How to Use LuxAlgo
To use LuxAlgo, put it on one clean chart, turn off every feature except the one you can state a use for, and record what it signalled for at least a month before changing anything. The tool marks conditions; the decision about what a condition means stays yours.
A signal suite marks conditions on a chart faster than you can find them. That is genuinely useful when you already know which condition you want marked, and it is a substitute for a method when you do not.
Before you start
A rule of your own that the tool is meant to help you execute, written before you subscribe. Then the question becomes whether it marks that condition, which is answerable.
One chart with nothing else on it, so the output can be judged in isolation. A suite added on top of five existing indicators cannot be evaluated at all.
A record of what it signalled, because the only useful evidence is your own. A spreadsheet with the date, the instrument and what happened next.
The steps
1. Write down the decision it is meant to speed up
“Mark me every failed breakout on my watchlist.” That is a job. “Tell me when to buy” is not a job, it is the whole of trading handed to a script.
2. Turn every feature off before turning one on
The default configuration displays a great deal at once, which is how a suite demonstrates value. Only one thing at a time can be evaluated.
3. Run it on one instrument you already know
Scroll back over bars you have already traded or studied. If its marks disagree with your reading there, you have learned something concrete about the tool in ten minutes.
4. Log every signal for a month before changing anything
Date, instrument, signal, what happened next. On this site’s shared series 52% of 571 single bars finished higher, so a run of correct-looking marks is what the base rate produces on its own.
5. Take the stop from the chart, never from the signal
A mark says a condition occurred. Where the idea is wrong is a structural question, and that distance is what sets the position size.
6. Resist the confluence settings
Requiring several components to agree makes the historical chart look far cleaner. It also removes most signals, and the ones it removes are chosen by what happened afterwards.
7. Judge it against your own log, not the marketing
After 30 days you can say whether it marked things you would have missed. That is the question, and nobody else’s screenshot can answer it for you.
How to tell it worked
You can state the 1 decision it speeds up, in a sentence written before you subscribed.
At most 2 features are enabled, so any effect is attributable.
A log covering at least 30 days exists, with what happened next recorded against each signal.
And every trade taken from a signal had a stop from the chart, so the size came from structure.
What a signal cannot contain
It cannot contain your risk. On this site’s shared series a round trip measures about 2% of the
median bar range of 0.493, and a suite that signals often charges that repeatedly. The figures are in
research/series-measurements.json.
It cannot see whether the instrument is fillable. The mark looks identical on something nobody trades, and that is where the difference between a signal and a trade lives.
What the subscription is actually buying
Speed on a condition you already look for. If you scan thirty names for one pattern, a tool that marks it is buying back real time every morning, and that is a defensible reason to pay.
Not a method. The features that appear to supply one — a directional bias, an overall confluence score — are the parts most likely to be describing the history they were tuned on.
And not an exemption from record-keeping. A paid tool with no log is the same evidence position as a free one with no log.
How to test whether it is adding anything
Run it alongside your existing process for a month, without acting on it. Mark the trades you would have taken anyway and the ones only the tool found. Two lists, kept separately.
Then ask whether the second list contains anything. If every signal you would have acted on was already on your first list, the tool is confirming rather than finding, and confirmation is the one service you can supply yourself for free.
If the second list has entries, look at what happened to them. A handful of found trades that went nowhere is a different answer from a handful that worked, and both are answerable from a month of records.
The comparison only works because you did not act on the second list. Trading both lists makes the result a fact about your execution rather than about the tool.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 88 mention this suite in the
title, at a median of 9,475 views across 41 channels, and 62% of those titles are instruction-shaped.
Trading signals generally appear in 24 at 24,268. The counts come from site/corpus_count.py and
site/rank_howto.py.
88 videos at 9,475 across 41 channels. A concentrated, well-covered subject — and the coverage is overwhelmingly demonstration on historical charts, which is the one form of evidence that cannot distinguish a useful tool from a well-fitted one.
The answer to the question on that chart is that a historical chart is where every tool looks excellent. The marks are placed with the outcome already visible. On this site’s series 54% of 566 ten-bar windows finished higher, which means a bullish mark is right more often than not before any skill is involved.
When it fails
The failure is the tool quietly becoming the method, and it happens without a decision. You subscribe to speed up a rule, the suite signals something your rule does not cover, the trade works, and within two months you are trading its marks. There was never a moment where you chose that system, and you cannot state what it is or when it stops applying — which means there is nothing to review when it stops working.
The second failure is enabling everything. Nothing can then be attributed.
A third is confluence mode. Cleaner history, fewer signals, chosen by outcome.
A fourth is taking a stop from a signal. Signals do not carry price levels.
A fifth is judging it on screenshots. Historical charts flatter every tool equally.
And a sixth is not logging. Without a record, month three is the same guess as day one.
Related
Trading signals covers what a signal can and cannot carry. Do indicators work is the honest version of the underlying question. And choosing indicators is how to decide what earns chart space.
The question I keep coming back to is what decision the tool is speeding up. If I already have a rule and it marks the condition faster than I can, that is real value. If I did not have a rule and the marks are supplying one, then the tool is not helping me trade my method — it has become the method, and I never chose it.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.