Trading Systems How to Tell if a Trading Strategy Works

Do you find yourself flitting from one way of trading to another? Or constantly tweaking the way you are trading, never quite finding that consistent winning streak?

The internet is an amazing thing – there are thousands of trading strategies described in forums, social media and YouTube videos etc. But how do you know if they work? The answer is much simpler than it seems. Test the strategy properly!

A proper test constitutes at least 30 correctly taken live or demo trades. The two elements to this are ’30 trades’ and ‘correctly taken’.

Why do you have to wait for 30 trades?
One of the biggest mistakes that newbie traders make is to give up on a trading strategy after a run of losing trades. The thinking behind doing this is understandable but very wrong. The thought is “If a strategy is losing trades, why keep doing it?” The point is that every trading strategy has losing trades!

There is no perfect system that never loses a trade. (Perhaps with the exception of a High Frequency Trading (HFT) hedge fund that is gaming the system by front-running large trades.) Maybe a trading strategy is very profitable over 30 trades – but just happens to lose 5, maybe even 10 trades in a row.

Sample size is the key
The point is that you need a large enough sample size to be able to judge whether the trading strategy wins more money than it loses. Take this analogy: If you had asked 10 US citizens if they voted for Donald Trump or Hillary Clinton in the 2016 Presidential election and 9 out of 10 people said they voted for Hillary, well why didn’t she win? It’s not enough people to judge how the whole country voted –the sample size is too small.

The same logic applies to a trading system. In fact, 30 is probably not enough trades to really tell – but it gives you a much better idea than a string of 4-5 losing trades that will put off a lot of traders from continuing a strategy they have started using.

What is a ‘correctly taken’ trade?
A trading strategy is defined by a set of rules. It is following these rules that give the system it’s ‘edge’ over a period of time. This edge produces a result that is better than random, and most importantly produces a profit.

It is not a valid test if you don’t follow the rules of the strategy. For example, a trading strategy says you should only place ‘buy’ orders when the price is above the 200 day moving average and only place ‘sell’ trades when price is below the 200 day moving average. If you place a buy trade below the 200 DMA you are basically trading a different system. The result of the trade – whether it is a win or loss – does not tell you anything about the viability of the trading system you are testing.

Don’t beat yourself up if you break the rules of the strategy you are testing. We are all human and make mistakes. The key is to realise that it was a mistake – we will call it an ‘error trade’. This trade does not count as one of the 30 trades. This is especially difficult if the error trade happened to be a winning trade. The best response to an error trade is to move on to the next trading opportunity and try to follow the rules.

In Summary
Remember your ultimate goal- a trading strategy that works. When placing the 30 test trades, the purpose is not to win every trade or trade on hunches or test other systems or get some ‘fun’ out of the market – it is purely to test this strategy. Jumping around from one type of trade to another is exciting but also disorientating. Using the same strategy over and over is perhaps a little less exciting but has a much better chance of long term profitability.

This article just scratches the surface of how to evaluate a trading strategy, but it outlines two guiding principles for how to judge whether a trading strategy works. At least 30 correctly taken trades are needed to test a trading strategy.

Jasper Lawler can be contacted at London Capital Group

 

[sminicooper]

A simple but really worthwhile article. Its easy methodology and important suggestion to actually do post-trade analysis ought to benefit anyone who doesn't already do it and is not achieving good results. I always think that it's the simple basics like this which have the most influence in making good trades.

 

[edmond360]

A simple but really worthwhile article. Its easy methodology and important suggestion to actually do post-trade analysis ought to benefit anyone who doesn't already do it and is not achieving good results. I always think that it's the simple basics like this which have the most influence in making good trades.

Well I think it's an utterly rubbish article written by someone who works for a broker, and whose thought processes and objectivity are completely clouded by wanting people to trade.

Ask yourself, how on earth can an arbitrary set of rules you've made up or read somewhere ever give you an edge in the market? Then, if you stumble upon some combination of said rules which happen to show profit, how and earth can that ever give you an edge? I promise you, it's just a statistical artefact and you're going to be made a fool of.

Be a little bit cynical. The common theme going on here is that whilst you work hard "testing" those rules, you're placing trades. Then once you go ahead and "apply" the chosen rules, guess what, you're going to carry on placing trades. There is no rhyme nor reason to those trades, all that has happened is the broker has succeeded in turning you into a mindless zombie who pushes the button and pays the commission.

 

[sigmund1]

Simple.

Simple and I'll even give you 2 for the price of 1.

You will need a Monte Carlo Simulator program.

If you are not using one of these then it's like buying a second hand car in the dark.

This formula will tell you what your optimum bet size should be if you just happen to be profitable in the first place.

K = W - (1-W)/R

 

[sminicooper]

Well I think it's an utterly rubbish article written by someone who works for a broker, and whose thought processes and objectivity are completely clouded by wanting people to trade.

Ask yourself, how on earth can an arbitrary set of rules you've made up or read somewhere ever give you an edge in the market? Then, if you stumble upon some combination of said rules which happen to show profit, how and earth can that ever give you an edge? I promise you, it's just a statistical artefact and you're going to be made a fool of.



Be a little bit cynical. The common theme going on here is that whilst you work hard "testing" those rules, you're placing trades. Then once you go ahead and "apply" the chosen rules, guess what, you're going to carry on placing trades. There is no rhyme nor reason to those trades, all that has happened is the broker has succeeded in turning you into a mindless zombie who pushes the button and pays the commission.

Who said anything about an arbitrary set of rules? The article is trying to encourage people to assess whether or not they stuck to their trading rules. Any form of post trade analysis is not going to help if you didn't adhere to your setups et cetera. Making a bad entry but ending up with a profitable trade is not going to improve your trading in the long term.

Why not forget that the author is a broker and may/may not have an agenda? (If you ignore everyone with an agenda you're not going to hear a lot are you?) Tyro traders need simple help and adopting a simple form of post trade analysis is a good start. You are of course entitled to your opinion of the article but it seems to me that you are the cynic.

 

[sminicooper]

Simple and I'll even give you 2 for the price of 1.

You will need a Monte Carlo Simulator program.

If you are not using one of these then it's like buying a second hand car in the dark.

This formula will tell you what your optimum bet size should be if you just happen to be profitable in the first place.

K = W - (1-W)/R

That’s fine advice but unfortunately out of context regarding the article! A Monte Carlo program is useful in solving probabilistic interpretations but that’s not what the author was talking about. The article describes a simple empirical method of post trade analysis – nothing more, nothing less.

 

[edmond360]

Who said anything about an arbitrary set of rules? The article is trying to encourage people to assess whether or not they stuck to their trading rules. Any form of post trade analysis is not going to help if you didn't adhere to your setups et cetera. Making a bad entry but ending up with a profitable trade is not going to improve your trading in the long term.

Why not forget that the author is a broker and may/may not have an agenda? (If you ignore everyone with an agenda you're not going to hear a lot are you?) Tyro traders need simple help and adopting a simple form of post trade analysis is a good start. You are of course entitled to your opinion of the article but it seems to me that you are the cynic.

Trading rules are always arbitrary. The article used the 200MA as an example. What is the average of the last 200 periods actually supposed to mean? Well it doesn't mean anything unless you first have a real reason to think the asset price will return to the average or else move away to some other average.

I agree with analysing past trades to make sure you are trading normally or to cut out mistakes and bad habits. But analysing them to reveal an edge? No chance. Post trade analysis doesn't help you identify reasons to be long or short in the next trade. How can it? Let's say I test some rules and they fail to work. I could work harder to improve the rules right. Or by the same logic of what constitutes an edge, I can just fade the buy/sell signals and I'll make a profit that way too. Ask yourself how confident you would be fading a system that lost money over 100 odd trades. It's so obviously silly when you look at it that way.

 

[aag100]

imho a strategy is working if you have more money than you started with

 

[NVP]

seriuously ?.........how can you tell if it works ?.......seriously ?

 

[sigmund1]

What Next?

OK, you've got "30 correctly taken trades" as per the explanation.

Lets assume your 30 correctly taken trades are showing a profit.

But what now and how do you tell if your trading strategy works over the LONG RUN?

Yeah, Yeah, I can hear you...it's profitable, what more do you want!

But not so fast.

It's one thing being profitable over 30 trades, but what about 300 trades?

I hate to be the harbinger of bad news, but I'm going to anyway.

There's a little quirky mathematical thingy that you might want to be aware of but don't want to hear.

That is, the more trades you take the greater your probability of going bust.

I hear you, I hear you, how can someone go bust if it's profitable in the first place.

The short answer is...

You not only have to be profitable, you have to be VERY PROFITABLE.

Only a Monte Carlo simulator can show you that, unless your some sort of maths whizz kid.

 

[edmond360]

Wait so what's the Monte Carlo simulator for? If you know your win rate, you can just calculate the probability of having X losers in a row given a sample size of Y. When Y >> X then that probability will tend towards 100%. It's common sense. You don't need to simulate anything, it's literally just typing a few numbers into your calculator.

But once you go down that road of using statistics, then at least be honest and admit that you've basically given up trying to do the hard work of understanding and thinking about the economy. You're just closing your eyes and placing blind faith in a statistic. Unless you do it properly, by firstly making a reasoned hypothesis, using that to predict what your statistical edge "should" be, and then going ahead and testing whether you get close to that edge. That is the way to do it, and it forces you to think about your rules and what they actually mean. But I don't see any traders doing it, so then you shouldn't kid yourself into thinking that you're taking a scientific approach. You're absolutely not, you're doing the opposite of science. Even people with maths and physics PhDs working at quant funds make this same mistake. They have no clue about how to apply mathematics to the real world, the way scientists of old used to do before they had Monte Carlo simulations.

 

[verial]

Only two things are necessary:

Explainability: If you cannot explain the trading strategy’s success with group psychology, you are unlikely to be working with something real. You could be overfitting.

Live trading: If you can make money with it consistently, it works. Use a brier score to calculate its success.

 

[sigmund1]

What went wrong!

Some think that all one needs to know is...

Your win rate...ho, ho, ho.

Our trusty hot shot trader (Gambler) gets 9 winners in a row. Golly Gee Whiz. Easy Street here I come.

What could possibly go wrong now?

Answer, that 10th trade.

The TENTH trade wipes out ALL the PROFITS from the previous 9 winners.

 

[foroom lluzers]

30 losing trades ? There is no edge here , if manual trading no more than 5 losing trades in a row .A greatsystem will give you no more than 5 losing trades in a row.

After 5 losses , the mind will not put on any more trades .For almost 12 days the trader was making no money.This makes it mentally difficult to trade , because it doesn't fit in with the human psyche.
Our psyche requires pleasure and reward , and trading does not offer it consistently.


In plain language, this means that we have to get a certain amount of pleasure and stimulation or rewards from our daily activities and what we put into our bodies. If we don't, then we create a pleasure deficit or what is known as "reward deficiency," and are subject to depression, anxiety and poor performance. Each day we have to stimulate our reward pathways adequately if we are to function well emotionally, mentally and physically

 

[foroom lluzers]

Here you have a nice buy signal on dax at 2210 stop 20 points price above support 1.2200 but mas not aligned

 

[CavaliereVerde]

A bit optimistic.
Unfortunatelly you need at leas 300 trades and 2 years to understand if a astrategy works.
And there is no guarantee it will work forever.