How far into the future can we really predict? - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hi there, this is a fascinating discussion. Certain patterns exist not only in intraday trading but also in positional trading. Here are a few of them:
- A recent high or low is highly likely to be breached.
- Before transitioning from an extended consolidation phase back into a trend, the price will pull back to establish a launchpad.
- Price action is inherently more range-bound than trend-driven.
- And there are others.
By leveraging these tendencies, one can manage long-term positions with a win rate exceeding 50%. These patterns also allow for the timely liquidation of poorly timed entries, which are bound to happen. After all, we are only human :)
Interestingly, there are actually quite a few traders who hold positions for months. That's exactly why I'm curious about the reasoning behind it.
From both a fundamental and a mathematical perspective, reality continuously generates new information. Of course, with very inert instruments like EURUSD, a long-term view can sometimes be justified. But even with such pairs, as we're witnessing right now, conditions can change significantly and surprisingly quickly.
And I completely agree that there will always be losing positions.
It's not just because we're human. Reality itself produces outliers. Even with a verified 95% win rate, there will still be that remaining 3–5% that requires careful management.
What I've personally found interesting is that those 3–5% are different when the edge is real and statistically verified.
They aren't completely irrational or chaotic.
It's a bit like a cool day in summer. Yes, it's cooler than expected—but it doesn't suddenly become winter. That remaining uncertainty still has structure, and that opens the door to additional ways of managing it.
To me, that's a fundamentally different situation from trading with no measurable edge at all.
Overall, I'm pleased that many people here seem to arrive at a similar conclusion.
The universe doesn't appear to be fully deterministic. And if any determinism exists, it seems to be limited to certain horizons—within which each of us tries, in our own way, to discover and exploit whatever structure is actually there.
That's why I keep coming back to the same conclusion: from an initial point in time, genuinely long-term forecasts—days, weeks, and certainly months—ahead are fundamentally limited.
If you want to hold a position for an extended period of time, consider using large Renko bricks in a custom chart:
Forum on trading, automated trading systems and testing trading strategies
Renko bricks ─ your opinion?
Ryan L Johnson, 2026.03.26 20:21
For those of us interested in positional LFT, here's a Heiken Ashi indicator applied to a wick'ed 200 pip GBPJPY Renko chart:
I think I see the question you're asking: what is the maximum trade duration that requires zero attention or active management? The kind where you enter a position, set your take-profit and stop-loss orders, and just walk away. It seems that this approach no longer works on any timeframe. You are absolutely right - the longer a trade remains open, the more market variables shift.
How can we isolate a quantum of time in which no market changes occur? Perhaps we could try defining it through market cycles. For instance, notice how volatility bumps slightly at the open of a hourly candle, or how distinct the intraday cycle is. However, this remains a probabilistic assessment.
I think I see your point: what is the ideal trade duration that requires zero attention or active management? The kind where you enter a position, set your take-profit and stop-loss orders, and just walk away. It seems that this set-and-forget approach no longer works on any timeframe.
Oh wow. Ok, I was clearly missing the OP's main idea─the invalidity of fortunetelling.🔮
Again, why would anyone attempt to engage in fortunetelling instead of statistical analysis/data forecasting?
Even if I were to create a Nearest Neighbor Weighted Coefficient strategy, I would reduce it to an EA... and run it through the Tester. At that point, I would have come full circle─right back to data forecasting.Agreed, analysis is the name of the game.
Let me add a point regarding the trading horizon: on the lower end, we are squeezed by spreads, and on the higher end, eaten up by swaps. The perfect balance lies somewhere within swing and positional trading.
Agreed, analysis is the name of the game.
Let me add a point regarding the trading horizon: on the lower end, we are squeezed by spreads, and on the higher end, eaten up by swaps. The perfect balance lies somewhere within swing and positional trading.
Let me clarify one thing first.
In my question about short-term versus long-term forecasting, I deliberately avoid defining strict boundaries.
As we know, time itself is a rather problematic concept. In fact, it doesn't even appear in several fundamental equations of physics. We need time as observers to describe and analyze change—to "slice" continuous processes into intervals—but whether time itself is fundamental remains an open question.
That's why my question is intentionally qualitative rather than quantitative. I'm not looking for an exact threshold or some mathematically perfect cutoff.
It's similar to the relationship between quantum mechanics and classical physics. On microscopic scales, one set of laws dominates; on macroscopic scales, deterministic behavior emerges. But where exactly is the boundary? Around the Planck scale? Not really. There are much larger systems that can still require quantum descriptions. The transition isn't a sharp line.
So, for me, many of the answers here are already valuable. I'm interested in people's experience—at what point does "long enough" simply stop working because too much new information has entered the system and the original assumptions gradually dissolve?
As for Renko charts...
With all due respect, I see them primarily as a convenience for us, constrained by our own perception.
Personally, I don't believe there is such a thing as "unnecessary" information. Every piece of data potentially carries information.
Renko is essentially a filter. It reduces noise, but it also removes information.
A quiet market is still information. A market that barely moves tells us something important as well.
That's one of the reasons I'm so interested in AI. It can process large amounts of information simultaneously, whereas my eyes and intuition inevitably simplify reality. What is comfortable for a human observer is not necessarily the richest representation of the market.
As for Renko charts...
With all due respect, I see them primarily as a convenience for us, constrained by our own perception.
Personally, I don't believe there is such a thing as "unnecessary" information. Every piece of data potentially carries information.
Renko is essentially a filter. It reduces noise, but it also removes information.
If a trader has experience with Renko charts, that trader understands that no information is removed. In contrast thereto, some information is merely reorganized:
Forum on trading, automated trading systems and testing trading strategies
How far into the future can we really predict?
Ryan L Johnson, 2026.07.17 17:40
My preferred chart structure is a custom chart and specifically, a Renko chart. As you probably already know, Renko charts are "timeless." I've enclosed that word in quotes because Renko bricks still have timestamps, even though the incremental scale of the time (x) axis of the chart is heavily distorted based on the speed in which price moves. If the bricks form quickly, the timestamp increments of the bricks are condensed. If the bricks form slowly, the timestamp increments of the bricks are expanded.
In other words, Renko time becomes a numerical inter-bar variable unlike time in traditional timeframes. Nothing is lost unless the trader fails to observe the transformation.
Furthermore, a quality Renko chart generator also includes volume/tick volume.With all due respect, I see them primarily as a convenience for us, constrained by our own perception.
Excellent take on things . This is what I am talking about humans and the need for narrative feeling safety in Renko Charts because they block out noise , Eh you need to hear a commotion to know there is a problem . It blocks out noise and you are last to know the World Trade Centres fell , how would that event look on a 4hr / Daily/ Week Renko chart .
Back Test lie
Any pattern you find by staring at historical charts is vulnerable to overfitting , with enough data and enough tries, you will find a rule that "worked" in the past purely by chance, the same way you can find shapes in clouds. That pattern has no predictive power going forward, but it feels real because you saw it "prove out" on the data you tested it on.
Forecasting
If a pattern in price data were real and exploitable stock X always dips on Tuesdays, professional quant funds with faster data, more capital, and better models would find it first and trade on it until the pattern disappeared. By the time it's visible in a retail charting app, it's either noise or already dead. This is the efficient market idea in practice: prices already reflect available information, so what's left over is mostly randomness, not a hidden signal you found.This is what I am talking about humans and the need for narrative feeling safety in Renko Charts because they block out noise , Eh you need to hear a commotion to know there is a problem . It blocks out noise and you are last to know the World Trade Centres fell , how would that event look on an 4hr / Daily/ Week Renko chart .
I respectfully disagree. I would feel sympathy for a trader waiting for a H4, D1, or even worse W1 bar to form/close on 9/11. Both Twin Towers collapsed in well less than day. In contrast thereto, imagine another trader who used a 10 pip Renko chart that dissects H4, D1, or W1 bars into smaller bricks. Any large H4, D1, or W1 bar would be transformed into a consecutive run of uniform bricks. If both traders had been using a stupid-simple moving average crossover, the Renko trader would have caught the swings earlier than the H4, D1, or W1 trader.
Any pattern you find by staring at historical charts is vulnerable to overfitting , with enough data and enough tries, you will find a rule that "worked" in the past purely by chance, the same way you can find shapes in clouds. That pattern has no predictive power going forward, but it feels real because you saw it "prove out" on the data you tested it on.
Following your "logic," the MT5 Tester would be rendered useless bloatware for everyone which is frankly, absurd. If you don't know how to properly and accurately backtest and/or code for Tester compliance (this is more common than it should be), that is on you─not on the Tester. Let's take a scalping strategy that profitably tests on 5 pip Renko bricks over the course of 5+ years, for example, wherein thousands of round trades are simulated. Assuming that no holy grail, grid, nor Martingarbage is implemented, the strategy merely has a slight edge that it exploits. If every month and year inside of that 5+ year period returned a simulated profit, you would be hard pressed to find a paradigm shifting event that invalidates those long-term statistics─not even 9/11.
IMHO, a large part of the problem (at least, more recently) is the viral status of adaptive/machine learning─especially among those who don't really know how to use it. If you're blindly updating your map as you go down the road then of course, there are going to be many wrong turns ahead.
A trader riding on luck has nothing to hang her/his hat on.
If a pattern in price data were real and exploitable stock X always dips on Tuesdays, professional quant funds with faster data, more capital, and better models would find it first and trade on it until the pattern disappeared. By the time it's visible in a retail charting app, it's either noise or already dead. This is the efficient market idea in practice: prices already reflect available information, so what's left over is mostly randomness, not a hidden signal you found.
That is a gross oversimplification of a real trading strategy. Having said that, there is no reason that it can't be backtested over time if so desired (just for fun, I guess). The average retail trader is not competing with professional quants, or even trying to do so. If they were, there would be no point in retail trading whatsoever. The Efficient Market Theory is just another theory as is the Random Walk Theory, etc. It is not a law of physics. Again, retail traders will never enjoy the same timely access to complete market data as that of professional traders. This is a markets-wide theory that has little to nothing to do with statistical testing/data forecasting. Liars can make statistics, but real statistics don't lie.
Unlike the trader who hangs her/his hat on collected and analyzed statistics, a trader who relies on luck has absolutely nothing to hang her/his hat on. If you're unable to devise and empirically prove a profitable strategy, that is quite common. That does not, however, mean that everyone else in the Forum cannot make it happen.I respectfully disagree. I would feel sympathy for a trader waiting for a H4, D1, or even worse W1 bar to form/close on 9/11. Both Twin Towers collapsed in well less than day. In contrast thereto, imagine another trader who used a 10 pip Renko chart that dissects H4, D1, or W1 bars into smaller bricks. Any large H4, D1, or W1 bar would be transformed into a consecutive run of uniform bricks. If both traders had been using a stupid-simple moving average crossover, the Renko trader would have caught the swings earlier than the H4, D1, or W1 trader.
Following your "logic," the MT5 Tester would be rendered useless bloatware for everyone which is frankly, absurd. If you don't know how to properly and accurately backtest and/or code for Tester compliance (this is more common than it should be), that is on you─not on the Tester. Let's take a scalping strategy that profitably tests on 5 pip Renko bricks over the course of 5+ years, for example, wherein thousands of round trades are simulated. Assuming that no holy grail, grid, nor Martingarbage is implemented, the strategy merely has a slight edge that it exploits. If every month and year inside of that 5+ year period returned a simulated profit, you would be hard pressed to find a paradigm shifting event that invalidates those long-term statistics─not even 9/11.
IMHO, a large part of the problem (at least, more recently) is the viral status of adaptive/machine learning─especially among those who don't really know how to use it. If you're blindly updating your map as you go down the road then of course, there are going to be many wrong turns ahead.
A trader riding on luck has nothing to hang her/his hat on.
That is a gross oversimplification of a real trading strategy. Having said that, there is no reason that it can't be backtested over time if so desired (just for fun, I guess). The average retail trader is not competing with professional quants, or even trying to do so. If they were, there would be no point in retail trading whatsoever. The Efficient Market Theory is just another theory as is the Random Walk Theory, etc. It is not a law of physics. Again, retail traders will never enjoy the same timely access to complete market data as that of professional traders. This is a markets-wide theory that has little to nothing to do with statistical testing/data forecasting. Liars can make statistics, but real statistics don't lie.
Unlike the trader who hangs her/his hat on collected and analyzed statistics, a trader who relies on luck has absolutely nothing to hang her/his hat on