Why We Killed the Press, and Four Other Ideas We Loved

The most useful page in any research log is the list of things that did not work. Here are five from the last month of Aqueduct work. Each one was an idea we liked, each one looked good on first pass, and each one died in the same place: a fair test. No figures; the lesson in each case is the shape of the mistake.

1. The press

"Press" is our word for adding to a position that is already working: a second ticket once the first is in profit, so a good trade becomes a bigger one. It is a beloved idea in trading and for weeks it looked like the best thing in the harness. Then we changed how the harness orders events inside a bar. On a coarse tape, the press was being credited with fills that, at five-minute resolution, could not have happened in that order; the add was filling before the stop that would have taken it out. On the finer tape, every version of the press did worse than no press at all. The idea was not wrong. The measurement had been flattering it. We turned the press off on every account we run.

2. Standing down after a sweep

We went through the cycles that ended at the floor and found a pattern: many of them had a day-two liquidity sweep on the same side as the trade. So we tried a rule: after a sweep like that, stand down for a while. It did nothing. The sweeps were real and the losses were real, but the sweeps did not *predict* the losses; they were just present in them, the way rain is present in car accidents. A pattern you find by looking only at the losers has to be checked against the winners before it becomes a rule, and in the winners the same sweeps were everywhere.

3. Averaging in

Instead of one entry at one level, a small ladder of entries, so a trade that goes against you first gets a better average price. It reads as prudent. In the harness it was the same bet with more exposure, and the twin averaged in just as happily as the real rule did. Averaging in does not add information; it adds size to the trades you were least sure about, which is precisely backwards.

4. The pre-filter

A condition checked before the order is even placed, to skip levels that looked poor. It showed a small, consistent improvement for two days. Then we made resting fills gap-aware, so an order no longer fills at a level the market has already run through. The improvement vanished. The pre-filter had not been finding better trades; it had been avoiding a fill artefact that should never have existed. When a refinement's benefit disappears the moment you fix the plumbing, the refinement was measuring the plumbing.

5. A second logic on the side

A separate rule of the founder's, a genuine reversal fade rather than the continuation model Aqueduct runs, put alongside Aqueduct on separate instruments and separate slots to smooth the record. Each was fine alone. Together, the combination was no better than Aqueduct by itself, because they were often right and wrong on the same days. Two edges only add if they disagree, and these two agreed too much.

What all five have in common

Every one of them was a human idea, which is why they were worth testing. Every one of them was mechanical, which is why the test meant something. And every one of them was caught by the same discipline: a fair null, a fine enough tape, fills the market would actually have given, and the honesty to write the result down. The twin test is that discipline. If you build your own tools, keep a page like this one. It will be the most valuable page you own.

Kingdom Portfolios LLC is an independent education publisher. It is not registered with the CFTC or the NFA, is not a Commodity Trading Advisor, does not offer or manage investments, and does not trade anyone else's capital. Nothing here is investment advice, a signal, an offer, or a solicitation.

No performance results of any kind are presented in this note and none should be inferred. Where this note refers to our lab, our harness or our measurement, it means simulated research on historical data. Hypothetical performance results have many inherent limitations: they are prepared with the benefit of hindsight, they do not involve financial risk, no hypothetical record can completely account for the impact of financial risk in actual trading, and no representation is made that any account will or is likely to achieve results similar to anything described. Past performance is not indicative of future results.

The algorithms described trade only Kingdom Portfolios' own accounts. They are not for sale, not licensed, and not run for anyone else. Trading forex and futures carries substantial risk of loss, including loss of the whole account. Education only.