Novice pair traders often choose diverging instruments, leading to unstable results.

Aleksander 2026.07.10 06:01 #2901

The main mistake made by novice ‘pair traders’ is to choose either the strongest and weakest, or the highest and lowest instruments… so to speak, if they’ve diverged, they’re likely to continue doing so… The key to selecting instruments is to choose converging pairs, roughly 4–6 per cent of the maximum spread between them, to put it simply....

Not the Grail, just A topic for traders. [Archive!] WRITING A COUNTRY

Aleksander 2026.07.10 06:03 #2902

By the way – it doesn’t matter where these instruments are on the ‘aggregate chart’

mvf358 2026.07.10 07:21 #2903

Aleksey Nikolayev # :

I also asked about stop-loss. AI's response:

In cointegration and mean reversion strategies, a classic stop-loss of a fixed number of pips for individual pairs doesn't work . If one currency continues to rise and the other continues to fall, your synthetic spread will go into deep loss.

Since you are trading the entire basket as one instrument, the stop loss must be calculated mathematically for the entire portfolio as a whole .

Here are 4 main methods for setting stop losses when trading cointegration:

1. Stop-loss by Z-score (Statistical stop)

This is the most popular and logical method. The deviation of your cointegration spread is measured in standard deviations (Z-score).

Trade entry: When Z-score = ±2.0 (the price has deviated significantly from the average).
Stop loss: When Z-score = ±3.0 or ±3.5.
Logic: The probability that a stationary series will deviate beyond 3-3.5 standard deviations is extremely low (less than 1%). If this happens, cointegration has broken down (the market has changed, and the fundamental links between currencies have been broken). There's no point in continuing the trade—you need to lock in the loss on the entire basket.

2. Time-stop

A stationary spread has a mathematical parameter called half-life . It indicates the average time it takes for the spread to return to its mean value after a deviation.

Logic: If the average return time is 24 hours, and your portfolio has been in the red for 72 hours (3 half-lives) and nothing has changed, then the model has stopped working.
Exit: All positions are closed at market price after the specified time, regardless of the current financial result.

3. Equity Stop (Money Stop on the Basket)

The easiest method to implement for trading robots.

Logic: You calculate the total floating profit/loss (Equity) specifically for those orders that are included in this cointegration portfolio.
Exit: If the total loss for all open currency pairs exceeds, for example, 2% or 3% of your total deposit, the robot forcibly closes the entire grid of orders with one button.

4. Stop-loss based on the historical maximum spread

Before starting trading, you analyze the spread history over the past few months.

Logic: You find the point of the deepest historical divergence (the maximum peak of the spread in history).
Exit: The stop-loss is placed slightly above this historical high. If the market breaks this historical high, the current movement is abnormal and the pattern has broken.

Important rule: No stops inside the basket!

It is extremely important not to set separate stop losses for each currency pair .

Example: If the EUR in your basket rises and the GBP falls, and a separate stop-loss is triggered on the EUR, your hedged contour is completely destroyed. You're left with an open, naked GBP position, which could result in an even greater loss. The basket opens and closes simultaneously (both by take-profit and stop-loss).

Well, there it is, a scientific description of the Sinus method. But everyone weaves their own basket.

Aleksey Nikolayev 2026.07.10 09:13 #2904

mvf358 #:
Where is there room here for Roman, the great mathematician whom I banished?

Do you not follow the rules of the Russian language on Fridays?

Roman has taken the theory a step further – dynamic portfolios using the theory of stochastic process filtering.

mvf358 2026.07.10 09:31 #2905

Aleksey Nikolayev #:

Do you break the rules of the Russian language on Fridays?

Roman has further developed the theory – dynamic portfolios using the theory of stochastic process filtering.

We need to focus on developing practical application, rather than theory, which takes us away from reality. Whenever I asked him to demonstrate his theory in practice and make a profit, he would put on a show and backtrack. For now, his theory remains just that – a theory – and at present it has nothing to do with Sinus’s subject matter.

From theory to practice Theorem on the presence A great book on

Vitaly Muzichenko 2026.07.10 11:27 #2906

Aleksey Nikolayev #:

I also had a chat with the AI, asking it a question:

There’s a trading strategy like this in Forex. We take the eight major currencies and rank them by ‘strength’ (the more a currency rises relative to the others, the stronger it is). Then we take the two strongest and the two weakest. We trade one strong-weak pair in the direction of the shift in strength (selling the strong one against the weak one), whilst hedging with the other strong-weak pair. Question: Isn’t this a special case of trading multiple cointegration between these eight currencies? Would it be better to use the Johansen test for this type of trading?

AI’s answer:

Yes, the approach you have described is a special case of trading based on multiple cointegration (or trading in currency baskets / stationary portfolios), and the Johansen test is indeed much better suited to this, although it completely changes

...

Well, that’s something at least.

Now carry on talking to him; get him to build that very portfolio and show you.

No matter how many times I’ve asked, he builds something completely different from what I’m asking for, and describes what he’s actually building. But that’s not what’s needed at all.

Why have subscriptions been Adviser to the whole FOREX - Trends, forecasts

mvf358 2026.07.10 12:18 #2907

Vitaly Muzichenko #:

Well, that’s something at least.

Now carry on talking to him; get him to put together that portfolio and show it to you.

No matter how many times I’ve asked, he builds something completely different from what I’ve asked for, and describes what he’s building. But that’s not what’s needed at all.

AI isn’t yet ready to build the right portfolios on its own. You need to select the portfolios yourself based on the actual performance history of the equity portfolio.

Aleksey Nikolayev 2026.07.10 14:14 #2908

Vitaly Muzichenko #:

Well, that’s something at least.

Now carry on talking to him; get him to put together that portfolio and show it to you.

No matter how many times I’ve asked, he builds something completely different from what I’ve asked for, and describes what he’s building. But that’s not what’s needed at all.

Many people have looked for multiple cointegration on major currency pairs, but as a rule, nothing interesting or stable has come of it. My limited experience confirms this. Dreamer confirmed that the topic is a dead end. The moderator above also made a slightly sarcastic remark, assuming that I was claiming that searching for cointegrations is profitable.

On the unequal probability Not the Grail, just Econometrics: why co-integration is