Which system is better?

[dom993]

Hi Dom993,

Very interesting study, indeed.

I don't understand why the maximum drawdown gets bigger when you stop trading if the equity curve is below a certain moving average, though ?!?

It should be the opposite, no? After all, you are preventing major drawdowns from ever happening in the first place

It doesn't prevent chop when the equity curve bounces around the MA, that's why. (think about it: you miss the win(s) that gets the P&L curve above the MA, but you get the losses that bring it back under)

I also tried using MA bands, Donchian channels ... these improve things a bit, but over the long run there are enough of these MA or channels crossing to impact the P&L and worsen the max DD.

A P&L curve isn't better behaved than price on a chart, it is actually way worse as you cannot find any statistically valid pattern in it.

But nothing can substitute to one's own experience, so use the MonteCarlo spreadhset I shared, put your own system(s) trade distribution in it, and try to trade that P&L curve for yourself

 

[dom993]

On the attached, I voluntarily degraded that CL151 performance by using tight stops & targets (20-ticks/20-ticks), for the sake of illustrating what can happen when your edge starts failing.

Remember, there are about 1800 trades included in that P&L curve.

 

[StarDust9182]

Is there a formula that can figure out what the maximum number of consecutive losses your system can have?

Assuming you are talking about Handle (BTW I would bet with Handle and against most new people on this site), isn't it his maximum capital divided by 2 times the position size on each average downs?

It is my understanding that Handle has a limit of doubling down (x times) anyways since he talks about 20% losses in his math of "scalping". If he really doubled down forever, then his formulas for losses should be 1% or something less. However, that 1% would be a killer!

One can calculate the math for sequences in a sequence of random numbers and use that to determine the probability of wiping out one's capital. If you do that then the 20% tells you his most probable "x" in x times.

 

[xelite777]

It doesn't prevent chop when the equity curve bounces around the MA, that's why. (think about it: you miss the win(s) that gets the P&L curve above the MA, but you get the losses that bring it back under)

I also tried using MA bands, Donchian channels ... these improve things a bit, but over the long run there are enough of these MA or channels crossing to impact the P&L and worsen the max DD.

A P&L curve isn't better behaved than price on a chart, it is actually way worse as you cannot find any statistically valid pattern in it.

But nothing can substitute to one's own experience, so use the MonteCarlo spreadhset I shared, put your own system(s) trade distribution in it, and try to trade that P&L curve for yourself

What if the trader trades only when the average monthly drawdown of his (profitable) system has been reached (on paper), will he make more risk-adjusted profit that way?

And more importantly, will the maximum drawdown now diminish by the same token?

Thank you Dom993 from Ottawa (lovely Canadian city by the way).

 

[dom993]

What if the trader trades only when the average monthly drawdown of his (profitable) system has been reached (on paper), will he make more risk-adjusted profit that way?

And more importantly, will the maximum drawdown now diminish by the same token?

Thank you Dom993 from Ottawa (lovely Canadian city by the way).

You're welcome.

Quoting myself: "nothing can substitute to one's own experience, so use the MonteCarlo spreadhset I shared, put your own system(s) trade distribution in it, and try to trade that P&L curve for yourself".

When you are trading a profitable system, you shouldn't fear the losing trades ... you should fear missing the winning ones, these are the only one that really matter.

Sometime, someday, your system (my systems, any system) will go on a losing streak beyond your system stop, no matter how big it is. This is a statistical certainty. So the day you start trading your system, you have to accept that idea. This system stop is your guaranteed downside. But until then, your job is to profit with the system. And the only way is to take every single trade, no matter how it looks like, no matter what the P&L curve looks like at this point in time, because the outcome of that trade couldn't care less about these things: it is a random event, as any other individual trade. You have to take them ALL to make money in the long run.

 

[fjpenney]

There's usually a "yes but ...".

Yes to what you just said but just because a system has worked for the past X years doesn't mean it will work forever. I suspect you are well aware of how to use a Monte Carlo analysis to determine if the past number of trades fall within the expected range.

No system provides a very smooth equity curve like most of us would love. Returns will be lumpy. However, a trader must be prepared to accept that a system is no longer in sync with the market and either smaller position sizes should be taken or the system shouldn't be traded. You can still monitor the system to determine if it gets back in sync with the underlying market. There are countless examples of systems that worked for over ten years but then were no longer in sync with the market and failed. That type of failure will happen again and traders must know how to determine when poor performance is beyond what is expected to happen.

 

[dom993]

There's usually a "yes but ...".

Yes to what you just said but just because a system has worked for the past X years doesn't mean it will work forever. I suspect you are well aware of how to use a Monte Carlo analysis to determine if the past number of trades fall within the expected range.

No system provides a very smooth equity curve like most of us would love. Returns will be lumpy. However, a trader must be prepared to accept that a system is no longer in sync with the market and either smaller position sizes should be taken or the system shouldn't be traded. You can still monitor the system to determine if it gets back in sync with the underlying market. There are countless examples of systems that worked for over ten years but then were no longer in sync with the market and failed. That type of failure will happen again and traders must know how to determine when poor performance is beyond what is expected to happen.

Re-read what I wrote ... it is a statistical certainty that any system will fail, system failure defined by drawdown exceeding the predefined system stop.

Yes, it is up to the trader to decide if he wants to trade the system or not. But let me tell you that a great system will still have drawdowns, and many of these will be impossible to decipher.

Attached are the performance & P&L curve for CLAlwaysIn v51. About 6000 trades in 7 years, plus 200 so far this year. Pretty decent Sharpe ratio and reasonably smooth equity curve.

The last chart is a 200-trades average of the trade P&L. Like any average, it lags the "instant" average, in this case by 100 trades. It shows how a profitable system's performance can vary in time, without having any kind of system failure.

All systems do that. All the time. And sometime, someday, they fail. That's trading life.

Make more money with the system than what you'll lose when it fails, that's the real goal. And trade several systems.

 

[Sergio77]

You will only know after the fact but in general higher win rate if R:R is 1 the better it is.

 

[cornholio]

cornholio:

The attached is an Excel template, working with the YASAI free MonteCarlo add-on for Excel (http://www.yasai.rutgers.edu/download.html).

It is a lightweight version of what I use.

The key is to study the drawdown characteristics, not the max. number of consecutive losses.

The tool gives you the average & stdev for whatever you set it to give you, here the P&L max DD.

Once you DD gets N stdev away from the average, your edge is likely gone. Unfortunately, trading outcome distributions are far from Gaussian, so you'll have to use a fairly large N (I use 5, most of the time) to avoid being "stopped" by some noise.

Damn, I haven't been on here for 2 almost weeks. Thanks I will check it out.