Not three strategies — one entry at three settings, posted together so you can see what the
parameter actually does. This is the cheapest test there is for whether an idea is real.
1. The three versions
# Tight — fewer trades
if ChangeAngle(60) > 30 and Strength > 130:
Buy()# Middle
if ChangeAngle(60) > 20 and Strength > 115:
Buy()# Loose — many trades
if ChangeAngle(60) > 12 and Strength > 105:
Buy()2. How to read the three
Run all three over the same period and compare
trade count first, result second.
| Pattern across the three | What it means |
|---|
| results change smoothly | the parameter is describing something real |
| the middle is dramatically better than both neighbours | you found a coincidence |
| only the tight version works | a small sample of a good period looking like skill |
| all three lose | at least you learned it cheaply |
What you are looking for is whether the middle and loose versions still work. An idea that survives
being loosened has something behind it. An idea that works at exactly one threshold found a pattern
in your sample — see
eleven conditions, four trades a month for where that ends.
3. Move the window with the angle
The angle and the window are not independent. Holding
ChangeAngle's threshold fixed while you
widen its window is not a mild change, it is a much stricter condition — a slope in degrees over a
longer stretch asks for a far larger move.
Both were moved when these three were swept. Otherwise the least sensitive version simply never
trades, and it is very easy to read that as a result rather than as a mistake.
4. When this test misleads
If all three versions share a condition that is itself overfitted, loosening the other numbers
proves nothing. Vary one dimension at a time.
It also misleads if the three versions produce very different trade counts and you compare average
return without noticing. Ten trades at +2% and two hundred at +0.3% are not comparable results, and
after costs they are not even the same sign — a round trip on this population costs about 0.75%
(
what day trading a US stock really costs).
5. What to change first
Nothing here. This file is a test, not a strategy.
Take the version that survives, then give it proper exits and a liquidity floor before you consider
running it —
exit rules every day trading strategy needs and
the liquidity filter. Run the sweep on your own recordings:
how to backtest a day trading strategy on US stocks.
Educational template for research and backtesting. Not investment advice and not a signal service.
