Originally published in French on abcbourse.com, June 18, 2014. The French original is still online at abcbourse.com; read it there if you want to check this translation against it. Translated, not rewritten: what follows is the original argument as it was made, including what time has since proven wrong. Where that happens, you'll find a note from the desk, dated 2026.
In my last article, I explained why it's a mistake to fixate too heavily on the entry price of a trade.
I'm going to carry on this mini-series by giving you the best entry setup for short-term trades: anything from a few minutes to a few hours.
As I explained, giving too much weight to the entry price completely blots out the accounting and statistical side of trading, and that kind of reasoning is what drives most traders to failure.
Here, then, is the solution I've come to rely on most over the years, one that should help you steer clear of that trap.
The first, obvious point is that you can't trade the same way all the time, or to put it more precisely, you can't run the same trade in every market condition.
You'll grasp easily enough that if the market's climbing over the long term, you can't trade it the same way as during a crisis move, something like 2008.
In the same way, it's easy enough to see that when the market's wide awake, you don't trade it the same way as when it's fast asleep.
Which is exactly why your trading strategy has to start with a FILTER.
The idea is to define a type of market: to put a number on it, so that whatever trading you go on to do actually matches those conditions.
This filter has to be simple. The risk is building one so complex that you end up with a hopelessly narrow selection of conditions. Imagine, say, a filter like: Monday afternoon, if the market closed up on Friday and opened slightly down that same morning. On paper, sure, the results might look razor-sharp, but in reality, the odds of those exact conditions ever coming round again are far too thin.
Remember: building a trading method isn't about drawing a pretty curve in the past. It's about giving yourself the best possible odds that the method keeps making money in the future, once you're trading it live.
So it matters that the filter stays simple. You'll find it easy to recognise these market conditions, and you won't run the risk of having selected too biased a sample.
There are all kinds of filters:
- The obvious ones, trying to quantify whether the trend is up or down.
- The ones that gauge volatility conditions.
- And then there are the ones people think about less, because trading software isn't very comfortable with them: time-of-day filters, for instance (I only trade in the afternoon, say, or the opposite, only shortly after the open).
The most common mistake is to blur the filter together with the condition that actually gets you into the market.
I see a huge number of traders explain to me that they bought because price broke above a moving average, say.
That's an entry point like any other, but why should crossing a moving average count as a good entry level? It says absolutely nothing about what's actually happening, or how it's happening.
You absolutely have to keep the filter separate: the filter is what defines how you're going to trade, and what you're actually going to do is a different question entirely.
To put it bluntly, here's what that looks like. I build a filter on volatility, then:
- If volatility is low, I buy on a small acceleration.
- If volatility is high, I buy on a very sharp one.
The goal is to give your buying strategy the best possible odds, relative to the market's general conditions.
See the difference with a signal based purely on the filter alone?
Every combination is possible.
For example, it's fairly obvious that in the morning, European markets are more inclined to correct the excesses left over from the US close the day before. So you'd favour counter-trend strategies through to the end of the morning.
In the afternoon, though, you're chasing after the US, so you're better off working with the trend.
So the filter here is the time of day, and it defines two completely different kinds of trading.
One of the few drawbacks of thinking this way is that it adds a layer of complexity to your decision process.
It's a well-established fact: the more complex a method gets, the more you can fit it to what already happened in the past, and the less likely it is to keep working in the future.
Our goal, then, is to keep both the filter and the entry conditions as simple as possible, so that a method proven against the past gives itself the best possible odds of still working in the future.
There's more than one advantage to working this way:
First, it's far easier to find signals that work once you're applying them inside market conditions that have already been filtered.
To exaggerate the point: say I hand you a filter that's perfect at telling you whether the trend is up or down, and I tell you: when the market's bullish, only trade long; when it's bearish, only trade short. You can see straight away how much more obvious the trading becomes, and how much easier success is to reach.
Second, since every strategy rests on a filter that pins down a type of market, it's far easier to find the right signals for it. If my filter tells me the market's wildly volatile, it's obvious the riskiest strategies will run you straight into ruin.
Another major advantage: since every strategy is most comfortable in one particular type of market, the one its filter defines, it's a great deal easier to diversify your portfolio.
For example, if one strategy, through its filter, trades counter-trend, it's simple enough to pair it with another that does better with the trend.
The aim is to bring your risk down.
And then there's the most important advantage of all: robustness. Because you keep the buying decision separate from the filter, it's far easier to lay down fairly simple rules for when you actually step in.
And remember: the simpler it is, the greater the odds it keeps working. That's a statistical given.
Good trading!
Part of a series. This article belongs to Market Regime: Every Strategy Is Built for One Kind of Weather. Related: Only the Exit Counts: The Part of a Trade Nobody Works On and The Value Added by a "Filtered" Strategy. Background: mean reversion in the Ouinex glossary.
About the author. Samuel Rondot has traded since 1995, self-taught. Between 2009 and 2017 he published around 300 articles on objectifeco.com, abcbourse.com and his own blog, many of them drawn from what he saw in hundreds of client accounts. Objectifeco wiped its archive in 2016; the Internet Archive kept a copy, and every article republished here links to its original. He now runs the Strategy at Ouinex.
Trading leveraged products carries a high risk of loss.
