None of the five below are bad indicators. Four of them are on my own screen. They get people hurt because each one is a description of something that already happened, and it keeps getting used as a prediction of what happens next. Here is what each one is actually saying, and the specific way the popular reading goes wrong.
RSI above 70 is a description, not a sell signal.
Wilder published RSI in 1978 for markets that mean-revert on a timescale of days. Bitcoin trends do not respect that assumption. In a real trend the daily RSI parks above 70 and stays there, and the people who sold the first 70 print spend the next month watching.
The useful version is the range shift. In an uptrend, RSI tends to oscillate roughly between 40 and 90 and bottoms out near 40 rather than 30. In a downtrend the whole band slides down and rallies die around 60. So the information is not "70 was hit" β it is "the floor moved". When a market that has been holding 40 on pullbacks starts failing at 40 and rolling over, that is a change worth acting on, and it usually happens well before the oscillator looks extreme.
The other thing that gets skipped: timeframe. A 4h RSI at 78 and a daily RSI at 78 are different statements about position size, and they routinely disagree. We took one specific case apart in RSI 4h versus 1d around the March 2024 high; the mechanics are in the RSI guide, and you can run your own series through the RSI calculator.
Negative funding tells you who is paying, not who is right.
Funding going negative means shorts are paying longs to hold the perp near spot. That is a payment, not a forecast, and it gets misread in both directions at once. One camp sees it and concludes shorts are trapped, so buy. The other camp sees it and concludes the market has turned bearish, so sell. Both skipped the same question: who is short, and why.
Because a large share of perp shorts are not bearish at all. Miners hedging production, desks hedging inventory, and funds who are buying spot and want the price risk off while they accumulate all sell perpetuals against holdings they intend to keep. That selling pushes the perp below spot, which is what makes funding negative, and those sellers are paying the funding as a cost of hedging rather than as a bet on direction. Negative funding on top of visible spot accumulation is a completely different situation from negative funding while spot is being dumped β and the number on the dashboard looks identical in both cases.
May 2022 is the case worth remembering, because it is the one that pays for the lesson. UST lost its peg and LUNA went to effectively nothing inside a few days. Funding on that pair had been negative going into it. Every person buying the negative-funding dip was buying an asset whose collateral model was in the process of failing. The indicator was not lying β shorts really were paying β it just had nothing to say about solvency.
What makes negative funding informative is what open interest is doing alongside it. Negative funding with open interest still climbing means new shorts are being added, and that is fuel. Negative funding while open interest falls means shorts are covering, and the squeeze you were waiting for is already being spent. Same funding number, opposite situation. That split is the whole subject of why funding turns negative, and the cost side is in the funding rate guide.
A falling dominance chart is not the same as altseason.
BTC dominance is a ratio. Ratios fall for two completely different reasons, and the chart looks identical either way: alts can rise faster than Bitcoin, or Bitcoin can fall faster than alts. The second one is not a rotation into risk, it is everything going down with Bitcoin leading, and it is a terrible moment to buy the smaller end of the book.
Two mechanical traps sit underneath the chart as well. First, check what your provider puts in the denominator β some total-market-cap series include stablecoins, some exclude them, and the same day can produce visibly different dominance readings depending on which one you loaded. Second, dominance moves when large new tokens are listed and counted, with no capital rotating anywhere.
The screen I trust more: alt pairs quoted in BTC, not in dollars. If the alts you follow are making higher lows against BTC while BTC itself is flat to up, something is genuinely rotating. If they only look strong in USD terms, you are just looking at Bitcoin. The dashboard is at BTC dominance and the longer argument is in altseason signals.
MVRV thresholds are a moving target.
The numbers people quote for MVRV and MVRV Z-score β sell above this, buy below that β were fitted to the 2013, 2017 and 2021 tops. Pull up any long MVRV Z-score chart and the shape of the problem is obvious: each cycle's peak has come in lower than the one before it. A threshold calibrated on the tallest peak in the series will keep you long through every subsequent top; a threshold calibrated on the most recent one will get you out early if the pattern breaks.
There is a second drift underneath. MVRV compares market cap to realized cap, and realized cap depends on when coins last moved. Coins that are provably lost, plus supply that has sat still for a decade, are still in the denominator at prices nobody will ever transact at again. That does not invalidate the metric, but it does mean the absolute level is not a constant across a decade.
Use it as a temperature reading relative to the last two cycles rather than against a fixed line, and check it monthly rather than daily β nothing about it moves fast enough to justify more. Inputs and arithmetic are in the MVRV Z-score calculator, with the rest of the cohort metrics in six on-chain metrics.
Coinbase Premium now has an ETF inside it.
Coinbase Premium was a decent proxy for US retail bid when Coinbase was mostly US retail. That is no longer what the venue is. Since the US spot Bitcoin ETFs launched in January 2024, Coinbase has also been the custodian and an execution venue for a large share of that complex, which means creation and redemption activity can show up in the same price gap that people still read as sentiment.
So a positive premium can mean US buyers are paying up, or it can mean an authorised participant is filling a creation basket on a day when nobody retail did anything. Those two require opposite reactions. There is also a plumbing component: the premium is a USD-settled price against USDT-settled prices elsewhere, so when USDT itself trades slightly off peg, part of the gap is the stablecoin, not demand.
The check that separates them is same-day ETF flow. Premium up on a heavy net-creation day is mostly mechanical. Premium up on a flat flow day, holding for several sessions, is the version that carries information. Data is on the Coinbase premium dashboard, the interpretation problem is in Coinbase premium: real demand or false comfort, and the flow layer is in the ETF flow note.
Same mistake five times.
Strip the specifics and two errors are doing all the damage.
The first is treating a lagging measurement as a contrarian trigger. RSI, funding, dominance, MVRV and the premium all describe conditions that already exist. Extreme conditions persist β that is what makes them extreme. "This is stretched" and "this reverses now" are not the same sentence, and the gap between them is where accounts die.
The second is running one indicator alone. I am not going to hand you a win rate for that, because I do not have one and neither does anyone quoting you a number for a single indicator on a single asset. What I can say is structural: every one of these five has a documented failure mode that a second, mechanically unrelated input would have caught. Funding is a derivatives measurement; check it against a spot or on-chain measurement, not against another derivatives measurement.
The rule I use instead of a threshold.
An extreme reading changes my size. It never changes my direction. If funding is at the top of its recent range and RSI is pinned, I take risk off β I do not flip short. Reversing on an extreme means betting against a trend on the evidence that the trend is strong, which is a sentence that should not survive being said out loud.
The second habit is cheaper than it sounds: before opening anything on an indicator, write the reading that would prove you wrong. "Long here because funding is negative" becomes falsifiable the moment you add "and I am wrong if open interest keeps climbing while price does not bounce." Most of the five misreads above cannot survive being written that way, which is exactly why the popular version of each never is.
Keep a short failure list per indicator as you go. Mine has entries like "dominance chart, denominator includes stablecoins" and "premium spike, check ETF flow first." It is four lines long and it has saved me more money than any of the indicators on it. If you want terminology while you build yours, the glossary defines the pieces.
Context and references.
Crypto indicators are routinely misread because their behavior differs from traditional finance counterparts. RSI overbought thresholds, funding rate signs, BTC dominance moves, MVRV ratios and Coinbase Premium all have crypto-specific interpretation rules. CoinDesk and The Block have both covered the most common analytical errors.
The five most damaging misreads: (1) treating RSI 70+ as automatic sell in trending markets; (2) reading negative funding as automatic bearish without checking spot accumulation; (3) calling altseason from any BTC.D drop without confirming BTC absolute direction; (4) using MVRV thresholds from 2017-21 in the ETF-era market; (5) interpreting Coinbase Premium swings without ETF flow context.
For each misread, the correction relies on multi-indicator cross-checking. Glassnode's composite cycle dashboard pairs MVRV with NUPL and RHODL for cross-validation. Coinglass pairs funding with open interest and long/short ratio. Farside Investors and SoSoValue provide the ETF flow layer that updates legacy on-chain interpretations.
The structural lesson: crypto indicators are useful but require regular recalibration. The ETF channel changed cycle dynamics meaningfully in 2024-2025. The MiCA regulatory framework changing in 2025-2026 may shift them again. Static threshold rules become obsolete; framework-based interpretation does not.
Crypto assets are volatile and not suitable for every investor. This page is editorial analysis, not financial advice.