### Use of Probability of win and Statistics in Trading

#### Having an ‘Edge’ and trading vs. Gambling

Gambling: The odds are against you and you have to rely on luck to win. For example, take the lottery. Who makes money with the lottery? Well, it’s used by governments to pay for schools, etc. That means that they make a profit on the lottery. If 1 million individuals pay \$1.00 each for a ticket, the only way they can make a profit is by paying out less than \$1 million. So, the math for the lottery is such that even if some individuals are lucky, they cannot usually exceed \$1 million in total winnings among all winners. So, in other words, buying 1 million tickets yourself with \$1 million, you would end up with less than \$1 million in winnings, even if you have winning tickets. The math has been done and the probability of the game has been fixed against you.

The operators of the lottery have the ‘edge’ in this game. The only way you could possibly win as an individual is by sheer luck. You would have to randomly get one of the winning tickets. You have no control. You cannot repeat any success (in the rare event that you do win). Most of the time you will lose. Statistically, over time, you will pay far more into the lottery than you get out (if you get anything out).

The same principle is used at the casino. All the games in casinos are fixed at specific win/loss ratios which always favor the casinos (the house has the advantage of the ‘edge’). Sure, they may pay out some large sums occasionally (losses to the casino) to a few people, but they collect small steady sums (winnings to the casino). As long as their income exceeds their expenses, they make money. And with the odds fixed, they do, lots of money. All gambling is set up this way. The house always has the ‘edge’ in all gambling. In reality, this edge does not have to be significant. Even 51% winning and 49% losing will make a casino lots of money over time. It’s a numbers game.

#### How do you get the edge? What really is an edge?

If you design a trading system, test it and it works to make you money, it should do better than randomly entering trades.

For example, My system is able to predict days the market goes up 65% of the time and I can buy that market based on my system. Alternatively, I could enter the market each day using a coin toss, giving me 50% accuracy in predicting correct market direction.

Calculating the edge for this example:

Edge = (percentage of win of system) – (percentage of win of random system)
Edge = 65 – 50 = 15%

If my own system is only as good as a random entry system, I need to change systems or make some adjustments on the edge would be zero. We need an edge greater zero to make a system work.

You don’t really need much of an edge. Even a tiny amount is sufficient. Being able to have 51% winners is enough, giving you an edge of 1%.

Look at the following hypothetical scenario:

I have a coin that comes up heads 70% of the time and this has been verified by 100,000 coin tosses. I’m now offering you the choice of two deals:

Deal 1: We can toss the coin exactly one time and if it is Heads, you get \$1 million dollars. If it comes up tails, you lose and get nothing.

Deal 2: We can toss the coin 1 million times and every time it comes up heads, you get \$1. If it comes up tails you get nothing.

Which one would you choose?

It may look like a trick question, but the answer is really quite simple if you think about it (and maybe you caught on right away).

The first choice represents gambling because you have no edge. You rely on luck. The odds are pretty good, but you only have one chance. It’s all or nothing. It may be tempting because you have a 70% probability of win, but no guarantee. And, everything is lost and won in one single event. No control. Not repeatable. One chance and you’re done.

This is the person who comes into trading trying to hit the one big home run all the time. He or she may mortgage the farm and try to just hit one big trade to be set for life or to get rich quick in Stock Market (as often advertised). Often, he or she may even get the first trade right. But then, thinking that the system works, the second or third trade is a total loss of everything gained (or often more). That’s gambling.

Now, let’s look at the second deal. Would you like a sure deal? The right answer here is that you want Deal #2 because it is guaranteed to get you \$700,000 in profit, with a 100% guaranteed chance of success. No doubt about it. No risk to take. Guaranteed. Deal one has a 30% probability of failure. Yes, you could make more money and you have a 70% probability of win, but that’s not good enough. In trading, you cannot risk losing your entire account in any one trade. You have to be able to be there for the long run and be able to handle some small losses.

On deal #2, you’d lose 300,000 small trades of \$1.00 each, but that’s ok because overall the system is profitable, making you \$700,000 guaranteed a profit. Also, you would not have to panic if the first 50,000 trades in a row were to be losers because you know that your system has a 70% winning rate in the long run based on previous thorough testing. So, this example actually illustrates several important points.

Talking about ‘edge’. In deal #2 you have the edge. You can prove that you are guaranteed to win \$700,000 by just doing the math.

The edge may also be represented in different ways. What about a system that has a 30% win rate? How would you calculate an edge? Obviously, the above example would not work, taking a random coin toss with 50% and comparing that to a 30% win rate. You could calculate what percent return your system generates on each trade instead. So, for example, your system returns, on average, 0.1% return on your account balance. Comparing that with a system that generates random trades, which has a percent return of zero.

A better way to assess the edge of a system with simpler calculations can be done (indirectly) by calculating the system’s EXPECTANCY instead of calculating the edge directly. Expectancy is very easy to calculate and is explained below. Once we have calculated the ‘expectancy’, we can then also calculate the ‘edge’ and ‘risk of ruin’.

Expectancy (an easy calculation to assess your edge)

The Expectancy is the average dollar amount that you can expect to gain or lose with each trade and can be calculated as follows:

Expectancy = (Probability of Win * Average Win Size) – (Probability of Loss * Average Loss Size)

To analyze this, let’s look at an example of two demo trading accounts. Trades are  from 01/16/2017 to 02/04/2017, we had the following results:

Probability of win: 87 / 94 = 0.9255 (92.55%)
Probability of loss:  = 0.0745 (7.45%)

Account size at beginning: \$10,000
Ending Balance: \$13,696.66

Average profit per winner: \$60.32
Average loss per loser: \$221.57

Expectancy = (Probability of Win * Average Win Size) – (Probability of Loss * Average Loss Size)
Expectancy = (0.9255 * 60.32) – (0.0745 * 221.57) = 55.82 . 16.50 = 39.32

This tells us that for every trade, we could expect to earn \$39.32.
With 94 trades x \$39.32 = \$3696.08 (rounding differences)

So, winning or losing, any trade that happens on average makes us \$39.32 in this example. Looks like a good system. It has a positive expectancy and should make money over time if it keeps producing the same results in the future that we have seen over the 2 weeks of demo forward testing.

Obviously, you want a system with a positive expectancy, the larger the better. Also, note that a forward test of only 2 weeks is not enough to establish that a system will work over a longer time period. Statistical significance and the need to have many trades in a test are discussed below (and you will see that 94 trades is not sufficient for statistical significance).

#### Risk Reward Ratio

Another way to look at system performance and profitability is the Risk-Reward Ratio.

Imagine entering a trade and placing a profit target 10 pips above the entry price and a stop loss price 20 points below the entry price. If you win, you gain 10 pips, if you lose, you lose 20 pips. This is a very poor risk-reward ratio of 2. For every dollar you put into a trade, you risk losing more than you can gain from the trade by a factor of 2 since you can expect to gain \$1.00 if the trade works and lose \$2 if it does not.

You would need to have a win/loss ratio of 2, meaning 2 winners to every 1 loser to break even (not counting commissions or slippage) and a ratio greater than 2 winners to every 1 loser to make money on trading this system.

Here are sample numbers to illustrate this:

Losers: 100
Winners: 200

Probability of win: 200 / 300 = 2/3
Probability of loss: 100 / 300 = 1/3

Average win 10 pips
Average loss 20 pips

Again, calculating expectancy:

Expectancy = (Probability of Win * Average Win Size) – (Probability of Loss * Average Loss Size)
Expectancy = (2/3 * 10) – (1/3 * 20) = 0 (break even, but losing commissions and slippage)

Now, this Risk-Reward scenario of 2 is fine as long as your winning percentage, in this case, is greater than 2/3 (plus the amount to pay commission and slippage).

Bottom line, you can have a system that wins 95% of all trades and still lose all your money very quickly because your few losing trades could be substantial, wiping out all the many small gains. Conversely, you could have a system that loses 95% of the time and makes money overall because the wins are much larger than all the many losses combined. You could have a system that has exactly 50% winners and 50% losers. The winning percentage by itself is meaningless. You also need to know how large the average winner and average loser is to calculate if the system will be profitable.

However, equally worthless is the risk/reward ratio by itself since it only considers the Average Win Size and the Average Loss Size of the expectancy equation. You want a positive expectancy, which also includes the probability of win and losses. In the end, you need to make money with your system and nothing else really matters.

You will see a lot of people get excited about the risk/reward ratios of systems. Many traders and people who write books about the trading rant and rave that systems must have a low risk/reward ratio on each trade. This may make you feel cozy about your system but really does not matter. I would rather have a system that wins 90% of the time with a risk/reward ratio of 2 than a system that wins only 10% of the time with a risk/reward ratio of 1/2. See for yourself:

 Risk/Reward ratio Expectancy Profit with 100 trades low: 1/2 (10 pips per loss, 20 per win) (0.1 * 20) – (0.9 * 10) = -\$7 per trade -\$700 high: 2 (20 pips per loss, 10 per win) (0.9 * 10) – (0.1 * 20) = +\$7 per trade +\$700

All that has changed in this example is the risk/reward ratio of the trades. Again, you have to consider the whole picture by calculating the expectancy.

The only way to figure out if a system is profitable is to look at the expectancy and do the math. For that, you need some real test data from backtesting and preferably from forwarding testing for a few weeks to months (the longer the better). If the numbers turn out to be great (including statistical significance), it’s worth trading the system with real money.

#### What is Risk of Ruin and why it is so important?

The probability of losing enough of your money to the point that you are unable to recover (and keep trading) is referred to as ‘risk of ruin’. In short, it is the probability that your account balance will go to zero (or near to zero).

For example, assume that you have a \$10,000 account. You lose \$9999 and are left with \$1. You will NEVER be able to continue trading with your \$1. Even if you were left with \$100 instead of \$1, you may never be able to regenerate your account or keep trading. However, losing \$4,000 or so would still leave you in a position to keep trading. So, the point of ruin is reached once you are unable to recover from your loss and you have to stop trading. The Risk of Ruin is the probability of reaching the point of ruin.

Looking at an example, if you have a system that has, on average, 20 consecutive losses followed by 1 winner, you would have to ensure that you can trade at least 21 times in a row without depleting your account to survive until you have a winner. If you risk too much on each trade and lose everything in your account over 15 consecutive trades (or any number less than 21 for that matter), you have a risk of ruin of 100%. If you have enough money and good money management allowing you to place 30 losing trades in a row without losing your entire account balance (and assuming that your winner exceeds all of your losses combined), you will make money and you will have a lower risk of ruin.

Now, even though your system may have on average 1 winner for every 20 losers, there is still a risk here that you could have 60 losers in a row and then 3 winners. This may be unlikely, but still possible. Therefore, there is still a risk of ruin in any system, as in this example, and you have to be aware of that. You just have to make sure that you don’t have a system that already starts out with the fairly high risk of ruin.

#### Here is some data to illustrate this point:

 Risk Losing trades until ruin 1% 100 2% 50 3% 33 4% 25 5% 20 10% 10 20% 5 50% 2

For properly backtested and forward tested systems, Amibroker or whatever software package you are using will generate reports for you. One of those numbers is max # of consecutive losers. This will give you an idea of how many trades in a row you may see ending in a loss. However, you should at least double this number, or even better, triple this number and then allow for that many losing trades in a row. You will need to have the ability to recover from such a losing streak and the way to ensure this is by making your risk per trade very small. Again, you should not exceed 5%. Better numbers are between 1 – 2 % of your account per trade.

Now, if your system has a positive expectancy, your system also has a positive edge. So, your account balance should be increasing with most trades, making it harder to delete your account since it is increasing (in essence decreasing your risk of ruin with each successful trade).

Let’s assume that we have tested a system with the following numbers:

Account size: \$15,000

Edge = Expectancy / Risk per trade
Edge = \$7.50 / \$150 = 0.05 = 5% (return per trade on money risked)

The formula for risk of ruin is as follows:

Risk of ruin = ((1 – Edge) / (1 + Edge)) ^ Capital units
(^ means ‘to the power’)

Capital units = dollar value risk per trade

So, risking 1% per trade on a \$15,000 account, each trade would have a dollar value of \$150. Capital units would then be the number of \$150 trades you could lose until your account is at zero if you had no gains in your account at all (essentially straight losing trades until your account is at zero).

Capital units = \$15,000 / \$150 = 100

This means that after 100 losing trades, your account is at zero (assuming straight losers without any gains).

Example 1: Using the numbers from above, assuming that your system has a 5% edge, using a \$15,000 account and 1% risk per trade, we can calculate the following:

Dollar value risk per trade = \$15,000 x 0.01 (1%) = \$150
Capital units = \$15,000 / \$150 = 100

Risk of ruin = ((1 – 0.05) / (1 + 0.05)) ^ 100
Risk of ruin = ((0.95) / (1.05)) ^ 100
Risk of ruin = 0.000045 (or 0.0045% = near zero = very unlikely)

Example 2: Compare that with taking significant more risk per trade as follows: 5% edge, \$15,000 account, 20% risk per trade.

Dollar value risk per trade = \$15,000 x 0.20 (20%) = \$3,000
Capital units = \$15,000 / \$3,000 = 5

Risk of ruin = ((1 – 0.05) / (1 + 0.05)) ^ 5
Risk of ruin = ((0.95) / (1.05)) ^ 5
Risk of ruin = 0.60 (or 60% = very likely = more than random chance)

In essence, the first example shows that your risk of ruin with your trading system is almost zero. You can pretty much guarantee yourself that you will still be trading many moons from now with an account that will be growing over time as long as you have a system with a positive edge.

The second example is a sure recipe for disaster as the calculations show. Risking 20% of your account per trade is making ruin the likely outcome.

This is why expectancy (or edge) alone is only part of the picture in trading.

You need two components of your system to work:
(1) Positive expectancy or edge
(2) Proper money management to prevent ruin

You have two choices to improve your success:
(1) improve your edge (typically there is a limit to this in system design)

Example 3: Significantly increasing your edge to 30%, \$15,000 account, 20% risk per trade.

Dollar value risk per trade = \$15,000 x 0.20 (20%) = \$3,000
Capital units = \$15,000 / \$3,000 = 5

Risk of ruin = ((1 – 0.30) / (1 + 0.30)) ^ 5
Risk of ruin = ((0.70) / (1.30)) ^ 5
Risk of ruin = 0.045 ( or 4.5% = less likely, although not as good as example 1)

As you can see here, if you have a great system with a significant edge, you can afford to risk more per trade. There is a direct relationship between edge and risk of ruin. If you are good with math and looking at the formula itself, this makes sense. It also makes sense just thinking it through.

If you have a large edge, this means that you have many winning trades (and few losing ones) and this, in turn, will increase your account rapidly, making ruin a less likely outcome with each winning trade.

In summary, Risk of Ruin DECREASES with:
(1) Larger Edge
(2) Lower risk per trade (smaller position sizing)

##### One more VERY important thing: Statistical validity of your system tests

You have a system, you have tested it, you have crunched the numbers and have found:

(1) Positive Expectancy
(2) Positive Edge
(3) Proper position sizing
(4) Low risk or ruin

Now what? Done? Let’s trade! Not quite.

How do you know that the numbers you have crunched for your testing are actually valid? Or, in other words, how reliable and how reproducible are the results obtained from testing on a real account?
Using statistical tools and number crunching as explained above requires one more step to ensure the validity of the results obtained.

There is actually a way to predict how well your testing performance will correlate to real performance, assuming that the general conditions of the markets during your live trading are similar to those of the testing period.

Imagine a couple of simple scenarios:

Your system makes one (1) successful trade over 3 months. You do your analysis (expectancy, edge, the risk of ruin) and it shows that your system is fabulous (obviously). Great. Or is it? Was this result due to pure luck, i.e. a ‘random’ win or is the system actually good?

You really cannot tell since you only have 1 trade total. How about 10 trades? Assume you have 10 successful trades over 3 months with 100% winners. Would that convince you? It would be better, but still not enough as you will see later.

How about 100 trades over 3 months? Better, yes. Good enough? Probably not.

Let’s look at a specific example:

Let’s assume you have a system that buys the EURINR. The target profit (TP) on each trade is 10 pips. The stop loss (SL) on each trade is set at 80 pips. With this ratio of TP to SL (10: 80), for each 1 trade that hits a stop loss, you have to win 8 times to make up the losses. So, 8 out of 9 trades have to be winners just to break even (not counting commission and slippage). To make money, you have to win more often than 8 of 9 trades.

8 of 9 trades is a winning percentage of 88.88%, so your system has to win more than this percentage (plus commission and slippage) to make you money.

Let’s assume now that you have done 3 months of forward testing with your system and it shows a winning rate of 92% with the average trade making you \$5.00 (which is the expectancy of the system).

You now look at this with great excitement, having found a system that shows all the right results that have been forward tested and verified with all the statistical tools we have covered so far. It even has a very nice expectancy of \$5.00 per trade. How could you go wrong?

To understand what’s missing, we have to talk about the ‘Margin or Error’. What is it and why should you care? The margin or error is a measurement of test validity and reproducibility. Before presidential elections, you always hear about the margin of error. ‘Candidate 1 leads against candidate 2 with 51% to 49% with a margin of error of 3%.’ What that means is that Candidate 1 could have as little as 48% of the vote or as much as 54%. It really means that the survey results in this example are worthless because either candidate could win since there is an overlap of the winning percentages if you add the margin of error to either number. The margin of error is directly proportional to the number of people polled for the survey. If you surveyed every person in the country, your margin of error would be zero since no statistical extrapolation would be necessary. If you only sample 1,000, your margin of error is quite large. That’s what pollsters do. They essentially take a small sample of people on the street and then extrapolate the results to represent the entire country. This can be done with statistics, but you have to include the margin of error. The same concept applies to trading system testing and winning percentages.

Back to trading and system testing, what if your test has a winning rate of 92% plus/minus 5%? In reality, this means that your real winning rate on a repeat test or live trading could vary from 87% to 97%. If it is really 87%, you’ll lose your entire account in a hurry since you need 88.88% winners to just break even. So, it’s really not enough to produce 92% winners in your test. You also have to establish that you can repeat the 92% (or have very close results) on a repeat test or real account.

You have to understand that ALL test results have a margin of error and you have to take this into account. Otherwise, you will go broke (and wonder why your system worked on a test account but not a real account). You have to ensure that your margin of error is small enough so that you can guarantee yourself repeat success, repeatable and reliable test results.

Again, If you produced the result with 1 trade over 3 months, you don’t know if it is random or not. You essentially have a 50% probability of repeating your success and a 50% of losing. Essentially, your result may not be reproducible and therefore not be valid. However, if you have 10,000 trades over the 3 months with this result, you could be very confident that this result is valid and reproducible.

The basic concept here is The fewer number of trades you have, the less reproducible and reliable it is. The larger your number, the more reproducible and reliable your results.

So, how many trades are enough to ensure reliability and reproducibility of your results? You guessed it. There are some formulas to calculate this and then below I’ll show you some numbers specifically.

We calculate the Margin of Error in 2 steps, by first calculating the ‘standard error’:

(1) Standard Error = square-root of (( winning rate * (1 – winning rate) / number of trades))
(2) Margin of error = 1.96 x Standard Error

Using a factor of 1.96 in the above calculation, we can be 95% sure that our results are valid. The 95% is generally accepted as the standard in statistics and science. This means that we are 95% sure that the test results are due to a good system rather than random events faking the results. Or, in other words, there is only a 5% probability that our test results have happened purely by chance/random events. This is also called a ‘95% confidence interval’.

You could increase your percentage to 99% validity if you wanted, but there are some drawbacks to this as you have to have many more trades to achieve this and the 95% is usually good enough. It is also used by scientists worldwide as the accepted standard.

So, let’s run some numbers to look at our system with the 92% win rate and \$5.00 expectancy per trade. We will compare this system having 92% win rate over 10 trades, 100 trades, 1,000 trades, 10,000, 100,000 and 1 million trades (using 95% confidence interval):

 Number trades Margin of error Low end win % High end win % Reproducible Good enough 1 54% 38% 100% NO NO 10 17% 75% 100% NO NO 100 5.4% 86.6% 97.4% NO NO 500 2.4% 89.6% 94.4% YES MAYBE 1,000 1.7% 90.3% 93.7% YES >YES 10,000 0.5% 91.5% 92.5% YES YES 100,000 0.16% 91.84% 92.16% YES YES 1,000,000 0.05% 91.95% 92.05% YES YES

What is this information telling us?
If you only have 100 trades in your test, you can be 95% sure (very sure, in other words) that you can repeat the performance of 92% plus/minus 5.4% winners. So, you have a possibility of only 86.6% winners which is not enough to break even.

This does not mean that this system will fail or that it is a bad system. In fact, we think this is a great system, but we cannot be sure. It means that we have not done enough (in terms of enough trades) to test this system which otherwise looks very good in terms of expectancy and all other numbers we have crunched. Realistically, no one is going to run 1 million trades on a test, but it should be very easy to test your system over longer periods of times.

Over our testing, we had 226 trades within about a 6-8 week time span. You can see above that at about 500 trades you can get fairly reliable results, so this is achievable within a fairly short period of time (months) on a demo account with forward testing. Running a backtest over a 12 month period should easily be able to give you 1,200 trades or more. Five years of data would give you 6,000 trades or more quite easily.

This would be very reasonable since this longer time frame would also test your system over many different market conditions. So, you get multiple benefits: many trades and many different market conditions over time. If your system holds up over longer time periods and does well, it’s likely a very solid system.

Just to complete the discussion about Margin or Error, if you change your confidence interval from 95% (95% sure the results are not random) to 99%, you will have to run much larger tests with even more trades.

The margin of error multiplier in the formula:
Margin of error = 1.96 * Standard error
Changes to:
Margin of error = 2.58 * Standard error

This is usually not worth it since 95% reliability is enough and not much is gained by taking it to 99%. However, lowering it to 90% is not acceptable for statistical validity and the 95% has been generally accepted as the threshold worldwide.

ANOTHER VERY IMPORTANT POINT: Reducing the winning percentage closer to 50% also causes huge changes in the results. The closer you get to 50-50 probability of win vs. lose, the more likely it is that your system is random, or at least resembles a random system. Therefore, larger numbers of trades are required to prove that it is not random. So, a system that has 55% winners will need much larger numbers to produce the same small margin of error when compared to a system that has a win rate of 92%. In other words, the margin of error will be much larger with the same number of trades. You are essentially trying to prove that the 5% difference (55% vs. 50% random win rate) can be directly attributed to your system, rather than chance (randomness).

So, here is the table with margin of error for a system with a 55% win rate, 95% confidence interval (compare to the one above for a system with a 92% win rate)

 Number trades Margin of error Low end win % High-end win % 1 99.5% 0% 100% 10 31.5% 23.5% 86.5% 100 10% 45.0% 65.0% 500 4.4% 50.6% 59.4% 1,000 3.1% 51.9% 58.1% 10,000 1% 54.0% 56.0% 100,000 0.3% 54.7% 55.3% 1,000,000 0.1% 54.9% 55.1%

Comparing the two systems above, one with 92% win rate, the other with 55% win rate, you’ll notice that the margin of error is close to 2 times as large for the 55% win rate system than the 92% win rate system, which is what we would expect.

##### Pulling it all together

In summary, you must do some number crunching with all of your systems after getting some data on backtesting and forward testing to calculate edge, expectancy, and risk of ruin.

To evaluate any system you need to go through these steps:

(1) Backtest the system over different markets and time frames, ideally over several years of data with many trades
(2) Forward test the system over several weeks to months on a demo account with many trades
(5) Calculate your risk of ruin
(6) Calculate your win rate and margin of error for statistical significance
(7) If expectancy, edge, risk or ruin, win rate, the margin of error and forward tests are favorable, trade on a live account.

##### The importance of having a system in the first place

A trading system essentially involves and defines the following fundamental parts:
(1) What currencies (or commodities, or stocks, or whatever) do I buy and sell?
(3) When should I sell and take a profit if things go well?
(4) When do I cut my losses if things go badly?
(5) How much money should I risk for each trade?

Your system is complete when you have come up with rules to answer all 5 components.

Here is an example of a very simple system (I have not actually tested this. It’s a random example for illustration purposes):

(1) I will buy the USDINR.
(2) I will buy (go long) every morning exactly at the opening of the NSE Session.
(3) I will sell exactly 30 pips above where I bought
(4) I will sell with a stop loss if I’m down 20 pips from where I bought
(5) I will risk 5% of my account balance on each trade.

You could have arrived at these rules by several ways:
(1) You have studied charts and have noticed this pattern would be successful (again, this example is random).
(2) You are randomly trying to come up with rules and systems to test (like this example).
(3) You read about strategies online or bought a trading book that showed you some successful ideas and concepts that you have copied or combined with your own strategies.

Often, coming up with your own trading strategy really just involves reading about other peoples strategies and then tweaking them until they work or work better. Most of the time people are willing to share quite freely what major components they use for their systems, but they won’t tell you the exact settings.

Using the above example, a successful trader may tell you that:
(1) You should trade the major currency pairs, may even tell you the specific pair.
(2) You should buy when one of the major sessions open
(3) You should set a pip target for taking profits
(4) You should set a stop loss pip target that is smaller than the profit target
(5) You should use a low-risk profile.

All pretty helpful generalizations and good advice, but worthless if you are new to trading and are trying to set up your own trading plan without a clue on how to do it.

Here is another (real) example of a trading system using some indicators:

The 50-200 Moving Average system (a very well known and simple system):

(1) Trade any major currency pair.
(2) Buy (go long, opening the position) when the Middle Moving Average crosses above the slow moving average from below AND the Fast Moving Average is above the Middle Moving average.
(3) Close your position when the fast Moving Average crosses below the middle moving average from above.

What’s left to your own interpretation and implementation is:
(1) What is a moving average?
(2) If you understand moving averages, what time interval is used for the long, slow and middle moving average?
(3) Same as (2)
(4) Where are stop losses?
(5) How much money do I risk?

Again, trading books and online resources could give you some ideas, but you will likely find many different settings for this system and would be left to experiment for yourself. Most of the successful settings are not freely shared by successful traders, or they may give you some basic ideas, but usually not the whole picture and strategy.

One of the major points we are trying to make here is that you have to have a system and then test it and tweak it until you have proven that it works and makes you money consistently. If it doesn’t, even with tweaking, you start over and redesign.

Another point is this: Don’t try to reinvent the wheel. Read about strategies that already exist and use them as a starting point for your own system. There are a few great books and websites to get you started and we’ll list some for you later.

Without a trading strategy or plan that defines each one of these parts in detail, you cannot trade successfully. You cannot trade on rumors or gut feeling but have to design a trading plan or strategy that will allow you to make money consistently over time.

To know that you have a successful trading plan or strategy, you have to do several things:

(A) Make observations about market behavior
(B) Design your trading plan with specific rules to exploit the market behavior
(C) Establish money management rules