The market conditions filter
Lesson 6 · about 8 min
A setup says what a trade looks like. A filter says whether today is a day to look for it at all. They are separate lines on the plan because they answer separate questions, and because the most common way a working setup stops working is that the conditions around it changed while the setup itself looked the same.
Why the setup alone is not enough
A breakout setup on an index future behaves one way when the overnight range is wide and the day has a data release, and another way when the overnight range is a few points and nothing is scheduled. The bars look the same. The follow-through does not. If your journal records both as "the setup", your win rate is an average of two different things, and neither number is real.
The filter separates them before the trade, so that the journal separates them after.
What a filter measures
A filter is a property of the whole market, or the whole instrument, that is known before the session starts or at least before the setup appears. Common ones:
| Filter type | Example | Known when |
|---|---|---|
| Range | Overnight range >= 0.4% of price | At the open |
| Volatility | 20-day ATR between 0.8% and 2.5% of price | Previous close |
| Volatility rank | Implied volatility rank >= 30 | Previous close |
| Trend regime | Index above its 50-day moving average | Previous close |
| Calendar | No scheduled top-tier data during the session | Days ahead |
| Liquidity | Average daily volume over 1M shares | Previous close |
| Session | London and New York overlap only | Clock |
A filter is binary. Either the day qualifies or it does not, and if it does not, you do not look for the setup. You do something else, which is usually nothing.
One or two, not five
Every filter you add cuts the number of trading days. Two filters at 60% pass rate each leave you with 36% of days. Three leave 22%. If you stack five, you will trade twice a quarter and never build a sample. Start with one filter, the one you most suspect matters, and add a second only if the journal shows the first is not enough.
Key idea: The setup describes the trade. The filter describes the day. Log both, or your statistics will be an average of two different markets.
Filters are for logging, not just gating
Even if you are not sure a filter matters, record its value on every trade. "Overnight range as % of price: 0.31" costs five seconds to write down and is the thing that lets you discover, forty trades later, that every loser came on a narrow-range day. Module 4 has a field for this, and Module 5 shows how to turn that observation into a rule change.
The difference between a filter on the plan and a filter you are merely logging: the one on the plan blocks trades. The one you are logging does not, yet. Be careful not to let a logged filter become a "soft" rule you apply on the days you feel like it. Either it is on the plan or it is not.
The no-trade line, revisited
The NO-TRADE line from Module 1 is a set of calendar and clock filters. Data releases, holiday sessions, the first minutes after the open, the last minutes before the close. They are on a separate line because they are absolute: there is no version of the setup that overrides them. The market conditions filter is more specific to your setup and is the one that will change between plan versions.
A worked example
A trader runs a pullback setup on EUR/USD in the London session. First month, 22 trades, 45% win rate, average winner 1.6R, average loser 1R. Expectancy positive but thin: 0.45 × 1.6 − 0.55 × 1 = 0.17R per trade.
She has been logging the previous day's range as a percentage of price. Sorted by that field:
| Prior-day range | Trades | Win rate | Avg R |
|---|---|---|---|
| Under 0.5% | 9 | 22% | −0.41 |
| 0.5% and over | 13 | 62% | +0.57 |
Nine and thirteen trades are far too few to conclude anything, and Module 5 will insist on that. But this is exactly the kind of split that tells her what to keep logging, and what the first candidate filter should be if the pattern holds at 60 trades.
Try it: Choose one filter from the table that you suspect matters for your setup. Write it as a number on the plan's FILTER line. Then choose a second one you are curious about and add it as a journal field only. In a month, sort your trades by the second field and see whether there is a split worth watching.
Recap
- A filter is a property of the day or the market, known before the setup appears, that decides whether you look for the setup at all.
- Filters are binary and belong on their own line so that the journal can separate qualifying days from non-qualifying ones.
- One or two filters. Each one cuts your trading days; five leave you with no sample.
- Log candidate filters on every trade even before they are rules, but do not let a logged filter become a rule you apply only when you feel like it.
- The NO-TRADE line holds absolute calendar and clock filters; the FILTER line holds the setup-specific one that will evolve.
See it drawn
Original diagrams for the ideas on this page. Illustrative, not real market data.