Rejection off a level and a break of a falling trendline on small caps

The two most recognizable chart moves — selling a touch of a level that does not break, and buying the break out of a falling trendline — gave nothing on small caps, neither against a random hour nor against the market in that same hour.

Why it sounded right

Both moves read off a chart instantly, and both have a sensible explanation. Price walked up to a level where it has already been turned around, poked at it and went no further — so the sellers are back there, and selling right at the level pays: you can put protection immediately above it, the risk is small, and the turning point is known in advance. That is the cheapest trade in terms of risk that a chart ever offers.

The opposite move is just as clear. A coin slides down for weeks at a steady angle. Draw a line across the highs, and price leaving it means the seller who has been pressing all this time is done. Now the market has to find a price again, and it usually starts doing that upward. On small caps both moves look especially juicy, because the swings there are big: if the mechanism exists, it should give tens of percent, not half a percent.

There is a third argument that rarely gets said out loud. Small coins live their own lives, the big market pulls them around less. So the chart pattern should matter most exactly there — nothing else is around to override it.

How we tested it

We fixed the rules in advance, before the run, and ran them across every small cap with a few years of history on hourly candles, fees included. Then three controls, each answering its own question. First: same coin, a random hour instead of the signal. That tests whether the moment matters at all. Second: a move of the same size on the same coin, but with no level and no line — that tests whether the line adds anything to the move itself. Third: other small caps in that same hour — that tests whether we are just buying the whole small-cap sector moving.

We looked not only at the average but at the typical trade, at the share of trades that finished up, and at how far a rare but big case drags the position against you. Both halves of the window were counted separately.

How it ended

Selling a touch of the level adds nothing. A move of the same size without a level gives the same thing, a random hour gives the same thing, other small caps in the same hour give the same thing. The shape of it is unpleasant: most of the time the trade is slightly up, but now and then it gets carried a long way up, and those rare cases are enough to make the average negative while the typical trade is positive. Entering on confirmation — waiting for the next hour to close lower — makes the result worse still.

The trendline break gave the one meaningful result of the whole test, but not the one you need: it is better than a move of the same size without a line. So the line really does select the moves that pull back less. But in money it is zero: zero against a random hour, zero against other small caps in the same hour, and fewer than half the trades finish up. And the decisive part — on the first part of the window the move worked confidently, and on the second, where there are several times more signals, it did not work at all. That does not hold.

Where this could be wrong. The levels and lines here are drawn mechanically, while a live human picks with his eyes which moves to take — so we tested the mechanics, not the picking. The protective order above the level was not tested at all, and you cannot tune it on this same data. And the sample is incomplete: it contains coins that still have history today, while the dead and the delisted — exactly the ones that "lived their own lives" loudest — are absent.

What this teaches

• "The observation is correct" and "the trade pays" are two different claims. A level really does turn price around often; that does not mean selling at it makes money.
• The control "the same move, but with no level" is mandatory. Otherwise you are measuring the strength of the move and crediting it to the line.
• If a move worked on the first half of the history and stopped on the second, where there is more data, that is not "the market stopped working." It was never there.
• When the typical trade is up but the average is down, the trade is living off rare big moves against itself. You cannot judge a rule like that by the share of trades that finished up.
• Testing a mechanical rule says nothing about picking by hand. If the move lives on eyeball selection, that has to be measured separately — and before the trades, not after.