METHODOLOGY
Everything the Radar computes is described here. If any choice on this page looks debatable, it probably is — and that is why it is written down.
1.CORRELATION: ON RETURNS, NOT ON PRICES
By default the Radar computes correlation on returns. The reason is simple and frequently ignored: two series that merely rise over time produce a high correlation even with no relationship between them. This is called spurious correlation, and it is the most common error in charts of macro against crypto.
For price series we use log returns. For rate series — interest rates, spreads, zero-centred indices such as the NFCI — we use the first difference, in percentage points. Applying a percentage change to an interest rate produces a number with no economic meaning: a rate going from 0.10% to 0.20% would have “risen 100%”.
Correlation on levels remains available as an option, but always accompanied by an explicit warning in the panel.
2.DIFFERENT FREQUENCIES
Bitcoin trades every day of the year. The 10-year Treasury only on U.S. business days. M2 is monthly. The Fed balance sheet is weekly. Combining them requires a rule, and ours is:
- For statistics, only dates on which both series have a genuine observation are used, and the comparison is made at the lower of the two frequencies — the only one carrying independent information. A carried-forward value never enters a correlation computation.
- For visualization, we carry forward the last known value of the sparser series, mark those points as carried forward, and preserve the actual observation date.
- The correlation window (rolling window) is converted from days into a number of observations at the effective frequency. A 90-day window on a monthly series amounts to three observations, which is reported as an insufficient sample.
3.LAG (LEAD-LAG)
Shifting one series in time makes it possible to test whether it tends to move before the other. When a shift is applied, the chart displays a permanent notice — the adjustment is never hidden.
The panel shows the full correlation-by-lag curve, not only the best value. This is deliberate: testing many lags and reporting only the best one is a classic way of finding patterns that do not exist. An isolated peak surrounded by low values is almost always noise.
4.COMPUTED SERIES
The series below do not come ready-made from any provider: they are built here, with these formulas.
| Series | Formula |
|---|---|
| ETH/BTC | ETH/BTC = preço_ETH_USD ÷ preço_BTC_USD |
| Bitcoin dominance | BTC.D = (capitalização_BTC ÷ capitalização_total) × 100 |
| TOTAL2 — altcoin market capitalization | TOTAL2 = capitalização_total − capitalização_BTC |
| TOTAL2 measured in Bitcoin | TOTAL2/BTC = TOTAL2 ÷ preço_BTC_USD |
| Fed net liquidity (estimate) | Liquidez líquida ≈ Ativos do Fed − TGA − ON RRP |
| Global M2 (Trade With Renato index) | M2 global = M2_EUA + (M2_zona_euro × USD/EUR) + (M2_Japão ÷ JPY/USD) + (M4_RU × USD/GBP) |
| Fed rate expectations (2-year minus Fed Funds proxy) | Expectativa ≈ rendimento_2anos − Fed_Funds_efetiva |
| Bitcoin/gold ratio | BTC/Ouro = preço_BTC_USD ÷ preço_ouro_USD_por_onça |
5.GLOBAL M2 — WHAT IS INSIDE AND WHAT WAS LEFT OUT
There is no official “global M2” series. The Radar index sums the broad monetary aggregates of the United States, the euro area, Japan and the United Kingdom, converted into dollars at the Federal Reserve end-of-period exchange rates.
- United States: M2 (FRED, series M2SL), monthly, seasonally adjusted.
- Euro area: M2 from the ECB Data Portal, monthly, outstanding amount.
- Japan: M2 from the Bank of Japan API, monthly, average outstanding balance.
- United Kingdom: M4 from the Bank of England. The country does not publish M2 under the same definition as the other economies, so we use the equivalent broad aggregate and record the difference.
- China: outside the index. The PBoC offers no stable public endpoint and its robots.txt restricts automated collection. Since China accounts for a substantial share of the world money stock, this is the most serious limitation of the index, and for that reason it is declared on every page that uses it.
One detail that breaks many global M2 indices: the direction of the Federal Reserve exchange-rate quotations is not uniform. DEXUSEU and DEXUSUK carry the dollar in the numerator — multiply. DEXJPUS and DEXCHUS carry the foreign currency in the numerator — divide. Inverting one of them produces an index that looks plausible and is wrong.
The FRED foreign M2 series that many analysts use (MYAGM2EZM196N and similar) have been frozen since 2017-2019. They respond normally and announce a next release, which makes the failure invisible. That is why the Radar goes to the primary sources and keeps an alert that fires whenever any series falls more than two frequencies behind.
6.FED NET LIQUIDITY — AN ESTIMATE, NOT AN OFFICIAL DEFINITION
Net liquidity ≈ Fed assets (WALCL) − Treasury General Account − Overnight Reverse Repo
No central bank publishes this measure. It is a conventional and widely used approximation, with debatable choices about what to include — there are versions that add or subtract other accounts. The Radar always presents it as an estimate.
The TGA series required special handling: the Daily Treasury Statement changed structure twice. Until around 2021 the account was called “Federal Reserve Account”; then it became “Treasury General Account (TGA)”; and from 2 May 2022 it became “Treasury General Account (TGA) Closing Balance”, with the value moving from the close_today_bal field to open_today_bal — while the old field began returning the string "null". A naive query against the old field returns empty for everything after 2022 without raising any error at all.
7.WHY WE AVOID THE DOMINANCE-AGAINST-DOMINANCE CHART
A chart titled “Bitcoin dominance × altcoin dominance” looks informative and is not. If altcoin dominance is defined as 100% minus Bitcoin dominance, the two lines are mathematical mirrors and the correlation between them is exactly −1 by construction, over any period, under any circumstance. The chart looks good and carries no new information.
Comparisons that do carry real information:
- BTC dominance × ETH/BTC;
- BTC dominance × TOTAL2/BTC;
- BTC dominance × stablecoin dominance;
- BTC dominance × percentage of altcoins outperforming BTC.
8.DOLLAR INDEX: WHY IT IS NOT THE DXY
The DXY is a proprietary ICE index made up of six developed-market currencies, and displaying it publicly requires a licence. The Radar uses the Federal Reserve nominal broad dollar index (DTWEXBGS), which is in the public domain and uses a trade-weighted basket of about 26 currencies, including the yuan, the Mexican peso and the Brazilian real.
The two move in the same direction most of the time, but they are not the same thing: different bases (January 2006 = 100 against March 1973 = 100) and different baskets. Comparing a level of the broad index with a level of the DXY as if they were interchangeable is an error.
9.DELAYED DATA AND FAILURES
When an update fails, the behaviour is always the same:
- the last valid observation stays in place;
- the series is marked as delayed and that is shown in the interface;
- the time of the last successful update remains visible;
- the value is never replaced by zero;
- the failure is logged and visible in Source status.
Three consecutive failures change the state of the series to “failing”. An isolated failure changes nothing beyond the log — providers fluctuate, and raising an alarm at every fluctuation trains the operator to ignore alarms.
10.LICENCES AND WHAT THAT MEANS FOR THE RADAR
Not every available data point may be displayed publicly. The levels of the S&P 500 and Nasdaq-100 indices require prior written authorisation from their respective administrators; the ICE high yield spread prohibits reproduction without written permission; the DXY is licensed by ICE. The Radar code blocks ingestion of those series — it is a lock, not a warning.
Where a legitimate substitute exists, we use it and state that it is a substitute: an ETF in place of the index level, the Fed broad index in place of the DXY. Where none exists, the chart stays marked “source pending” until the licence is contracted. Under no circumstance do we fill the gap with an estimate or with simulated data.
11.GLOSSARY
Each entry answers what it is, why it matters, how to read it and which common mistake to avoid.
M2
A broad measure of the amount of money in an economy: currency, checking deposits, savings, and short-term funds.
Why it matters: It is the most direct way to track whether money is being created or destroyed in the system.
How to read it: Watch the growth rate, not the level. M2 growing 2% a year is a different world from M2 growing 25%.
Common mistake: Treating M2 as if it were liquidity available to invest. Much of it sits in household accounts and never reaches the markets.
Global liquidity
An umbrella term for the amount of money and credit available in the global financial system.
Why it matters: Risk assets tend to appreciate when liquidity expands and to suffer when it contracts.
How to read it: There is no single official measure. Each construction — aggregate M2, central bank balance sheets, net liquidity — captures a different piece.
Common mistake: Presenting a proprietary global liquidity index as if it were an official figure. It is not: it is always a construction, with debatable choices.
TGA — Treasury General Account
The U.S. Treasury's checking account at the Federal Reserve, where taxes and debt issuance come in and government spending goes out.
Why it matters: When the TGA rises, money leaves the financial system and moves into the government. When it falls, the money flows back into circulation.
How to read it: It is an account that drains or injects liquidity without the Fed changing anything. Large bond issuances tend to raise the balance quickly.
Common mistake: Ignoring the TGA when looking at the Fed's balance sheet. The balance sheet can hold steady while the TGA alone tightens financial conditions.
Reverse repo (ON RRP)
An operation in which funds and banks lend money to the Fed overnight in exchange for securities as collateral.
Why it matters: It is a parking spot for idle liquidity. Money sitting there is not buying assets.
How to read it: A high balance signals excess liquidity with nowhere to go. A falling balance can indicate that money is being reallocated.
Common mistake: Reading a decline in reverse repo automatically as money flowing into risk assets. It may simply be migrating into short-term Treasury bills.
Real yield
The interest rate net of expected inflation. In the U.S., it is observable directly in the yield on inflation-indexed securities (TIPS).
Why it matters: It sets the opportunity cost of holding assets that pay no income, such as gold and Bitcoin.
How to read it: A high real yield makes it expensive to hold non-yielding assets. A negative real yield does the opposite.
Common mistake: Calculating real yield by subtracting past inflation from the current rate. What matters is EXPECTED inflation, not inflation that has already occurred.
Inflation breakeven
The difference between the yield on a nominal bond and an inflation-indexed bond of the same maturity.
Why it matters: It is the average inflation rate the market is pricing in for that period.
How to read it: A 10-year breakeven of 2.4% means the market expects average inflation of 2.4% a year over the next decade.
Common mistake: Treating the breakeven as a pure inflation forecast. It also embeds a risk premium and a liquidity premium.
Yield curve
The relationship between a bond's maturity and its yield. Longer maturities normally pay more.
Why it matters: When the curve inverts — short-term paying more than long-term — it has historically preceded an economic slowdown.
How to read it: Watch the 10-year-minus-2-year spread. Negative means inverted.
Common mistake: Assuming the inversion marks the start of the recession. Historically, the stress shows up at DE-INVERSION, when the curve steepens again because the central bank has started cutting rates.
Credit spread
The extra yield required to lend to a company instead of to the government.
Why it matters: It is the price of default risk and tends to be the first market to show deterioration.
How to read it: Rapid widening signals risk aversion. The speed of the move matters more than the level.
Common mistake: Looking at the absolute level. A 400-basis-point spread in a high-rate regime means something different from the same spread in a low-rate regime.
VIX
An index measuring expected volatility for the S&P 500 over the next 30 days, derived from options prices.
Why it matters: It functions as a risk-aversion thermometer for the U.S. equity market.
How to read it: The signal is in the extremes. Spikes above 30 mark acute stress.
Common mistake: Calling the VIX the "fear index" and reading any rise as pessimism. It measures expected volatility in both directions, not price direction.
Bitcoin dominance
Bitcoin's share of total cryptocurrency market capitalization.
Why it matters: It indicates whether capital is concentrated in Bitcoin or spread across the rest of the market.
How to read it: A decline in dominance tends to accompany rotation into altcoins — but only when the market as a whole is rising.
Common mistake: Comparing Bitcoin dominance with "altcoin dominance." Since the latter is defined as 100% minus the former, the chart is a mathematical mirror image: the correlation is always exactly -1 and carries no new information.
TOTAL2
The market capitalization of all cryptocurrencies except Bitcoin.
Why it matters: It measures the size of the altcoin market as a block.
How to read it: Measured in dollars, it shows size. Measured in Bitcoin (TOTAL2/BTC), it shows relative strength isolated from the market's overall direction.
Common mistake: Forgetting that the market cap includes extremely illiquid tokens with prices that could not withstand actual selling. TOTAL2 overstates the segment.
TOTAL3
The market capitalization of all cryptocurrencies except Bitcoin and Ethereum.
Why it matters: It isolates the most speculative segment of the market.
How to read it: Comparing TOTAL3 with TOTAL2 shows how much of the altcoin move is just Ethereum.
Common mistake: Using TOTAL3 as a proxy for "adoption." It measures price, not usage.
TVL — Total Value Locked
The total value deposited in decentralized finance protocols.
Why it matters: It approximates the economic size of the DeFi sector.
How to read it: Separate growth from new capital inflows from growth driven by the appreciation of the deposited assets.
Common mistake: Reading a rise in TVL as new money coming in. If the price of the deposited tokens rises, TVL rises without anyone having deposited anything.
Funding rate
A periodic payment between longs and shorts in perpetual futures contracts, which keeps the price tethered to the spot market.
Why it matters: A strongly positive funding rate indicates an excess of leveraged long positions.
How to read it: Extreme values signal stretched positioning and the risk of a cascading liquidation.
Common mistake: Treating high funding as an automatic sell signal. It can stay elevated for weeks in a strong trend.
Open interest
The total number of derivatives contracts outstanding at a given moment.
Why it matters: It measures how much leverage exists in the system.
How to read it: Rising open interest with a rising price indicates new money entering long. Rising open interest with a falling price indicates leveraged selling.
Common mistake: Confusing open interest with volume. Volume is how much was traded; open interest is how much remains outstanding.
Realized cap
The sum of each unit of the coin's value at the price it last moved on the blockchain.
Why it matters: It approximates the aggregate average cost basis of holders, rather than the current market value.
How to read it: It serves as a baseline for judging whether the market is in aggregate profit or loss.
Common mistake: Treating it as "money actually invested." Not every on-chain movement is a purchase.
MVRV
The ratio between market value and realized cap.
Why it matters: It indicates whether holders are, on average, in profit or in loss.
How to read it: Very high values have historically coincided with tops; very low values, with bottoms.
Common mistake: Using fixed thresholds from past cycles. The distribution of MVRV has shifted as the market has matured.
Correlation
A statistical measure of how much two series move together, ranging from -1 to +1.
Why it matters: It allows relationships that look real to the naked eye to be compared objectively.
How to read it: Close to +1: strong co-movement. Close to 0: little linear relationship. Close to -1: strong inverse movement.
Common mistake: Calculating correlation on price levels instead of returns. Two series that only rise over the long run will produce a high correlation even with no real relationship — the classic spurious correlation.
Causality
A relationship in which one variable actually brings about the change in another.
Why it matters: It is what most people actually want to know, and it is what correlation does NOT answer.
How to read it: Establishing causality requires theory, experiment, or statistical identification — it is not enough to observe two lines rising together.
Common mistake: Concluding that A causes B because the correlation is high. Often a third factor moves both, or the relationship is a coincidence of the chosen period.
Lead-lag
A technique of shifting one series in time to test whether it tends to move before the other.
Why it matters: It helps distinguish which variable leads and which one reacts.
How to read it: Look at the shape of the correlation-by-lag curve, not just the peak. An isolated peak surrounded by low values is usually noise.
Common mistake: Testing dozens of lags and reporting only the best one. The more combinations you test, the higher the chance of finding one that looks good by pure chance.
Base 100
A normalization in which both series start at a value of 100 at the beginning of the window.
Why it matters: It allows growth pace to be compared across series with completely different scales.
How to read it: The distance between the lines shows relative performance since the start of the window.
Common mistake: Forgetting that the result depends entirely on the chosen start date. Changing the start date changes the story the chart tells.
Fed net liquidity
An estimate of effectively available liquidity: Fed assets minus Treasury cash and minus reverse repo balances.
Why it matters: It captures the net effect of three forces that the balance sheet alone hides.
How to read it: It is meant for reading trends over weeks and months, not for day-to-day moves.
Common mistake: Presenting it as an official measure. No central bank publishes this: it is a conventional approximation, with debatable choices about what to include.
DXY and the broad dollar index
The DXY is a proprietary ICE index covering six developed-market currencies. The Fed's broad index uses about 26 currencies, trade-weighted.
Why it matters: Both measure dollar strength, but not the same thing.
How to read it: This site uses the Fed's broad index, which is public domain. It includes the yuan, the peso, and the real, and so it can diverge from the DXY.
Common mistake: Comparing a broad-index reading with a DXY level as if they were interchangeable. The base periods and the currency baskets are different.