3 Steps for Finding a Real Trading Edge

In his excellent “Market Wizards” book series, Jack Schwager said, “There are a million ways to make money in the markets. The irony is that they are all very difficult to find.”

That’s true. If it wasn’t, we traders would all be multi-gazillionaires, wouldn’t we? Finding a trading edge is a lot of work, so it’s important to have an overall plan. In this first article of my series on finding an edge, we’ll look at an overview of the process. Future articles will delve into the details and different techniques that you can use in the search for your next profitable trading strategy.

Currently I’m primarily a Forex trader, but I’ve traded stocks, bonds and options as well. The procedure that I outline here can be useful to any trader, no matter the market.

What is an edge?

Before getting into our 3-step course of action, let’s first define what a trading edge is. We need to start with some basic trading math. This is a whole subject in its own right, so I’ll discuss trading math more fully in future articles. For now, let’s just concentrate on the ideas of risk, reward, and expectancy.

If I buy XYZ stock at 35 with a stop loss at 30, then my risk is 5 points. Let’s say I decide to sell if it reaches 50. Then my goal is a reward of 15 points. Of course I might not set a specific profit goal, waiting instead to see what the market gives me. If the stock rises to 44, and then stalls, I might sell there for a reward of 9 points. In the first example, my reward to risk ratio was 15 to 5, or 3:1, while in the second example it was 9:5.

Some newer traders naively think that they’ll automatically make money on average if they always set their profit targets higher than their risk. What actually happens is that they just get stopped out more often. This is because it’s more likely that the price will hit the stop before it reaches the target. So out of four trades, they may lose 5 points on three and earn 15 on the fourth, for a total “expectancy” of exactly zero (minus commissions and spreads).

The formula for expectancy is:

(Win Rate)(Average Win Amount) – (1-Win Rate)(Average Loss Amount)

Suppose I do 100 trades using some specific strategy, and that my win rate is 40%. My loss rate (or 1-WR) is 60%. If I have an average win of 8 points and an average loss of 5 points, then my historical expectancy for this strategy is:

(0.40)(8) – (0.60)(5) = 3.2 – 3 = 0.2 points/trade

The “edge” is the expectancy expressed as a percentage of my risk per trade, which in this case is 5 points. So the edge is just 0.2 points per trade divided by the 5 points I risk per trade, or 4% of amount risked. So for every $100 I risk with this method, I expect to gain $4. We say that my strategy has a 4% edge.

Gather and explore the data

The first step in finding a trading edge is to gather and explore historical price data. A simple internet search should yield several sources for this, ranging from tick by tick data to daily, weekly, or even monthly bars. In any case, you’ll want to get this data into a spreadsheet. Some data sources provide you with a quick way to do this, while others may require some copying and pasting.

If you’re not familiar with spreadsheets, then now is the time to learn. Using the powerful tools in MS Excel or Open Office, you can answer just about any quantifiable question you have about the data. What’s the frequency of inside bars vs. outside bars? If a bar breaks the previous bar’s high, what’s the probability that the next bar will do the same? And so on. If your creativity needs an initial kick-start, check out my blog site. It has oodles of free archived research notes, filled with examples of data explorations.

This is the stage that statisticians call “Exploratory Data Analysis” or EDA. You’ll want to look at overall features of the data such as bullish or bearish biases, the average price move per bar, and so on. This provides you with realistic profit expectations, and can lead you to further ideas for exploration.

Develop a trading idea

Great. So now you’ve got thousands of price bars in a spreadsheet, and you’re slicing and dicing the data to uncover its secrets. At this point, you’re bound to have a few “aha! moments.”

As an example, just a week or so before writing this, I was examining hourly bar data for the EUR/JPY currency pair, concentrating on just the 4-hour period during the London and New York session overlap. “Aha!” I said. I had just found that 72% of the time, the high or low for that overlap period occurred during the first hour, as opposed to the other three hours. Could I exploit this knowledge to create an edge? Well, I’m still working on that one, so you’ll have to stay tuned.

So the second step is to use what you’ve found in the exploratory stage to develop a specific trading idea. When developing your trading idea, be careful that you’re not just “data mining” and concentrating on some meaningless statistical artifact.

As a non-trading example of this, suppose I gathered data on the amount of rainfall in Boston over the past year, and organized it by the day of the week. It’s extremely improbable that any two days would have exactly the same average rainfall, so I could rank the amount of rain by day of the week. There’s clearly going to be a day, say Tuesday, that had the highest rainfall, and another day, say Friday, that had the lowest. But is this meaningful? Should I plan to have picnics only on Fridays, but never on Tuesdays? Of course not.

The storm clouds don’t know what day it is. This is what I mean by a statistical artifact found through data mining. There’s an old saying in statistics that if you torture the data long enough, it’s bound to tell you something.

So when you develop your trading idea, it’s important to have some theory or model, grounded in the real world, which explains why the idea should work.

For example, if you notice that price often makes big moves when it crosses the 50-bar moving average as opposed to other moving averages, what could be causing this? Could it be that this MA is often a favorite among speculators? If so, then this isn’t just a statistical artifact, it’s the result of prevailing trader psychology.

If you notice that currencies often make large moves after three consecutive positive trade balance reports, is this just a statistical artifact? Probably not, as there is a clear fundamental connection between a country’s trade balance and the demand for its currency. Maybe some big bank out there has a strategy of accumulating currencies with good trade numbers. This is a model grounded in reality, and supported by your data.

Test it rigorously

Now that you’ve developed a trading idea supported by your data and a logical model grounded in reality, it’s time to test it. This last stage can often be disappointing, and consequently is sometimes ignored by traders, to the peril of their account balances.

During this stage, you’ll be using basic concepts from probability and statistics, so it’s a good idea to brush up on those subjects. You’ll want to be familiar with such ideas as statistical power and significance, sensitivity and selectivity, type-1 and type-2 errors, and a few others. Again, I’ll explore many of these tools in future articles.

Not only do we want to know how often a signal correctly predicts some behavior, we also want to know how often the signal fails, and how often the lack of a signal correctly or incorrectly predicts absence of the behavior. It’s these latter three statistics that traders often overlook.

In the case of a purely mechanical method with a well defined signal, traders will usually back-test the signal with historical data. In this case, it’s often a good idea to do “out of sample” testing. This avoids the self-fulfilling practice of confirming your hypothesis using the same data you used to come up with it.

In cases where the trading method is a bit more qualitative and difficult to define for back-testing, you may want to forward-test the method in a live account. It’s best to use either a demo account or a small amount of money at first. In this way, you can gather actual expectancy data before committing more funds.


So now you’ve seen the 80,000 foot overview of the 3-step process for finding a trading edge. Gather and explore your data. Develop a trading idea. And finally, test it rigorously. If, during this last step, you find that your brilliant trading idea turns out to be a dud, don’t get discouraged.

And above all, don’t ignore your results and trade the idea anyway! That’s a sure path to emptying out your trading account. Instead, go back to your data and keep looking for ideas.

Remember, there are a million ways to make money in the markets. The trick is finding them. And now you’re on your way to knowing how to do that. Good luck, and keep pipping up!


Deixe uma resposta

O seu endereço de email não será publicado. Campos obrigatórios marcados com *