coin flip trading system

 

Hi,

Has anyone out there used or know of a trading algorithm/system that somewhere in the decision making

process flips a coin (uses a random number generator) to make some decision?

For example, in a google search I see that there are trading strategies where a coin flip is

used to get in to trades or make some other decision, i.e. how many pips to wait before getting out, etc.

As I understood it, the theory behind using random decision making at some point seems to be that humans typically falter at making certain decisions at a rate greater than 50%.

Anybody with knowledge of this kind of system?

I would greatly appreciate any information on such a system.

Thanks in advance

 
marketpirate:

Hi,

Has anyone out there used or know of a trading algorithm/system that somewhere in the decision making

process flips a coin (uses a random number generator) to make some decision?

For example, in a google search I see that there are trading strategies where a coin flip is

used to get in to trades or make some other decision, i.e. how many pips to wait before getting out, etc.

As I understood it, the theory behind using random decision making at some point seems to be that humans typically falter at making certain decisions at a rate greater than 50%.

Anybody with knowledge of this kind of system?

I would greatly appreciate any information on such a system.

Thanks in advance

It is trivial to make a system that either gets into a trade or gets out at random. You can also get LONG or SHORT at random. The problem with such a system, as written, is that it would be loss making simply due to the spread.

I did see somebody selling a "dart thrower" system, with no details given at the time. I assume it was a scam since there are more FOREX scams than the real thing by a massive factor.

 
 
robofx.org:

Check this out:

https://www.mql5.com/en/code

Ah yes, what a pile of pooh. Simulated as PF=1.13 and people still say they are trading it!

I simulated it over 2011 on EURUSD and it has a PF of 0.90, which is more in keeping with what one would expect.

 
The way I see it is that a coin flip system over several back tests would give a mean and standard deviation aganst which to backtest other EA's. So if another EA proforms significantly worse or bettter than the base line found with the statistics found with a coin flip then it is probable you could go live with it (reversing the buy sell and testing again on one that performs significantly worse). Also such data from the coin flip could be used to check for significant change in market behaviour thus warning when a working EA may be failing.
 
marketpirate:

Has anyone out there used or know of a trading algorithm/system that somewhere in the decision making process flips a coin (uses a random number generator) to make some decision?

https://www.mql5.com/en/code/9987, Was created by a friend of mine.

My Opinion: in-order to make random systems work, one need very advanced money-management(mm) and advanced trade management(tm). Those same management styles would yield slightly more on non-random signals hence puts the trader back in trading. The above only offers gamblers progressions. :)

 
dabbler:

Ah yes, what a pile of pooh. Simulated as PF=1.13 and people still say they are trading it!

I simulated it over 2011 on EURUSD and it has a PF of 0.90, which is more in keeping with what one would expect.


Sorry what does PF stand for? What is PF=1.13 mean?

Thanks.

 
Profit Factor.
 
marketpirate:


Sorry what does PF stand for? What is PF=1.13 mean?

Thanks.

PF= Profit Factor, the ratio gross profit divided by gross loss.

PF=1.0 would be break even

If you test a losing system several times you will occasionally get above break even for one set of trades. There are some (very stupid) people who think that with "advanced money management" you can take a losing system into a winning system. So, for example, you can go to Las Vegas and win at Roulette, which has a PF below 1 of course.

Any system will demonstrate some variation of profit factor over different simulation runs (different time periods) or forward tests. With a PF close to one you are therefore unsure if the system is a winner or a loser. At 1.13 it is safe to assume it is a loser.

 
dabbler:

PF= Profit Factor, the ratio gross profit divided by gross loss.

PF=1.0 would be break even

If you test a losing system several times you will occasionally get above break even for one set of trades. There are some (very stupid) people who think that with "advanced money management" you can take a losing system into a winning system. So, for example, you can go to Las Vegas and win at Roulette, which has a PF below 1 of course.

Any system will demonstrate some variation of profit factor over different simulation runs (different time periods) or forward tests. With a PF close to one you are therefore unsure if the system is a winner or a loser. At 1.13 it is safe to assume it is a loser.


ok interesting.

but i am a little confused - why is 1.13 a loser? if PF = P/L ratio = 1.13 means profits are greater than losses, NO? by the same token PF of anything less than 1 would be a loser, i would think. what did I miss?

 
marketpirate:

ok interesting.

but i am a little confused - why is 1.13 a loser? if PF = P/L ratio = 1.13 means profits are greater than losses, NO? by the same token PF of anything less than 1 would be a loser, i would think. what did I miss?

Very good question. If you could guarantee that the system had a PF of 1.13 it would indeed be a winner. The thing is that you are not sure. The number comes from a back test and it is subject to a margin of error depending on the number of tests done. A random system is known to be unprofitable just by mathematics. So when you simulate a positive result on say 100 trades you know that it is not true in the long run. If, for example, you were playing roulette and there were 36 numbers and 1 zero, the chance of RED would be 18/37. At 2:1 payout that system has a profit factor of 0.95 and you know that the majority of people will lose all their money. Yes a few will walk away with some money by luck but if they keep at it they are guaranteed failure. It's all in the statistics.

With trading the statistics depend on the spread. Random trades done with 10 pip SL and TP are much harder to win than trades with 50 pip limits simply due to the spread. As an experiment you can try an excel spread sheet simulation of trading at a random 50% win rate and see that sometimes you can win nicely over a limited run.

Reason: