Lesson 0.4. The Honest Frame
Video version (in Russian) · 11 minutesLet me start with a number that looks like a ready-made trading system.
I took 1,332 trading days — from the start of 2020 to today. For each day I found the strike that held the largest mass of open interest near the price, and looked at what price did when it got there.
It arrived and reversed in 59 cases out of 100.
A beautiful number. Almost six out of ten. Numbers like this are what courses are built on, what signals are sold on, what arrows are drawn from. On its own it means nothing — I'll show you why in a moment, and along the way it will become clear what actually does work in options data.
A number with nothing to compare it to
The problem isn't the count. It was measured on canonical data, the rules were declared in advance, and the price is the same one you see on your chart. The problem is that this number has nothing to be compared against.
Here is the question that has to be asked: how many would there have been if the level hadn't been there at all?
That's easy to test. Take the same day, the same distance from yesterday's close — but measure it in the other direction. You get a point with exactly the same geometry and no relationship to options whatsoever. An empty spot. And you look at it through the same eyes, by the same rules.
The empty spot reversed price in 57 cases out of 100.
The difference is 1.3 percentage points. On a sample this size that is indistinguishable from chance: the probability of getting a difference like that by accident is around two thirds. Which means there is no difference.
What follows from this, and what doesn't
It does not follow that option levels are useless. Exactly one thing follows, and it needs to be stated precisely:
The rule 'the largest level reverses price' does not work.
In that exact form: find the maximum mass, wait for the touch, expect the reversal. Like that — no. Tested over six and a half years, not confirmed.
The difference between an observation and a rule is the whole point of this lesson. Observation: price often reverses at large levels. That is true. Rule: therefore, at a large level you should expect a reversal. That is false — because price reverses just as often where there is nothing at all. Markets reverse constantly; that is their normal state.
An observation only becomes a rule after it has been compared against emptiness. Without that comparison, any observation about the market is simply a report that the market moves.
What can actually be predicted inside a day
The second source of illusions is misjudging the scale of the task. Here is the EUR/USD daily move in numbers, over 1,766 days:
| what we measure | median |
|---|---|
| range of the day, low to high | 68 points |
| result of the day, open to close | 29 points |
| share of the range that reached the result | 47% |
Over the last year the numbers are more modest: a range of 58 points, a result of 24.
Read it like this: price covers around 60 points from edge to edge during a day, but less than half of that travel survives to the evening. The rest is movement there and back. On 42% of the days of the last year the result of the day was under 20 points — price moved all day and came back to almost where it started.
Which leads to this: 'guess the direction of the day' is a task that, on four days out of ten, has no clear answer at all. Direction on those days is weak and chop dominates. And a source that confidently names a direction every morning is, on those days, naming noise with exactly the same confidence.
That is not a reason to give up — it is a reason to understand what to count on. The work is not with every single day in a row, but with those market states where the map speaks clearly.
A map of interests, not a timetable of moves
Everything we have covered across three lessons comes down to one sentence: options data is a map of interests held by large participants. Where the money sits, where it arrives, where it leaves from.
A map of interests is not a timetable of moves. Another number from the same measurement: on 56% of days price never reached the largest level at all during the whole day. It stood there, it was visible, it weighed thousands of contracts — and it had nothing to do with that particular day.
The map shows where things are placed. What price will do today, it does not show.
What levels can and cannot do
They can:
- map out the terrain: where the large interests sit and where they are shifting;
- show where the sellers' risk is concentrated — from lesson 0.1 we know that it is sellers who are forced to defend themselves and who leave traces in the futures;
- give a scenario its frame: where it is confirmed and where it breaks;
- show what changed over the day: where new mass arrived, where it left from;
- pick up a tendency: a shift of interest reads quite clearly, and usually earlier than it becomes obvious in the price.
They cannot:
- give a literal prediction: 'so many points up today';
- work as a standalone simple rule like 'a level means a reversal';
- cancel out news, central bank decisions and gaps in liquidity;
- guarantee anything. The word 'guarantee' next to a market is a marker of fraud.
So what does work
The right conclusion from the first half of this lesson is not 'the data is empty', but 'there are no simple rules in it'. Not a single standalone rule survives testing: not 'a large level', not 'lots of calls', not 'open interest is rising'. We tested every one of those, and every one of them fell apart at the control.
Something else works — the combination. An options map is made of dozens of mutually contradictory factors: where the mass sits and where it is shifting, what arrives and what leaves, how the expiries line up, what the future is doing, where the sellers' risk is and where it isn't. Each factor on its own is close to noise. But their joint state at a particular moment does give you a preferred direction, and the levels that matter on that day.
That is exactly why our reports never say 'this level, therefore down'. They say 'on the balance of the map, a preference for a decline; the areas to watch are these'. That is not caution for the sake of caution — it is a precise description of what the data allows us to say.
And honestly, about the scale of it: interpreting the whole body of options data is closer to a craft learned over years than to a formula you can write out on a sheet of paper. Readability comes with practice — when the same picture passes in front of you for the hundredth time and you finally see how it differs from the previous ninety-nine. We are on that path ourselves and we promise no final destination.
Why the misses get shown too
If you only publish the good days, any instrument looks magnificent — that is how the entire signals industry works. There is only one way to check analysis: the forecast is fixed before the event, the result is compared afterwards, and the statistics are kept across every case in a row — the inconvenient ones included.
Behind the scenes the same discipline runs that you saw in the first half of this lesson. Before an observation makes it into a report, it goes through a test on history with a mandatory control: would an empty spot give the same? Most beautiful ideas do not survive that test — and you will never see them. What remains is not secret knowledge but working probabilities: an edge of a few percent, not a miracle.
The lesson worth taking away from here: a statement about the market without a control is not knowledge, it is an impression. Test any source that way, including ours.
How to read what we publish
Now you have the frame to read our wording correctly.
- 'A preference for a decline towards such-and-such levels' is a weighted scenario: the most likely development given the current map. Not a promise.
- The levels named are the areas where the scenario gets tested. The reaction to them either confirms the picture or breaks it.
- 'The map is contradictory, there is no edge' is an honest 'I don't know'. That is worth more than a forced forecast, and it will come up regularly.
And most importantly: we publish analysis, not trading signals. What to do with your own money, and when, is each person's own decision, and the risk of that decision is not compensated by anything. This analysis is not individual investment advice.
Check it yourself
The best habit you can take from this course: write down the forecast and the outcome.
Record what was said in the morning. Compare it in the evening. In writing, not from memory: memory keeps the forecasts that came true and quietly loses the ones that didn't.
A few weeks of a table like that will tell you more about a source than any testimonials, and at the same time it will show you the true scale of the task: how many days were actually predictable at all.
What's next
Stage 0 is complete. You know who creates option levels and why sellers are forced to defend themselves; how an option works and what its four coordinates are; what open interest is and how it differs from volume; and the frame all of it operates in.
Stage 1 is reading practice: we take our screens one by one and go through every column. The first lesson is about the basis: why the strike prices in the tables do not match your spot chart, and how to convert them properly. That single correction is what separates a level that 'missed by 25 points' from a level that worked exactly.
And after that comes the combination of factors this was all built for: how to read the map as a whole rather than as isolated numbers.
Next lesson — 1.1 Basis: why the strike doesn't match your chart. New lessons appear here every Wednesday and Saturday. Meanwhile: the live option flow map.