tested with, not for the future.

Optimum f is something similar to Kelly formula?