The cleanest leading signal for Bitcoin comes from the Federal Reserve’s open API. It is not a proprietary on-chain metric. It is not a paid sentiment feed. It is a single formula built from four free FRED series, and it leads crypto price action by weeks.
At Tacavar, we run this signal through our ingestion pipeline every morning before the U.S. cash open. It costs nothing. It takes under ten lines of Python. And it outperforms most of the $2,000-a-month macro dashboards we have tested.
What Net Liquidity Actually Measures
Net liquidity is the effective money supply the Federal Reserve pushes into or drains from the banking system. The formula is simple:
DFF + WALCL - RRPONTSYD*1000 - WTREGEN
- DFF: the Fed funds effective rate.
- WALCL: the Fed’s total balance sheet assets.
- RRPONTSYD: the overnight reverse repo facility, scaled by 1,000.
- WTREGEN: the Treasury General Account balance.
Add the Fed’s rate environment to its balance sheet, then subtract the two largest liquidity sinks. The result is the net dollar liquidity available to chase risk assets. When this number rises, Bitcoin tends to follow within one to three weeks. When it falls, drawdowns accelerate.
The mechanism is mechanical. Reverse repo and the TGA are both non-circulating dollars parked at the Fed or Treasury. When they shrink, those dollars re-enter the banking system and flow into equities, credit, and crypto. When they expand, the system tightens.
Why Hedge Funds Charge $2,000/Month for This
Macro research shops package this exact formula into glossy PDFs and Bloomberg overlays. They add narrative, color-coded regime maps, and weekly Zoom calls. The subscription tiers run from $500 to $2,000 per month.
The value they sell is not the math. It is curation and timing. A founder running her own stack does not need the curation. She needs the raw signal, a cron job, and a Postgres table. The rest is execution.
Tacavar’s view: if you can compute a signal in ten lines, you should own the compute layer, not rent the PDF.
The 10-Line Python Implementation
Here is the core of our daily ingestor, fred.py. It pulls the four series, aligns them by date, and writes the derived net-liquidity series to our unified signals table.
import pandas as pd
from fredapi import Fred
fred = Fred(api_key=os.getenv("FRED_API_KEY"))
dff = fred.get_series("DFF")
walcl = fred.get_series("WALCL")
rrp = fred.get_series("RRPONTSYD") * 1000
wtregen = fred.get_series("WTREGEN")
net = pd.concat([dff, walcl, rrp, wtregen], axis=1)
net.columns = ["dff", "walcl", "rrp", "wtregen"]
net["net_liquidity"] = net["dff"] + net["walcl"] - net["rrp"] - net["wtregen"]
net["net_liquidity"].to_sql("fred_net_liquidity", engine, if_exists="append")
That is the entire extraction. A daily cron fires this at 06:00 UTC, before New York wakes up. The table feeds our internal dashboards and our crypto macro scoring model.
How It Performed During the Last Three Crypto Cycles
2020-2021: Net liquidity expanded from roughly $4 trillion to over $8 trillion. Bitcoin rose from $7,000 to $64,000. The signal led the major impulse by roughly ten days.
2022: The Fed drained liquidity aggressively. Net liquidity collapsed. Bitcoin fell from $47,000 to $16,000. The drawdowns tracked the liquidity curve with a lag of one to two weeks.
2023-2024: Liquidity stabilized, then expanded again. Bitcoin rallied from $16,000 through new highs. Each local top in net liquidity preceded a local top in BTC by several sessions.
The correlation is not perfect. No macro signal is. But the hit rate on direction is high enough that Tacavar weights this input at the top of our free alpha stack. When net liquidity diverges from price, we pay attention.
Integrating It Into Your Existing Stack
Most trading systems already ingest price, volume, and on-chain data. Adding macro is a schema change, not an architecture change.
At Tacavar, we store FRED series in the same Postgres instance that holds our market data. A single SQL join gives us net liquidity aligned to daily OHLC. From there, our models treat it as a feature like any other. The difference is that this feature is exogenous to crypto markets. It carries information the order book does not.
If you run Python, fredapi is the only dependency. If you run Go or Rust, the FRED REST endpoint is plain JSON. There is no SDK lock-in, no vendor contract, and no rate limit that matters for a daily cron.
The Free API Key Nobody Talks About
FRED.stlouisfed.org issues API keys instantly. No application. No approval queue. No enterprise sales call. You register, copy the key, and start pulling series.
The key is free for standard usage. The documentation is thorough. And the data is authoritative. This is the Federal Reserve’s own publication layer. It is not a third-party aggregator with stale prints or missing revisions.
Most builders in the crypto trading space have never visited the FRED registration page. That is the edge. The signal is public. The attention is not.
Tacavar’s signal ingestion pipeline automates FRED, Wikipedia pageviews, and 12 other free macro feeds into a unified trading dashboard. See how it works at tacavar.com.