Not being a trader, when developing my BTC trading strategy, I had to be clear about one thing: Which data can actually predict Bitcoin price movements, and which data will only add noise to the prediction.

Let's start with the conclusion: After completing my work, I tested this system for a week. At every key signal point, it preemptively gave me the direction.

Here is the full logic.

1. Research Background: I Reviewed All Methods of "Predicting BTC"

I am not a professional trader. So, instead of jumping straight to indicators, I did something dumb first—

I reviewed all Bitcoin prediction methods from 2017 to 2025 available in the market.

They fall into three categories:

Category 1: Celebrity Opinions. VanEck predicted $180K by 2025. Missed. Bitwise predicted $200K. Missed. Tom Lee, Arthur Hayes, Novogratz, Cathie Wood—almost all major price predictions over the past 8 years had a systematic bullish bias, with an average deviation of over 50%.

Category 2: Analytical Methods. Stock-to-Flow model (PlanB's version), logarithmic growth curves, cycle theories, Wyckoff method, Elliott Waves... each had its own "historical accuracy," but when tested beyond 2024, almost all failed.

Category 3: On-Chain Signals. MVRV Z-Score, SOPR, NUPL, Puell Multiple, Hash Ribbon, Reserve Risk... This category is what I researched the most because it's not about "prediction" but rather about "state description."

After reviewing all three categories, I began to filter.

2. Filtering and Analysis: More Data Doesn't Mean More Accuracy, It Often Means More Noise

After screening, I discovered something counterintuitive:

When a massive amount of data points in different directions, your judgment actually gets worse.

After analysis, I divided them into two categories—

Unreliable Category (Discard)

Celebrity Predictions. The incentive structure dictates that they must make bold statements. Saying "$500K" gets them headlines, followers, and repeated mentions. Saying "$80K sideways" gets zero attention. No one questions them when they are wrong, but if they are right, they are forever hailed as "gurus." This structure doesn't change, so the predictions are never accurate.

Stock-to-Flow and similar pure models. Very accurate before 2021, completely off track after 2022. Why? Because the model's assumption was "price is determined by the supply curve," but post-ETF entry, the price is determined by capital flow, not supply. The model itself is not wrong; the world it describes has changed.

Single sentiment indicators (pure Fear & Greed). Historically, when Fear & Greed stayed below 20 for an extended period, sometimes it marked a bottom, and other times it was a precursor to "plummeting to -30." When used alone, there are too many false signals.

Reliable Category (Retain)

MVRV Z-Score. Measures the deviation of the current market value from the average cost of all holders. Every time it entered the green zone in history, it precisely marked a bottom within ±2 weeks—a triple success in 2018, March 2020, and 2022. But be aware: after 2024, its ability to predict tops failed ($73K signaled overheating, BTC rose to $126K) because ETF trading is off-chain, and it cannot see the institutional share of the pie. Therefore, only retain the ability to predict bottoms.

SOPR 28-day moving average. Measures how much BTC on the move was sold at a loss. Persistently below 1.0 = holders are capitulating = nearing the bottom. This indicator has been very consistent in predicting bottoms throughout history.

ETF Net Fund Flow. A core indicator added post-2024. Institutional marginal behavior must be observed from here because on-chain data cannot. Net inflow accumulating for more than 5 days consecutively >$1 billion = institutions are accumulating; net outflow for more than 5 days consecutively = institutions are withdrawing.

Macro Liquidity. Fed Direction + M2 Growth Rate. Long during easing cycles, reduce exposure during tightening cycles. No short-term timing, only set the overall direction.

Fear & Greed as a Supplement. Not used alone, only weighted when resonating with other signals.

After filtering, we are left with four dimensions. Adding one more would be redundant.

3. Strategy Formation: Four-Dimensional Resonance, Take Action Only When Three or More Point in the Same Direction

After clarifying "which criteria and why," I turned it into a trading strategy.

Core Logic: Do not chase price targets, only assess direction and positioning.

Bottom Assessment: MVRV enters the green zone + SOPR falls below 1.0 → On-chain hodlers capitulate, a historically high probability buy zone.

Top Assessment: On-chain indicators overheated + ETF consecutive outflows → Institutions retreat, reducing positions.

Macro Background: Fed Direction → Long during easing, reduce exposure during tightening.

Sentiment Supplement: Fear & Greed < 20 → Extreme fear, supplementary weighting.

No single signal is sufficient to act. Only when three or more point in the same direction is it a true entry basis.

Then I turned it into an automated monitoring system:

· Automatically fetch BTC price, Fear & Greed, on-chain data, ETF flows daily

· No push notifications if the signal is not triggered

· When triggered, I receive direct Telegram notifications

· Not a daily report, not noise. Only alerts when something genuinely worth attention happens

Current Signal (April 15, 2026)

This system's current readings for me:

BTC $71,631. Fear & Greed = 12, at a historically extreme level of fear. MVRV Z-Score in the green buy zone. SOPR below 1.0, indicating holders are selling at a loss.

On-chain Triple Confluence all confirmed.

The only counter signal: ETF flows have been relatively weak recently, with institutions not definitively starting to accumulate.

Historically, the on-chain Triple Confluence (extreme fear + MVRV in the green zone + SOPR < 1) has only occurred three times: at the bottom in late 2018, in March 2020, and at the bottom in 2022. Subsequently, there was a 100%+ return over the following 12 months.

This is not about predicting how high BTC will rise. This is an objective description of the current market state.

From my research, my biggest takeaway is:

Predictions are someone else's views; frameworks are your own tools of judgment.

If your prediction is wrong, you have nothing. If your framework is wrong, at least you know where the issue lies and can iterate.

You can incorporate your own preferences, such as leverage and cycle timing preferences. This way, the signals AI provides you are tailored to your operational characteristics.

Disclaimer: The above is based on historical patterns and is not financial advice.