Not an entry — a market regime filter to place in front of everything else, which stops the
strategy on days it should not be trading at all.
1. The rule
# BUY — regime gate in front of your own conditions
if DayAmount > 10000000 and Volatility(1800) > 0.3 and Volatility(1800) < 3.0:
if Price > 0: # replace this line with your entry condition
Buy()
Too quiet and there is nothing to trade. Too wild and your assumptions about how far price travels
stop holding. Either way you end up blaming the strategy for what was really the weather that day.
2. Why a filter that does nothing is worth having
Most strategies are tested in one market regime and then traded in every regime. A gate that
refuses to trade outside the conditions you tested in is the cheapest protection available, and it
costs nothing to add because it sits in front of your existing entry.
The band matters more than the floor.
| Bound | What is happening | Why not to trade |
|---|
| below 0.3 | nothing is moving | your target will never be reached and the spread is the trade |
| above 3.0 | price is moving violently | stops are hit by noise rather than by being wrong |
The strategy was tested somewhere between those two. Trading outside them is not running your
strategy — it is running an untested one that happens to share its code.
3. Why this is worth more than it looks
The gain from a regime filter does not show up as a higher win rate. It shows up as trades you did
not take, and each avoided trade saves a round trip.
That matters more than most people expect. On the US movers population we measured, a round trip
costs about 0.75% (
what day trading a US stock really costs). A typical 2% target needs
a 33% win rate on paper and 58% after that cost. On numbers like those, not trading is a much
larger lever than trading slightly better.
4. When it fails
The band can be fitted to the sample like anything else. If your strategy only works with
volatility between 0.31 and 0.34, you have not found a regime filter — you have found a way to
select the days that happened to work. See
the same entry at three sensitivities for the test that catches this.
5. What to change first
Widen the band until the strategy starts losing, then step back one notch. A filter you cannot
loosen without breaking the result is not a filter, it is the strategy.
Test the widening on your own recordings rather than reasoning about it:
how to backtest a day trading strategy on US stocks.
Educational template for research and backtesting. Not investment advice and not a signal service.
