Transcription
The RSI is one of the most loved indicators in trading, and it's loved for a good reason. The concept behind it is simple, easy to understand, and makes intuitively a lot of sense to our pattern searching human brains. When it works, it works beautifully.
However, in my opinion, it is also an extremely outdated indicator that has not aged well when it comes to dealing with modern markets. The RSI was published by J. Wells Wilder in 1978. That's nearly 50 years ago. The markets Wilder designed it for were stock markets that traded a few hours a day with mostly institutional participants in price action that moved at a fraction of today's speed. Crypto didn't exist. Retail trading almost did not exist. High frequency trading didn't exist. 24-hour markets didn't exist. Flesh crashes didn't exist.
Today's markets are faster, more extreme, and change direction harder than anything Wilder could have imagined when he picked the number 14 for the default RSI period. And yet, that same number 14 is still what almost every charting platform ships as default. The same fixed overbought and oversold levels of 70 and 30. The same calculation unchained apply to markets it was never designed for. So, it was about time the RSI got an update. And the good news for you is you don't have to do that work yourself. I did it. I spent the time researching, prototyping, validating, debugging, and stress testing. And I'm sharing the result with you completely for free on Trading View. So, you do not have to pay for all this garbage out there that others try to sell you as premium indicators.
In this video, I will not only explain the indicator itself, when to use it and when not, but I will also give you some settings for different asset classes such as crypto, forex, indices, and gold. As always, all I ask in return is this. Hit the like button, subscribe to the channel, and leave a comment underneath this video. That's it. And that's how this channel keeps existing and gaining more and more traction of becoming a true community for us retail traders.
Now, before we get into the details of my version of a modern RSI indicator, two things you need to know. First, this video is a complete technical walkthrough. I'm going to show you exactly how this indicator works, what it does well, what it does poorly, and where the dangers are if you rely on it blindly. This is not a hype video. I do not hype on this channel. I will not tell you this is going to make you a millionaire in two weeks. I bit tools such as indicators based on data and share with you what works for me. If you find my work useful to incorporate into your own trading style, mission accomplished.
Second, and this is important, you will probably want to watch this video more than once, maybe twice, maybe three times. The indicator is genuinely simple to use, but understanding why it works the way it works takes a bit of effort. And if you turn off this video after 3 minutes, slap the indicator on your chart, and expect to make money, you'll be disappointed. I can make your tools better, and I can save you the work of building them. What I cannot do is take the thinking and learning part off your shoulders. That part is non-negotiable if you want to become a successful trader. Anyone who tells you otherwise is selling you something. By the way, you'll, as always, find a link to the indicator in the pin comment and description of this video. So, with that out of the way, let's go into what this indicator actually does.
The classic RSI has three problems when applied to modern markets. Problem one, fix parameters. RSI 14 means looking at the last 14 bars. But what's a good look back depends entirely on the market regime. In a fast, choppy crypto market, 14 bars might be way too slow. In a quiet, drifting markets, 14 bars might be way too fast and noisy. There's no universally correct number, but the indicator pretends there is. Problem two, fixed thresholds. The traditional 70 and 30 levels assume the RSI distribution is roughly symmetric and that price reaches those levels at meaningful turning points. In a strong trend, RSI can sit above 70 for weeks without reversing. In a tight range, it might never reach 70 at all at obvious local highs. The thresholds don't adapt. Problem three, Wilder smoother is dated. RSI internally uses Wilder smoothing function, a onesided exponential moving average from the 1970s. It's slow, it's noisy, and there are much better digital signal processing techniques available today. John Alas, for example, has spent decades developing superior movers specifically for financial time series. So, we have three weak points. fixed period, fixed threshold, dated smoother. I think my adaptive modern RSI addresses all three.
However, my idea was simple, but the implementation actually took weeks to nail down. On every single bar, the indicator looks back over the last 300 bars, which is the optimization look back that you can adjust, by the way, and asks itself a question, out of all possible parameter combinations I could be using right now, which one would have called the pivots and reversals best over this recent history. It does so by testing 60 different combination, five different RSI periods, seven, 10, 14, 21 and 28, three different smoover strength, four different overbought oversold thresholds pairs. So 60 combinations in total. For each combination, it walks through the look back window and finds every bar where that version of the RSI would have triggered. Meaning every time the RSI crossed into oversold or crossed back out of overbought, then it measures what happened after each of those triggers. Did price actually mean revert and did the trigger pay off on average? The combination with a best forward return record gets selected as the active configuration. That's [clears throat] the version you see plotted on the chart. The aqua line, the red and green level lines, those reflect whatever parameter set is currently winning.
Now, you might be asking yourself, dear, if the indicator basically finds the optimum on every new bar, does it not repaint all the time and never give me real-time signals? The thing is, the indicator doesn't switch parameters every bar, even when the optimizer finds a new winner. It only switches when the new combination is meaningful better than the current one by at least 10%. This prevents the parameters from from flipping back and forth between near equivalent settings and making the chart unreadable. And because we are computing this incrementally using rolling sums instead of recomputing everything from scratch, the indicator runs fast even on long charts. That was a deliberate design choice because not everyone has a Trading View premium plan that allows higher compute time limits for indicators.
So at the top of the indicator pane, you'll see one of five regime labels at any time. Mean reverting shop. This is the RSI natural habitat. The market is doing exactly what RSI was designed to catch, pushing to extremes and reverting. Use the triggers actively here, but remember that even in shop, single bar reversals can fail. Wait for the trigger to print on a closed bar before acting and respect your stops. chop becomes trend without warning. Quiet range, mild swings and modest signal. The RSI works but the edges are smaller and the moves are shorter. Treat triggers as light confluence. Useful as one of many votes but don't size up aggressively on them alone. The adaptive thresholds will remain moderate probably around 25 to 75 range. Price reaching them is meaningful but not extreme. A trigger that aligns with key levels, divergence on a higher time frame or volume confirmation is worth taking. A trigger alone isn't trending. This is where most traders blow up using RSI. The instinct is to short overbought and to buy oversold. And in a trend, those are the worst possible entries. The indicator is telling you the regime has changed and you should respect that. In strong trends, even pullback entries can fail spectacularly. RSI is a secondary tool here. The primary tool should be trend identification and structure. Noisy transition. The market is between regimes and the indicator is telling you it cannot find a clean read. The optimizer selected heavy smoothing because the underlying signal is choppy and even with optimization the score is near zero. Remember, no strategy will be successful in every market regime. You have to adapt as a trader or face the consequences.
Next to the label, you see held n bars. That's how many candles the current regime classification has been stable. If it says held four bars, the regime just changed and you should weigh and read lightly. If it says held 200 bars, the indicator has been confidently calling this regime for a long time and you can trust it more. And the regime label only updates after the new classification has been held for at least 10 candles. I implemented this to prevent jitter. Small temporary parameter switches don't change what the label shows. Only sustained regime shifts do.
In practice, when you put this on a chart, you'll get four things. First, an RSI line that adapts to current market conditions. When the market is fast, the RSI is fast. When that's slow, the RSI is slow. Second, overbought and oversold levels that adapt with it. You'll see red and green horizontal lines changing occasionally. That's the optimizer adjusting the thresholds. You're no longer stuck with 70 and 30 in markets where those numbers don't mean anything. Third, clean trigger markers. The red triangles fire when the RSI rolls over from overbought. The green triangles fire when it enters oversold. These never repaint. They fire on the close of a bar where the crossover happens and they stay where they fire. So you are always seeing exactly what happened in terms of real signals and then decide for yourself if this is currently a good time to trust the indicator or not. Fourth, the regime label, a plain language read on what kind of market the recent past has been and how stable that read is.
Now, let's talk about some kay feeds because like every indicator, this one also has shortcomings when used just as an individual tool and I think it's just fair to share them with you as well and not overhype things. This indicator is backward adaptive. It optimizes its parameters based on what worked over the for example last 300 bars. The implicit assumption is that the next bars will rhyme with the last bars. And that assumption is true most of the time. Well, except at the exact moments you care most about regime transitions. When the market shifts from chop to trend or from trend to crash, the parameters were optimal for the previous regime will not be optimal for the new regime. The indicators will lack the transition. The standard look back is 300 candles. That's roughly 12 days on a 1 hour chart and the indicator cannot react faster than its own look back turnover. So as a rule of thumb, when the regime label has been stable for a long time, you can trust the signals more. When the regime label has just changed or when the health counter is low, treat the indicator asformational only. Don't bet your account on a regime that just flipped.
Second danger, this is one layer of confluence. And I know I've been mentioning confluence a lot in my videos, but I really need to highlight this every time because so many viewers are always asking for optimal settings or trading system based on my indicators that guarantee them to earn XY Euro dollar rupia per month. Indicators are basically tools such as Excel. If you learn how to use them, they can be a powerful tool to make you more effective or profitable as a trader. But they cannot replace your brain and your common sense. If, for example, Donald Trump shares something on social media that is moving the markets like crazy. No indicator will protect you from this. So, as always, this is not a standalone trading system. The RSI tells you something about momentum and mean reversion potential. It does not tell you about trend, volume, market structure, key levels, fundamentals or risk management. If you trade off RSI alone, even this version of it, you will lose money. Use it as one piece of evidence amongst several. Trading is not about finding the perfect indicator. It's about combining decent indicators with good judgment, sound risk management, and the discipline to follow your own rules. I can give you better indicators. I cannot give you the rest. That part is on you.
The defaults work well for cryptocurrencies on the 1 hour time frame, especially Bitcoin. But let's talk about some settings for other asset classes. Forex. Forex behaves quite differently from crypto. Lower volatility, much stronger trends when they do form, for example, carry trade flows or central bank cycles, and far less mean reverting shop on intraday time frames than crypto. The major pairs trend more than they revert. What that means in practice, the look back of 300 for one hour is fine, but on lower time frames such as 5 minutes or 15 minutes, consider using 500 to 700 bars. Forex regimes are more persistent than crypto regimes. The evaluation bars forward horizon of five is reasonable, but you might extend to 8 to 10 on Forex. Mean reversion in forex tend to play out over longer windows than the violent snapbacks you see in crypto. Important forex has sessions. London New York overlap behaves completely different from Asian session. If you are scalping intraday the indicator can't see session structure. You might want to add a session filter externally or run separate charts per session.
Indices closest in character to crypto of all the asset classes but with two big differences trading hours and strong directional bias from monetary police and equity flows. What that means? Defaults work well on 1 hour and higher. On lower time frames you might need to bump look back to 400 to 500 since intraday equity action has session driven cycles that 300 bars might not fully capture. The gap problem indices a gap often at the open. The indicator treats the gap candle like any other bar, which means a large gap can artificially generate a trigger that wouldn't have fired in continuous trading. Either use continuous contract data, which trading view provides for some indices, or be aware that the first one to two bars after market open may produce false triggers. US indices have a strong buy the dip character during bull regimes. The regime label often reads meanreverting chop even in obvious uptrends because the dips actually do revert. This is honest. The indicator is correctly identifying meanreverting behavior. But you should still bias long only or trend aligned entries during bull regimes regardless of what the green or red triangle say.
Gold. Well, gold is its own beast. It behaves like a defensive asset during risk off, like a commodity during inflation regimes, and like a US dollar inverse most of the time. It has multiple personalities depending on the macro environment. What this means for gold, I suggest the most aggressive parameter tweaking because gold has a long cycle regimes that switch slowly. Try look back at 500 to 700 on one hour. The default 300 will turn over too quickly to catch gold's character. Regime high sterosis is worth bumping to 12 to 30 on gold. The asset has slow regime transitions and you want the labor to reflect that. Quick flips will be misleading. Gold mean reverts beautifully in chop regimes but trends violently when it goes. The trending label on gold is a strong signal to either write the trend or to stay out if you like more mean reversion strategies. Do not fade overbought gold during a trending regime. One specific note, gold has long duration consolidation multi-week sideways action that can produce the quiet range label for extended periods. During those RSI signals are the most reliable. Once the consolidation breaks, regime usually transitions to trending immediately.
If you found this video useful, hit the like button, subscribe to the channel if you haven't already, and leave a comment below telling me what asset and time frame you are testing this on. I read and try to reply to every comment. And always remember, we are doing here no hype, just state there. We want to verify instead of just visualizing things. See you again next week.