How to determine whether a trading system would be successful in the long run

Trading Systems Free Forex Trading Systems
1

To be successful, a trading system must have a profit factor of more than 1.28 over 1000 trades for trading capital of at least $100,000.

Profit factor = Gross profit divided by Gross loss

Therefore, Profit factor is a rough estimate of positive expectancy

Take for example a coin flip. If you were to take a coin flip over 1000 flips, the profit factor would be close to 1.0.

Why 1.28?

two words: [B]Oppurtunity Cost [/B]and [B]Risk premium[/B]

[B]Oppurtunity Cost[/B]

In order to justify its existence, a trading system has to make more money than simply investing it somewhere else.

[B]Risk premium[/B]

In order to justify the risk a trading system poses to a trader, the system has to make more money than other lower risk approaches.

  1. Studies by Eugene Fama have a shown that a simple passive low risk buy and hold strategy for stocks with low PE ratios yields an annualised return of 20.1 % over the course of 60 years after inflation. Thats a profit factor of 1.201

Therefore, a trading strategy has to have returns above at least 1.201 in order to justify the increased transaction costs and increased risk. There is no point in trading if you can get higher returns anyway from not doing much.

  1. Additionally, there is an average 3% inflation rate per year. This means that the system must have a profit factor of 1.231, to compensate for the inflation per year.

  2. Also, there must be a [B]risk premium[/B]. This is the reward for taking on additional risk. I’m sure many people would agree that there is really no point in taking additional risk for a one or two increased percentage points of return.

Risk premium is more difficult to calculate.

I like to use time oppurtunity cost to calculate the risk premium. How much time are you spending with an active trading system?

Then multiply that time with the minimum wage, which is currently $20 in higher wage countries like Australia. You have to calculate the time you have sacrificed for forex as compared with simply working in a mall somewhere.

For example, if a system takes one hour per day on average to manage, that is 250 x 20 = $5000. The system must make AT LEAST $5000 in addition to the profit factor of 1.2301.

Assuming you are rich and started out with capital of $100,000, that would put your required profit factor up 1.28.

[B]Problems[/B]

  1. If you are like the majority of traders out there who don’t have such an amount of capital to start trading forex, your profit factor has to be far higher. For example, if you started with $10,000, your profit factor has to be 1.73!!!

  2. Furthermore, a one hour per day trading system is considered fairly “set it and leave it”. Most systems are far more intense, and my require full time trading, possibly 8 hours per day. If it was 8 hours per day, your profit factor has to be a near impossible sounding 4.23 to justify the cost!!!

[B]Suggestions[/B]

  1. Make sure you start with at least $100,000, otherwise you would need an impossibly powerful trading system akin to a holy grail

  2. Go for trading systems that take up the least time, or employ algorithmic/robot trading. With algorithmic trading, you can pretty much abolish the risk premium factor in your calculations, but be careful.

  3. Adequately backtest your system and make sure it has a profit factor of at least 1.23 + risk premium over 1000 trades.

Now do you see how daunting the prospects are??

This post is also the reply I was preparing for the 3 individuals who accepted my challenge: namely master tang, simbafx and I think Matt. The above explains the parameters of success which I promised to take some time to prepare. Also, you are allowed to use position sizing at 2% risk to the capital.

For those who are unaware, the older thread concerns the impossibility of trading forex successfully BECAUSE of the above. I think a better way to phrase it is “impossible for trading to offer additional returns”, which in effect, would means its pointless. Granted, I was open to suggestions otherwise, and the 3 gutsy indivduals above accepted my challenge.

But to date, there have been no trading systems with a positive profit factor. Its pure gambling.

This was the reason for my bleak original thread, which was, in my opinion, unjusitifiably closed. I apologise for not putting this post on sooner, as this would have explained my position thoroughly. I simply want to share some insights, but I came across a bit harsh. You see, back from where I come from, we like taking a piss at each other.

2

Or no profit factor…

No Profit Factor.jpg700×201 18.6 KB

I realize it’s short of your “robust” 1k trade parameters, but it’s a demo. I won’t post live anything here.

But I assure you, my live results mirror this closely.

3

wow, thats very impressive.

However, what I worry about is whether you are using a very big stop loss:target profit ratio to get 31 wins straight.

I once made an ea that had a stop loss of 1000 pips and a take profit of 10 pips. It netted me 40 wins in one week. But thats only because the take profit was so small compared to the enormous stop loss. It was still a negative sum program, and had a negative profit factor and expected payoff.

But because it did not suffer a drawdown during the small period of the test, it seemingly had a positive expected payoff.

4

By thinking in terms of long run, one question that you ask yourself is “What happens if I keep doing this?” If there is more chance to win the game, it is a positive expectation game while the negative expectation is vice versa. The negative expectation game is the one that you have more chance to lose.

An edge refers to one’s systematic advantage over an opponent. Without an edge in games of chance, you will lose money in long run.

So, what about your trading? Consider your trading method, what happens if you keep using it? If it will give you profits over long run. That means you are trading with an edge.

5

rklee1 - I don’t disagree with your premise that to properly judge a trading system, or an individual’s performance, we need to look at risk-adjusted performance against other potential options and should be factoring in non-trading expenses as well as opportunity cost considerations. That said, you’ve got flaws in your arguments.

Firstly, there’s the profit factor (PF). Dividing gross profit by gross loss may be a quick and dirty way to get to expectancy, but you cannot then take that and compare it to annualized returns. You’re comparing apples and oranges when you match your PF to the 20.1% Fama performance figure. (by the way, doesn’t this low PE portfolio’s performance imply the use of fundamentals can produce excess returns?)

For example, Trader X has a $1000 account and does 1 trade per day with a $1/day net expectancy. At the end of the year he’d have $260 in profits ($1 x 260 trading days). That would be a 26% return, but would it be a 1.26 PF? The only way it would be is if his gross profits added up to $1260 and the gross losses came in at $1000. Any other combination leads to a greater or lesser PF. If his gross profits were $10260 the PF would have to be 1.026 ($10260/$10000), while if his gross profits were $760, then the PL would be 1.52 ($760/$500). Thus, Fama’s 20.1% is almost definitely not the same as a PF of 1.201, so let’s throw PF out as the benchmark and stick with risk-adjusted returns.

Now, on the risk-adjusted returns front, I think it’s pretty safe to say that everyone here is including transaction costs in their net, so the “justify the increased transaction costs” bit can be tossed as meaningless. If Trader X is making 26% per the example, it doesn’t matter what his transaction costs were. If they were high, they were offset by higher returns. Bottom line is 26% is greater than 20.1%, even when taking out 3% for inflation.

Of course the trader should definitely include any other costs he may have that aren’t transactional - like data, software, etc. This will vary by trader from zero to potentially large numbers.

To the point of determining risk premium, it’s defintely true that this can be more tricky. I would in no way, however, equate it at all to opportunity cost as you have done. I’ll get back to the latter. Risk premium, as the name implies, is a measure of the comparative risks taken. It normally is considered to be something like the relative variance in portfolio value or in returns. From that perspective, Trader X’s 26% system return (assuming it was consistent) could be anywhere from much, much better than the Fame result to much, much worse depending on the comparative volatility in each’s performance. This is what the Sharpe Ratio and other similar metrics attempt to capture. To get a proper risk-adjusted comparisson you need to come up with some reasonable metric that can be applied across all compared systems and/or benchmarks.

Getting back to opportunity cost, this is a very tricky thing to quantify. Obviously, the most basic element of it is forgone returns from alternative approaches (assuming comparable risk, which is why a proper risk premium must be assessed). As is rightly pointed out, however, there is also the time element. Because people value their time in different ways, we cannot assign a time value from outside, but it can certainly be applied in a uniform fashion where the individual’s work to achieve returns - be it assessing 401k options, selecting ETFs, or following a trading system - is factored into the equation.

My contention is that you need a seperate metric for looking at this time element. A simple $/hr calculation, as was suggested, will suffice. But beware that the time element of trading/investing cannot just be viewed in cost terms because things like entertainment and education value come in to play. For example, when I first started coaching volleyball I calculated what my hourly rate was when factoring in all the time I was putting in to it. The result was something below $1/hr as I recall. It bothered me not at all, however, because I enjoyed the work and was developing myself as a coach such that I could increase that effective hourly rate moving forward.

6

Surely not “whether”? Surely there can be no doubt that the stop losses used must be great enough to ensure that a loss or small string of losses should be sufficient to wipe his account clean?

Given what your belief is and all that…

8

It’s not really an important distinction to the discussion so long as you use the same for measure all your options. The 20.1% noted is an annually compounded rate of return.

For the ‘trading system’, if you were only making $20 (spending 1 hour) per day on $100,000 then the compounded return is 5126.6 rather than 5,000. Not a huge difference I grant you but…

A correction: If you’re making $20/day there is no compounding. It’s just a flat notional return. You’d have to speak in % terms as you did with the 0.2%/day thought.

Now, let’s assume that we follow your 2% risk rule and take 1 trade each day, suddenly achieving a return of 0.2% per day seems both reasonable and possible wouldn’t you say?

Keep in mind that you can average 0.2%/day, but that doesn’t necessarily mean having a compounded 0.2%/day rate of return.

10

I love this post ! :cool:

11

True!!!, can’t see any DD numbers. Without the DD numbers, profit or expected payoff do not mean anything. It’s about risk reward. I do have backtests with profit factor of > 10.

Statistics1jan2010-30jun2010.jpg700×173 23.8 KB

Account of $100.000 to start trading forex? You must be out of your mind!

$1000 is all you need to trade safe with the system I’m using.
13 pips/trade, 60 trades in 2,5 month = about 1000 pips each 3 months.
Till now I have not seen DD > 600 pips. LIVE and TEST (worst DD in 1/2 year backtest in image above)

Scaling up the lotsize first 6 months -> 10% profit each month with 10% risk on original deposit.

PF of 1.23 ppff 4. blablabla… ???
I do not even use PF’s.

max DD, average pips trade, average return month, Perc win. All I need to know…

then I’m able to calculate the average profit month (using perc of risk = 10%) and certain probability not to hit the max risk level of -10% (88%)

Then profit is 10% month…
using max 10% risk original deposit…
probability of not hitting the -10% level = 88%…
costs of spread is 2/15 pips at the moment (LIVE) = 14% transaction costs.

12

Yup Pip - Pfft pretty much sums it up perfectly :smiley:

13

sorry :wink: whoops…

14

Sorry, my replies aren’t the most intelligent at the moment - mainly because this is still tickling me… :smiley: :smiley:

Thought I was still still searching, but by the suggestions in the first post and seeing as I certainly didn’t start with 100k (but getting a little closer every day), I’ve got “the grail” right here. It’s available for sale for $9.99 … any takers? (just kidding) Meh - perhaps naysayers are better off just opening a 3% savings account. :–)

I wonder if Buffet worked out his required PF at the start - you know, back when he had a fortune and the grail already to hand?

I’m sorry, but that line is still getting me… :–) :–)