Prudent Risk Management + No Edge = Positive Expectancy??

[ironchef]

Don’t try to “day trade”. Be hyper focused on uncovering opportunities regardless of the timeframe. For example, if you understand options, it might be a good idea to explore 0dte and how they interact with the broader market. More specifically, are there “anomalies” you can take advantage of?

OK thanks.

Options are my bread and butter trades and I continue to trade while learning to day trade.

Once I am done with finding a profitable day trading system, my next project is trading butterfly, the @destriero way.

Best wishes to you for Thanksgiving.

 

[ironchef]

I asked chatGPT and it gave this formula to calculate the sample size:

N= [Z^2*p*(1-p)]/E^2

Where:
N=sample size
Z=Z-score
p=probability of win
E=Margin of error
Assuming R:R=1:1 for simplicity

For win/lose of 54/46
Z=1.96 for a 95% confidence level
p=0.54
1-p=.46
E=1% if I want a margin of error of 0.01

N=9,542, I need about 10,000 trades to be absolutely sure it is real.

That is too much, so I reduce the margin of error to 5%, then

N=382.

PRM could be the edge because I don't think I am doing anything special, basically just trend following, using generic stuff that everyone on ET uses?

I now have over 500 trades, I do notice a gradual degradation of win rate but still hovers around 54%. R:R, as much as I could tell is close to 1:1.

I just have to keep running it for a few more months in small size and see if things will hold.

Update:

If I use the standard margin of error of 3% (most political polls use 3%), then the sample size to use is 540.

As of today, after 9 weeks going live (but small size), I have done 538 trades with 296 winers and 242 losers. R:R is 1:1. Within a 3% margin of error, it is a profitable system.

As far as I can tell, PRM can produce a positive expectancy in a system with no obvious edge.

 

[tiddlywinks]

Update:

If I use the standard margin of error of 3% (most political polls use 3%), then the sample size to use is 540.

As of today, after 9 weeks going live (but small size), I have done 538 trades with 296 winers and 242 losers. R:R is 1:1. Within a 3% margin of error, it is a profitable system.

As far as I can tell, PRM can produce a positive expectancy in a system with no obvious edge.

The problem is you are trying to prove the wrong thing.

If as you say "PRM can produce a positive expectancy in a system with no obvious edge" then
you should just buy or sell at will, at random, and you will be profitable over a large enough number of trades.
And everyone lives the lifestyle they want, happily ever after!

What you should be attempting to prove is your ENTRY EDGE, which apparently is not "obvious" to you.
That would involve looking at the individual criteria that goes into your trade selection
and making a case, quantitatively, for existence of that criteria in the trade selection.

Every trade for every style of trade, for every trader, for every asset, has risk that can be managed or not.
If PRM is really an edge, than risk as trade selection criteria adds nothing.

Dimiti, lobster and cracked crab for everyone. Extra primo good Mr. Coleman sir!

Happy Thanksgiving!

BTW... what does Prudent mean in the context of Prudent Risk Management?

 

[ironchef]

BTW... what does Prudent mean in the context of Prudent Risk Management?

That, you have to ask the originator, @Buy1Sell2.

 

[ironchef]

The problem is you are trying to prove the wrong thing.

If as you say "PRM can produce a positive expectancy in a system with no obvious edge" then
you should just buy or sell at will, at random, and you will be profitable over a large enough number of trades.
And everyone lives the lifestyle they want, happily ever after!

What you should be attempting to prove is your ENTRY EDGE, which apparently is not "obvious" to you.
That would involve looking at the individual criteria that goes into your trade selection
and making a case, quantitatively, for existence of that criteria in the trade selection.

Every trade for every style of trade, for every trader, for every asset, has risk that can be managed or not.
If PRM is really an edge, than risk as trade selection criteria adds nothing.

Dimiti, lobster and cracked crab for everyone. Extra primo good Mr. Coleman sir!

Happy Thanksgiving!

BTW... what does Prudent mean in the context of Prudent Risk Management?

Excellent point, a good entry will give me an edge and I should concentrate on improving that.

 

[schizo]

BTW... what does Prudent mean in the context of Prudent Risk Management?

It means "Don't be a friggin' hero!"

 

[ironchef]

It means "Don't be a friggin' hero!"

And don't be greedy.

I made one adjustment going from paper to live: Instead of letting my winners run (many into losses that upset me to no end), I now take profits aggressively and improved win rate from under 50% to over 50%.

There is no free lunch. Net profit went down significantly. But I am more comfortable with higher win rate than higher profits.

 

[ironchef]

If as you say "PRM can produce a positive expectancy in a system with no obvious edge"
If PRM is really an edge, than risk as trade selection criteria adds nothing.

1. PRM is easy to preach but extremely difficult to practice. I am convinced properly practiced, it is an edge.

2. Trade selection is accretive to PRM.

 

[smacdtrader]

I understand that. The $64,000 question is the win/lose ratio greater than 1, do I have enough sample size to definitively say it is not random.

There are simple stat tests to determine this. I would start with a bootstrap test and if promising results then try something more formal and parametric. If you want to send me the series of returns I'd be happy to run a few tests if that's not in your wheelhouse.

 

[smacdtrader]

I asked chatGPT and it gave this formula to calculate the sample size:

N= [Z^2*p*(1-p)]/E^2

Where:
N=sample size
Z=Z-score
p=probability of win
E=Margin of error
Assuming R:R=1:1 for simplicity

For win/lose of 54/46
Z=1.96 for a 95% confidence level
p=0.54
1-p=.46
E=1% if I want a margin of error of 0.01

N=9,542, I need about 10,000 trades to be absolutely sure it is real.

That is too much, so I reduce the margin of error to 5%, then

N=382.

PRM could be the edge because I don't think I am doing anything special, basically just trend following, using generic stuff that everyone on ET uses?

I now have over 500 trades, I do notice a gradual degradation of win rate but still hovers around 54%. R:R, as much as I could tell is close to 1:1.

I just have to keep running it for a few more months in small size and see if things will hold.

Yeah, you don't want you margin of error to be 5% in this case.

A very dirty way to do this is take a sample of like 100 of the returns, calculate win rate. Do this 1000 or 10000 times. If the 1:1 win rate is between the first and 3rd quartile, then it's random and you got nothing. 3 lines of code. No fancy parametric stats necessary and no more time wasted on it.