Every time someone builds a liquidity signal, they reach for the same two series. M2, and the size of the central bank balance sheet. Both feel authoritative because they are the headline numbers everyone quotes. The problem is that by the time those numbers land on your screen, they are describing a world that already moved on.
M2 comes out weekly or monthly depending on where you look, and it gets revised. The balance sheet total is weekly. That sounds fine until you remember that markets reprice liquidity intraday, and a weekly figure that also revises is telling you where you were, not where you are. If your scorecard is fed by these, it is looking in the rear view mirror and calling it a forecast.
The good news is that most of what drives the liquidity that actually reaches risk assets is buried in a few daily series that publish faster and revise less. Once you know which levers move, you can build a proxy that leads the headline instead of trailing it.
What actually moves liquidity day to day
The number you care about is roughly how much cash is sloshing around the banking system and available to chase assets. In the US plumbing, three daily variables move that around more than almost anything else, and they net against each other in a way that is easy to reason about once you see it.
- Reserve balances. This is the cash commercial banks hold at the central bank. When reserves go up, the system has more fuel. When they drain, funding gets tighter at the margin. This is the closest thing to a direct read on system liquidity, and it updates far more often than the balance sheet total that people quote.
- The Treasury General Account, or TGA. Think of it as the government's checking account. When the Treasury builds up its cash balance, say after a bond issuance or around tax dates, that money leaves the private system and sits idle. When the TGA is drawn down, cash floods back in. A rising TGA is a drain, a falling TGA is an injection.
- Reverse repo, or RRP. This is a parking lot for cash that money market funds do not want to lend elsewhere. Money in RRP is money on the sidelines. When RRP falls, that cash is leaving the parking lot and looking for a home, which tends to be additive to system liquidity.
The rough mental model most macro people use is something like: net liquidity moves with the balance sheet, minus the TGA, minus RRP. You do not need to treat that as a precise formula. Treat it as a direction finder. If reserves are climbing while the TGA and RRP are both falling, liquidity is loosening, and you can usually see that days or weeks before the tidy weekly aggregates confirm it.
Building a proxy that leads the headline
The point of tracking the daily series is not to replace the official numbers. It is to front-run them. The official M2 or balance sheet print is a lagging confirmation of moves you can already infer from the plumbing.
A workflow I have found holds up reasonably well:
- Pull the daily reserve, TGA, and RRP series and compute a simple net liquidity proxy. Do not overfit the weights. Additive reserves, subtractive TGA, subtractive RRP is enough to start.
- Track the rate of change, not just the level. Liquidity turns show up in the slope before they show up in the level. A proxy that is still high but has clearly rolled over is often more informative than one that is low and stabilizing.
- Watch the calendar. A lot of the sharp TGA moves are mechanical and scheduled around tax dates and auction settlements. If you know a rebuild is coming, you can separate a drain that is a policy signal from a drain that is just the plumbing doing plumbing.
- Only then look at the weekly headline, and use it to check that your proxy has not drifted from reality. If the proxy and the headline keep disagreeing, your proxy is broken, not clever.
The honest caveat is that these series interact in ways that are not always clean. A falling RRP is usually additive, but if the cash is going straight into the TGA it can be a wash. So do not read any single line in isolation. Read the net.
Timing the lag to the crypto response
Say your proxy turns. Liquidity is loosening. The instinct is to expect crypto to rip immediately. It rarely works like that, and the gap between the liquidity turn and the price response is where most of the mistakes happen.
Liquidity is a slow tide, not a trigger. When system cash starts expanding, it does not teleport into the riskiest, longest-duration assets on the same day. It tends to move outward in stages. Front-end funding eases first, then it shows up in broader risk appetite, and crypto, which sits near the far end of the risk curve, tends to respond with a lag that historically runs on the order of weeks rather than days. I would not hard-code a specific number, because the lag stretches and compresses depending on how tight funding was to begin with and how levered the crypto market already is.
A few rules of thumb that keep me honest here:
- Treat a liquidity turn as a change in the odds, not a signal to size up on day one. It shifts the backdrop from headwind to tailwind. It does not tell you the entry.
- Direction is more reliable than magnitude. The proxy is good at telling you liquidity is loosening or tightening. It is bad at telling you how much price that is worth. Do not build a position that only works if a specific dollar move shows up.
- Beware the head fake around scheduled TGA rebuilds. A sharp mechanical drain can look like a liquidity turn and shake people out right before the underlying trend reasserts. Knowing the calendar is half the edge.
- Confirm with something orthogonal. If your liquidity proxy is turning bullish but on-chain flows and funding are still saying stress, wait. Two independent reads agreeing is worth more than one read shouting.
This is roughly how we think about the macro tab inside Blockcircle, layering a fast liquidity proxy against on-chain and positioning data so the scorecard is not leaning entirely on numbers that revise. The proxy gives you the early lean, and the slower confirmations tell you whether to trust it.
Where this breaks
The main failure mode is treating the proxy as gospel. It is a model of the plumbing, and models leak. The relationship between net liquidity and risk assets is loose in the short run and only tightens over longer windows. Anyone who tells you a daily liquidity number gives you a clean trade signal is selling a backtest that got lucky on one regime.
The second failure mode is ignoring the rest of the world. US plumbing is a big driver, but foreign central bank action, currency swings, and offshore dollar funding all move the same tide. If you are only watching the domestic series, you will occasionally get blindsided by a move that started somewhere you were not looking.
None of this is a crystal ball. What it buys you is a few days to a few weeks of lead time over people still waiting for the revised weekly print, plus a clearer sense of when a liquidity turn is real versus when it is the calendar playing tricks. That head start is usually enough to be early instead of late, which in this game is most of the job.