Why I stopped predicting and started measuring.
I turned £800 into £1,300 in a month, then gave it all back in an hour.
I was eighteen. It was my own money in a live FX account, and I had spent months learning the market properly before putting any of it at risk. For four weeks it worked. Then one session I stopped following what I had built and started following what I felt, and the whole month was undone before the afternoon was out.
What bothered me afterwards was not the money. It was that I could not explain what had happened in a way that would stop it happening again. I had no record of what my method actually was, no measurement of whether it had ever worked, and no way to tell the difference between a bad run and a bad system.
I had opinions about the market. I had no evidence about myself.
So I stopped buying courses and started measuring. Not because measurement is glamorous, but because it was the only thing I had not tried.
It is not a maths degree. It is a method.
Relating to, measuring, or measured by the quantity of something rather than its quality.
From the Latin quantitas, "how much". Its opposite is qualitative, which describes a thing by its character. A qualitative account of a trade is that it looked strong. A quantitative one is that price closed 2.3 standard deviations above its twenty period mean.
Applied to trading, that gives a working definition with nothing to do with hedge funds or physicists:
Quantitative trading means every decision is expressed as a measurable quantity, so it can be tested before you risk anything on it.
Which makes the real difference not intelligence, and not access to expensive data. It is whether a decision survives being written down.
"I go long when the market looks strong" cannot be tested. It changes shape depending on how you feel when you read it. "I go long when price sweeps the prior session low and closes back above it within three bars" can be tested, and inside an hour you will know whether it has ever worked.
A system, in this sense, is not software. It is a set of statements precise enough that two people reading them would take the same trade. Automation comes later and only matters once that exists.
Most retail methods are the first kind. That is why somebody can follow the same strategy for two years without ever establishing whether it works.
The second column is not harder. It is less enjoyable, and nobody sells a course on it, because there is nothing left to keep buying afterwards.
The most useful thing I know about trading.
Earlier this year I set out to prove a volatility fade setup worked on the one minute chart. I marked every trade by hand rather than automating it, because I did not trust a script to be honest with me.
Seventeen hours of clicking. 1,398 trades.
Four out of five trades correct, and the money still went the wrong way.
The average winner came in under five pips. Spread and commission together cost roughly three pips per round trip. An 81.5% win rate on a four pip edge against three pips of cost is not an edge. It is a rounding error you are paying a broker to generate on your behalf.
What matters is expectancy, and it takes ten minutes.
// W win rate as a decimal 0.815 // Aw average winning trade in pips // L loss rate 1 − W // Al average losing trade // C full round trip cost spread + commission E = (W × Aw) − (L × Al) − C // the one that caught me E = (0.815 × 4.2) − (0.185 × 6.1) − 3.0 E = 3.42 − 1.13 − 3.0 E = −0.71 pips per trade if E <= 0 then "does not survive execution"
Run it tonight on whatever you are trading. If what is left is at or below zero, the win rate does not matter, the entry does not matter, and no indicator will fix it.
A system that works does not work all the time.
This is the second thing that took me too long. A method has conditions it was built for, and when those conditions are absent it does not get slightly worse. It stops working.
Most losses from a system that genuinely has an edge come from running it in the wrong conditions, not from the system being wrong.
So the question before a trade is not only whether this is a setup. It is whether this is the kind of market the setup was built for. And that question is measurable: volatility either sits near its own recent average or it does not, trend strength is either present or it is not. Combine the two and market quality has a grade.
The uncomfortable part is that a run of enormous candles usually pushes the grade down, not up. Volatility that has broken away from its own average is not opportunity. It is the market doing something other than what your system was measured against.
Every instrument I build reports this grade, including when it is bad, and including when it is bad for my own configuration.
Certainty, and then dependency.
Trading education has an unusual problem. The honest version of the product is unsatisfying, and the unsatisfying version does not sell, so almost everybody sells the other one.
The honest version says: here is a method, here are the conditions it needs, here is where it fails, here is the drawdown you will have to sit through, and here is the arithmetic that decides whether it survives your broker's costs.
The version that sells says: here is a setup that works.
Notice what follows. If a strategy comes with no stated failure mode, every loss must be your fault. You were undisciplined. You lacked conviction. You did not follow it properly. The product can never be wrong, which is precisely why it keeps selling.
The second problem is dependency. A signal you cannot generate yourself. A group you have to keep paying to stay inside. A setup you cannot evaluate without somebody confirming it. None of those leave you owning anything, and the day you stop paying you are back where you started, only older.
The test I would apply to anybody selling you something, including me:
If the answer is more, you are not buying an education. You are renting a feeling.
A research and software company. Not a signals group.
I build instruments that measure the market rather than predict it, publish what I find whether it flatters me or not, and hand people the tools to check my work on their own instrument rather than take my word for it.
Every system we publish is one we run and test ourselves, and every result arrives with its boundaries stated first.
Nothing here forecasts. The instruments describe what has happened and what conditions currently exist. What you do with that is yours.
One of our instruments stops working above the fifteen minute chart. That is written into the material rather than buried in a footnote. One of the research papers grades a configuration of our own work as DEGRADED, and it stayed in.
The Build system hands you components rather than a finished chart, because a chart configured by somebody else is a chart you cannot fix.
That last one matters most. Every claim in this document is checkable inside the product, on your instrument, with your data. If the reports contradict my defaults on your pair, believe the reports.
You do not need any of this to use what you have just read.
The expectancy formula is yours. Run it tonight on whatever you are trading. If the answer is negative, you have learned more in ten minutes than most people manage in a year, and it cost you nothing.
If you want the rest, Research opens the library for thirty days without charging anything.
And if a month goes by and you have not opened anything, cancel before you are charged. I would rather have fifty members who read everything than five hundred who forgot they subscribed.