I Tested the Range Trade and the Retest

A range trade buys the low and sells the high of the same band, so a rising market cannot flatter it the way it flattered the breakout.

In the breakout video the trade came out slightly positive, and one split showed where that came from. All of the edge was on the long side, in a window where six instruments went up. So the number was partly the market, not the setup. This video takes the one trade that cannot have that problem.

Buy at the low, sell at the high — the same band. It has no direction by construction. And every return to an edge is a retest, of a line where price has already turned. By the end you'll know which number to fix, by measuring your winners in stops.

Which makes it a test of two things at once. Does it work, and is the direction-neutrality real when you measure it rather than assume it.

First, what counts as a range. And the first honest thing in this video is about that definition.

My first attempt said: a range is when the band of the last twenty days is no wider than three daily ranges.

That produced thirty nine signals across six instruments. SPY had none.

No signals is not a measurement. And loosening the threshold until the sample is big enough is fitting the definition to the answer.

So the threshold comes from the instrument itself. A range is when the twenty day band sits in the narrowest third of that instrument's own bands.

Nothing is chosen by hand, every instrument gets the same rule, and an edge touched again while the first trade could still be open counts once. The sample is a hundred and seventy six signals instead of thirty nine.

One more condition: both edges have to have been touched inside those twenty days. Otherwise a band that price slides across diagonally counts as a range, and there is nothing there to trade.

The control group is the same touch on the same kind of edge when the band is in the widest third. Same question, different market.

Then the same three decisions as the bounce and breakout videos. Where to enter, where the stop goes, how to get out.

Entry: at the edge, where a limit order would sit — or the next morning, if the day closed back inside the band.

Stop: half a daily range or a full one beyond the edge, or behind the extreme of the touching bar.

Exit: the opposite edge of the band, a fixed target of two stops, a time exit, or a trail.

One combination is excluded before anything is measured. Entering at the touch with a stop behind the extreme of the touching bar is impossible.

The extreme of a bar is only known once the bar has closed, and the order sits inside the day. That combination returned exactly minus one R in the bounce video and looked like a measurement.

Now the headline, and it is not what the topic promises.

In a range, the average across every legal combination is plus zero point zero one of an R per trade. In the control — the same touches when the band is wide — minus zero point two six.

That looks like a range helping, by a quarter of an R. It is not a gap this many signals can tell apart — and computing the same threshold from a hundred and twenty days of history instead of sixty turns it around: minus zero point zero two in a range, plus zero point three two in the control.

That is the whole premise of the setup, and this data cannot say whether it holds.

Before anything else, the direction check that broke the breakout video.

Longs, at the bottom of the band: plus zero point one. Shorts, at the top: minus zero point zero nine.

The longs make a little, the shorts lose a little. The side the rising window helped did better, as in the breakout video — but by a gap well within what this many signals can tell apart.

Which is worth stating plainly. The construction promised a direction-neutral trade, and this many signals can neither confirm that nor rule it out. And on average, in a range, the trade made about nothing. Those are two separate questions.

Now the decisions, and here the accounting has to be careful.

The bar stop only exists with the next-day entry. So a table comparing stops across all entries would be comparing entries, not stops.

That is the error that broke three versions of the risk video. Each decision is measured inside one entry.

Entering at the touch: minus zero point zero nine. Entering the next morning, after the day closed back inside: plus zero point one five.

Confirmation wins here, on the same stops — as it did in the bounce video, but by far more. Note that the breakout went the other way — there, waiting cost you. So this is not a rule about waiting; it is a fact about this setup.

Inside the next-day entry, the stops. Half a range: plus zero point two three. A full range: plus zero point zero five. Behind the bar: plus zero point one eight.

And the exits, still inside the next-day entry. The opposite edge: plus zero point one eight. A time exit: plus zero point three two.

A trail: plus zero point one two. A fixed target of two stops: minus zero point zero one, the worst of the four exits.

I had an explanation ready for that, and it was wrong.

My explanation was that a range has a ceiling, and a target of two stops does not fit inside the band.

Edge to edge: seven and a half tight stops at the median, under four full-range ones. From the next-morning entry, where the exits were compared, a two-stop target fits in eighty seven percent of trades with the tight stop, and sixty six percent with a full-range stop.

So the room is mostly there with the tight stop, and the explanation covers only part of it. The outcomes say the rest.

The same trades, one exit against the other. With the fixed target: stopped out sixty percent of the time, target reached twenty eight percent.

With the opposite edge: stopped out sixty one percent of the time, a large win twenty one percent.

Almost the same stop rate. The difference is not how often you are right — it is how much you are paid when you are.

Capping the win at two stops when the band is seven and a half stops wide throws away the part that makes the trade survive its own stop rate.

Now one more condition, and it points the same way.

Inside the ranges, split at the median width. The narrower half: minus zero point one five. The wider half: plus zero point one nine.

The tighter range came out worse — which sounds backwards, though the gap is within what this many signals can tell apart. Still, the obvious suspect is the stop.

The stop is half a daily range beyond the edge. But even in the narrower half the band is about six of those stops wide — so the stop is not the reason, and I do not have an explanation I can measure.

That may not be an effect at all. With this many signals, a gap this size is a lead, not a rule.

Then the half-window test again.

Chosen on the first half — entry at the touch, a stop of half a range, a time exit — plus zero point three six of an R over seventy seven trades.

The same rules on the second half: minus zero point two seven over ninety one trades.

It did not hold either; by now that is the normal result.

Three limits. All of it is before costs, which the risk module measured at zero point zero three to zero point one two of an R per round trip, and that is more than most of the numbers here.

Two: a twenty day band is one choice among many, and a range on the weekly chart is a different object. Three: only six instruments and one window.

So, what to go and do, and it takes about half an hour.

Take your last thirty winning trades and measure each one in stops, not in money. Divide the profit by the distance to the stop you actually had. No trades yet? Take thirty setups from a chart six months back, write down entry, stop and exit as if you had taken them, and measure the winners the same way.

If the median is under one stop, your win rate has to be above half just to break even — and now you know which number to fix. The range trade measured about zero, and whether it is direction-neutral this data cannot say. Educational content only. Nothing here is financial advice.