The Macro Investing Tool

The Data Did the Work

Last week I said the market had a September hike priced at about one in three, and that I expected the rest to get priced out as the data did the work. The data did the work. Payrolls contracted in July, headline CPI came in at 3.4% and falling, PPI printed flat against expectations of a rise, and retail sales fell 0.6%.

The dollar was the thing I said mattered most, and the Risk Monitor has flipped it to falling this month, with DXY at 99.4 and sinking after the retail sales miss. Gold flipped to rising on the Monitor the same week and sits near $4,400. I've long said gold follows the yield curve, and as the curve re-steepens and the hikes come out of the pricing, it should keep rising from here.

Now the miss. Last week I leaned against the page on crude at $77 and said I expected it to go lower... Alas, it's gone to $82, and oil is still the one asset where the Monitor and the regime agree on up. Wrong so far. I still think the supply premium unwinds when the fear does, but the tape hasn't agreed yet.

The long bond stays on the watch list at 4.69%, a touch higher on the week, though Treasuries caught a decent bid on Friday as the hike bets came out. This is where the shortage of US liquidity shows up: the government keeps issuing into a system without enough spare money to absorb it, so the long end stays stickier than anyone wants. It needs solving, but it's far from a crisis.

Now to the new stuff, because this week the dashboard grew a whole new way of looking at the relationship between liquidity and bitcoin, built from member feedback. The famous chart has become a full workbench: the long-run overlay, the year-on-year waves, the leads and lags, and the excess liquidity view, all live and all playable.

It settles an argument that's been running all year. The famous 0.9 correlation is a levels relationship: liquidity and bitcoin share the same long debasement trend, and that's as true as ever. The timing relationship is looser and always has been, roughly 13 weeks of lead with a correlation around 0.4. Two different claims were living in one chart, which is why this year confused so many people.

Because this year...the waves broke apart. Measured era by era, the liquidity wave and the bitcoin wave have moved together in every period since 2012, and the 2022 to 2026 window is the one true anomaly. Liquidity remains the dominant factor. What we have is a dislocation, and the new tools show you its size and, more importantly, its cause.

The cause is excess liquidity. Money only reaches the speculative end of the market when there's more of it than the economy needs, and the surplus goes hunting. Global excess has only just crossed back to neutral after a year in scarcity, while the US narrow measure still sits around -4%.

That's enough to fund one trend, not two. The marginal dollar is going to the AI trade, which is why tech sits above its long-run trend while bitcoin sits below it. There just isn't enough excess money to fund both trends at once. Yet.

The new tools also clean up a statistical trap: around two-thirds of bitcoin's year-on-year deterioration over the past six months is a base effect, the calculation lapping the October 2025 peak, and only about a third is price. The dislocation is real, but smaller than the headline number makes it look.

So crypto remains deeply oversold against what liquidity is doing and where it's going. The catalyst is the missing piece. What I most want to see is US narrow liquidity turning up as the curve steepens, because in this era, the plumbing measure is the one that tracks bitcoin, and it hasn't turned yet at -0.8% year on year. The data isn't real-time... early days.

While you're in there, play with the lead drift monitors. The leads between conditions, liquidity, and the cycle aren't fixed; rather, they drift with the regime, and the monitors now measure that drift continuously. As more people learn the leads, the leads shrink, and the liquidity-to-bitcoin lead has compressed towards coincident as the market front-runs it. Reflexivity comes for every good signal, so we measure it rather than assume it.

I'll also correct myself on financial conditions. The big drop in the GMI FCI isn't mainly the dollar, but mainly the long end: the attribution on the page runs the yield leg at roughly three times the dollar leg. The dollar's damage went into global liquidity instead, taking the year-on-year rate from 11% in February to under 7%. Both legs are turning: the dollar is rolling over, and the long-end leg resolves as the hikes get priced out.

We're also running two constructions of the FCI with different weightings, and right now they disagree, with the new build reading tight and the old build reading easy. We're testing which one earns the job. In the meantime, the house call is logged in advance... the market likely looks through this tightening.

The reason is breadth, and it's on chart 24 of the business cycle tab, one of the most useful charts on the dashboard right now. A tightening only transmits when enough legs squeeze at once, and only two of the four currently are: the long end and oil — while the front end and the dollar ease. Squeezes this narrow have historically been absorbed. It starts to bite at three legs or more, and the swing leg is the dollar... which is busy doing the opposite.

Underneath all of this, the business cycle is strong, and inflation isn't arguing. The ISM sits at 55.6 with new orders at 56.7 above the headline, core CPI is 2.5% and falling, and the labor market is weak in level but turning in lead, with temp help crossing positive for the first time in over three years. Weak retail sales, a weak jobs print, benign inflation: that mix retires the hike case without threatening the cycle.

Remember the calendar too. This is a midterm election year, and September is historically the weak patch, but there's no sign of pre-election weakness yet, with the S&P sitting at all-time highs. Speculative positioning is broadly neutral, and the one stretched reading, bitcoin futures at +2.5 sigma long, I still read as basis trades rather than conviction. Whether any softness lands pre-election or post-election is still open, and nothing in my DeMark counts concerns me.

One quiet mover worth a line: China. The PBoC balance sheet is growing 9.3% year on year, up two points in three months, the biggest standardized move on the page this week. The liquidity wave is building outside the US even while Washington sits on its hands.

So I stay firmly risk-on. Technology carries the baton until the liquidity mix shifts, and when excess liquidity comes through, the tables turn, and crypto's expected outperformance starts to play out. Into next week: the dollar first, September pricing going to zero, the long bond at 4.69%, US narrow liquidity for the turn, and my oil lean against the page. Expect seasonal chop along the way. It's the price of admission to the strongest part of the cycle.

— Raoul