AI ORASignal - First test results (we are always in profit)

Hello everyone!

It's quite difficult to talk about our AI without revealing its secrets, but we still need to share something

After a year of research, we decided to bring our AI into the public space. For the past five months, ORA has been working 24/5, generating live signals. You can check the public accounts where the early versions are presented, as well as the new separate working bots which, in our opinion, are now fully ready for serious use.

The AI is built on a series of key hypotheses, which we obviously cannot disclose in detail. However, after a significant period of time, one thing is clear — they work.
For example, the key metrics: win rate never drops below 92%, and the Profit Factor is consistently above 2.0 (to put it modestly), with real values often between 3.0 and 4.0.

Important note: ORA resembles a scalper, but it's not exactly that. It can trade on longer timeframes, but right now we need statistics from a large number of trades (hundreds), which can only be obtained on short intervals. In other words, research progresses much faster when results are known quickly.
De facto — yes, it is a scalper.

So what does ORA actually do? It thinks, calculates using our special metric, determines the best entry points across 17–19 different Forex pairs, chooses direction, and places lots. At the moment of placement, it immediately publishes the signal in our Telegram channel.
We have also separately tested a manual replication strategy on a dedicated account — i.e., what happens if our subscribers simply follow ORA’s signals. It works as well.

It’s important to understand that ORA finds entries with a certain reserve (edge). We talk about this regularly on our X account. Moreover, manually repeating after it can sometimes be even more profitable — for example, by entering at an even better price if the quote moves against the initial signal for a while.
Currently, all ORA signals are completely free, as we are still in the research and data collection phase.
Therefore, we are very interested in your honest opinion.

To be frank, we are somewhat puzzled ourselves. It works too stably and confidently.
Of course, knowing the hypotheses and the metric behind ORA, we expected good results (unfortunately without going into details), but still — in our opinion, it’s almost too good, which is why we keep testing and trading non-stop

What do you think? Is this just another ordinary system, or does it have serious potential?

Thank you for this wonderful sharing. First of all, both your courage in making such a comprehensive research process public and your transparency in sharing the results are truly commendable. Suggestions for Making the System and AI Usage Methods More Explained: I think it's necessary to encourage many developers and followers on how to navigate the world of AI.

Thank you for your kind words.
Regarding the transparency of the results - this is indeed the key point for us.
From the very beginning, we have been building an AI that completely excludes the logic of “if not today, then maybe tomorrow we’ll get lucky.”
That’s why showing real, live results allows us to evaluate everything honestly and transparently. In other words - either it works, or it goes in the trash and we try something new
By the way, regarding AI developers: Interestingly, attempts to apply AI and machine learning in trading have been going on for over two decades now, maybe even longer.
What’s strange is that after reading a huge number of scientific papers and research articles, we have never come across an approach similar to ours. Almost everyone seems to be approaching the problem from the wrong side.
The market (a chaotic multitude) simply cannot be taken head-on )))

For example, this is exactly how our AI works — the very thing we talk about in our X posts.
No one knows what will happen tomorrow (the future doesn’t exist), but up to this moment, ORA continues to work exactly like this

#EURJPY.
Here's another interesting case from our AI today.
We always try to study the reserve.
Important clarification: don’t expect simple linearity. This is hyper-linearity — a trend vector that, even when it pulls back, provides additional opportunities.
However, we then look at the boundary, and it is assumed that it should exist — and it really always does, at least up to today
Reserve is the key aspect. Its size is the true quality of the prediction, and this applies to any trading strategy.
Simple guessing is the sure path to blowing your deposit )

It's strange — we seem to be the only ones in this section of the forum actually showing real live accounts )))
The FF administration moved us to the commercial section for some reason, even though we are not making any profit, not selling signals, and nowhere offering to buy our system.
At least, that is not our goal at the moment.
Today and in the near future, we are strictly following our roadmap — research only.
Live, honest trading with constant public demonstration of results, fully transparent.
No one knows what will happen tomorrow, but it is precisely the long and stable statistics that gives us reason to believe that our "tomorrow" will be similar to the previous months of the AI’s performance )))

#GBPUSD — just recently**
Beautiful lots.
Of course, in hindsight it doesn’t seem like anything fantastic.
Hindsight makes everyone a hero, as they say
But we honestly publish every single lot placement in our TG channel, second by second — when the future hasn’t happened yet

Anyone would say: “But this is impossible” (judging by their own experience and measuring others by it).
However, ORA AI has its own trading experience, which is honestly demonstrated on Forex Factory — every single lot is visible. Cheating is impossible

In short — personally, I’m a skeptic to the bone, and every single time, literally every time, I’m still amazed: does this really work?
And so it continues month after month — we keep testing and testing
And up to this moment — it works!
What do you think about our AI?
We’re very interested in your opinion. Any opinion, no matter how skeptical.

Hello

do you have a myfx account whith analises for this sistem?

They haven’t opened it yet, but does MyFXBook really offer more possibilities?
In our opinion, Forex Factory is more transparent and detailed when it comes to analyzing real trading performance (every single trade, equity curve, profit factor, win rate, pips gained/lost, trade duration, and so on).
Personally, I think that from the perspective of evaluating the quality of the system, there is more than enough information here. They take literally everything into account and even make the statistics look worse than it actually is — meaning it’s a stricter and more conservative evaluation
For example, they show the average trade duration here.
Strictly speaking, using a simple average is not entirely correct. It would be more accurate to use the median, standard deviation, and a histogram of trade durations. That would give a fuller picture. But never mind. Even if the stats are presented in a stricter way, in my opinion they still look quite impressive

ok

Thanks for your message.

So what is the idea with this sistem?

Will you offer signals, copy trading, ?

What is your idea?

Thank you for your interest and for asking the key question

Important clarification first:
The system is not tied to any specific pairs, timeframes, or direction.
In simple terms, right now we gave it 17 Forex pairs and it trades them. It could just as easily work on 30 pairs, stocks, or crypto. We’ve tested it and have strong reasons to believe it doesn’t really care. At its core is our special metric and a set of hypotheses that appear to work very well, as shown by our long-term statistics.

As for your question — what is the idea behind this system?
First of all, of course, the basic idea is to make money, just like everyone else.

But the bigger question is: how to actually use something that understands the Forex, stock, and crypto markets?
We see it as a kind of expert advisor.
It’s like a guy who “shouts” — it will go this way or that way.
You can connect trading bots to it automatically so they place lots based on its signals.
We also manually copy its trades by simply watching the TG channel — and that works too.
Formally, there are many ways to extract profit from something that statistically works.

Right now, we are very interested in people’s honest opinion.
We want to understand if such systems are interesting at all.

And yes, personally I have a “template broken” feeling ))
I’m a skeptic at heart — the kind who needs to be shown 100 times before believing.
On one hand, that’s exactly what we’re aiming for. On the other hand — it’s kind of crazy how it refuses to let us lose money
So, it would be really interesting to know — would something like this be useful or in demand for you?

Another important point.
Many different systems offer the same main goal — to successfully grow your deposit.
Many of them are very demanding when it comes to settings and understanding how they function.
And yes, many probably work quite well (although I don’t understand how one can talk about success without showing open and verifiable statistics — even here, there are dozens of threads, and almost no one posts real accounts).
Then we come along with our approach — no layers of charts on top of charts, no complicated indicators…
It probably looks like: “What? You can understand the market without dozens of charts and indicators stacked on top of each other? This must be nonsense…”

We completely understand this kind of reaction.
However, every thing is identical to itself.
Experience with complex systems full of settings and charts is experience with those kinds of systems.

We have a different system. That kind of experience (analyzing through multiple charts and settings) does not apply here.
It does everything by itself, and the output is simple: “THIS direction” or “THAT direction”, on this pair or another.
It may seem too little on the surface, but in reality there is a lot going on under the hood, while the person using the AI doesn’t need to dive deep into it.
Although some basic skills are still required.

And if this is interesting to you (even just a few likes on this post would be nice), we will be happy to explain in more detail.


P.S.:
We ourselves are engineers and mathematicians, and of course we love beautiful charts and complex calculations.
But sometimes with pain we watch how some people think that the importance lies in the calculation process itself — in some obscure empirical formula, often further divided into something else — and that this suddenly seems significant.
We sincerely feel sorry for many people, as well as for the time wasted on seemingly smart and complicated counting processes and drawing lines.
That’s why we are creating something that doesn’t do anything meaningless.
A lot of what is considered “serious and scientific” in trading is actually pointless and only creates an illusion of importance, complexity, and value — especially when there is no proven statistics behind it.

Hello once more.

Recently i have join a project that envolves copy trading. The trader has a prove nice sistem with good sucess.

I think it is a good idea offer copy trading as a service. Like the idea. I invest and no need to look at charts. Since i have a daily job cant be al day in front of charts waithing for signals.

This is my opinion.

I belive with a toll like that you can do alot.

Whyen are you ready to go alive with it? from what i understand you are still testing it.

Thanks

Hello,

Thank you for your message and for the great question — it really hits the core of the matter
Let me answer in more detail, as others might be interested too.
Formally, we could launch tomorrow.
Considering that 99% of what’s offered on the market doesn’t even provide any verifiable proof of effectiveness (not even basic metrics), our system is more than ready by those standards ))

However, we probably respect both our own money and the potential money of investors and subscribers much more than that.
We want to create a tool that will genuinely bring them profit and peace of mind — without endless “ifs” and vague disclaimers.


Here’s what’s happening right now, for those who are interested.

We reached a win rate above 92% more than a year ago.
That is — on more than 16,000 lots and tests, it has been holding steadily.
So it’s already clear: the system is not guessing — it knows.
The probability of guessing at this level is 10 to the power of 30 — meaning it’s impossible to guess like this even throughout the entire lifetime of our universe. It may sound overly dramatic, but it’s important to understand: we actually calculate and verify even this, including through probability theory.
In other words, you always have to believe in something. Believing that it’s simply guessing with such accuracy is already more absurd than believing that it actually knows.

Why am I saying all this?

Because simply “knowing” — which so many people dream of — is actually not enough!
Here I put ten exclamation marks.
Many people think: “If I know, then that’s it, the job is done.”
Perhaps our main discovery is that this is not the case — and that it’s simply not enough!

It’s impossible to know anything with 100% certainty.
That means there will always be some “conditionally bad” lot (some outlier).

Why “conditionally”?
Because that’s exactly what we are testing right now.
It happens literally yesterday and before.
Dozens of lots are excellent, but one stays open longer.
Is it a mistake?
Time after time we become convinced that it is not. It’s more like a temporary gust in the opposite direction (pullbacks, stopped-out positions, other unexpected events that temporarily shift the trend). But in the end? We still come out in profit, and that “conditionally bad” lot either closes in profit too, or we simply don’t wait for it — because we’re already sitting on double profit from the current series of lots (yesterday was exactly like that — wait another 5 minutes and the “bad” lot would have been in profit too. Everything is visible on the account). But why wait when the Profit Factor on the series is already >2.0?
In short, we are testing and refining the strategy, and right now we are mainly focused on situations with those “conditionally bad” lots.
Important: we have never seen a truly bad lot — one that was completely “off” and from which we couldn’t recover.
In fact, that’s exactly what we are verifying.
The system has an understanding of the trend. Okay. But the trend can rarely and temporarily change (stops, other aspects). We are checking whether we always have a way to survive? Not to lose the daily profit, and God forbid more? So far — it has always been the case!

How much longer do we need to test?
We think another month, maybe two.

I’ve said it before — I’m a skeptic to the third power

But to finish, I’ll say this:
Our AI is not pre-trained. That is, it’s not a system that memorized a sextillion patterns from the past with the hope that the future will be the same. No. In our view, that’s nonsense and won’t work.
And that means — everything that works stably now will likely continue to work tomorrow. That’s exactly what we are verifying.

In short, I joke like this: if it continues like this — the million won’t run away, and the statistics will only strengthen our confidence

ORA constantly gives us reasons to post something — and these reasons are always pleasant
For example, yesterday we tweeted that #USD was clearly undervalued because the reserve on a certain trend was small.
We tweeted that the reserve would likely expand.
That’s how we read ORA.
And we have dozens of such “prophecies” on our X account

And voilà — by morning, everything turned out exactly as predicted )

We’re not even writing about the beautiful morning series on #CADJPY anymore. It just works.
At least up to the second this article was written

Not sceptical, just hard to follow along without knowing how you are doing the trades. Sure looks like you are on to something. All the best. Keep posting.

Thank you for your message — it’s really great!

At first, I also had a strange feeling.
We love control. Before making a decision, we form a foundation that then fundamentally determines our attitude toward the trade and its trend.
We wait because we have decided so and calculated it through experience.

At first glance, this seems to be missing here.

In general, interpreting AI is a big separate discussion in any field.
So far, many developers have come to the main conclusion when working with AI: it must be logical. That is, we set a task, and the result should be logical. Not just accurate — that’s not enough. It also has to be logical. It should somehow make sense.
What’s interesting is that any AI is essentially a matrix with millions of parameters and connections.
It’s hard to tell what kind of logic it might sometimes follow, and therefore it can see logic where we haven’t noticed it yet and haven’t even come up with a name for such a connection or correlation — which for some reason works.
However, with the market, everything turned out to be simpler (thank God).

We have a binary task — GROWTH or DECLINE

And if you look closely at where our AI places lots, to be honest, sometimes you feel both shock and delight
Almost ideal levels. Moreover, sometimes you can see on H4 and even on W1 that it found such places in some mystical way that many have already forgotten about
And the real mystery is that no one told it about them. We don’t teach the AI past years.

In short, by now (after probably thousands of checks of its work), I have developed a different experience and feeling.
Something like: it has worked hundreds of times before, so most likely it will work again now — and to hell with the context

Hello

Tnaks for the message

I would like to know more about this. I have join the Telegram group.

What more i can follow you?

Thanks

Thank you very much for your interest.

At the moment, we publish all the news on our X account.
You can also find more information on our website: OraSignal.com

We will continue to support this forum thread and keep sharing the results.
As I mentioned earlier, our current goal is to test stable and reliable growth - that it continues as it is now, regardless of the market conjuncture, including special situations such as #TrampStartEndWarNews. So far, even this does not break the AI

In other words, we look at situations where there are many "+" and suddenly a "-", but overall there should always be a profit, and not less than 2:1, or as it is now — 3–4:1.
If this does not change over the next few months, until summer, then I think the beginning of selling the bot will become an obvious step. By then we will have the highest confidence that we will not let anyone down and will be able to provide a clear, understandable, and by the way quite simple logic of "how to play this piano"
We will definitely announce it here as well.

One thing I've learned after 5 years of backtesting forex strategies: most traders don't realize their indicators are repainting.
Here's the issue. When you use close[0] (the current bar's close) in your strategy logic, the value changes with every tick until the bar closes. So your backtest sees the final close, but in live trading you'd be entering on a value that hasn't settled yet.
The fix is simple: always reference close[1] (the previous confirmed bar). Same for any indicator that derives from price. RSI(close, 14) should be RSI(close[1], 14) if you want your backtest to match live execution.
I ran a comparison on a basic EMA crossover on EURUSD M15 over 2 years:
- Using close[0]: 67% win rate, 1.8 profit factor
- Using close[1]: 54% win rate, 1.1 profit factor
That 13% gap is entirely fake. The close[0] version was "seeing the future" on every single bar.
If your strategy only works with close[0], it doesn't work. Period.
Anyone else run into this? What's your approach to validating that your backtest results are actually reproducible in live