Reports on the death of month-end were greatly exaggerated

¡Payments!

The Jackson Hole program is out, and Warsh’s slot is called “Opening Remarks” and it’s only 30 minutes. This adds to my view that Warsh will simply try to move through and past this event without saying anything substantive. The whole program sounds extremely not market moving.

https://www.kansascityfed.org/research/jackson-hole-economic-symposium/2026/

Next week has some interesting activity with the RBNZ probably tops for me, plus NFP, BoC, and Canadian jobs data. Monday is month-end, and while it’s the U.K. bank holiday, there will still be a WMR fixing because the only time you don’t get a WMR fix is when both U.S. and U.K. are away.

Stocks are 4% higher this month, which normally leads to USD selling at month end, in theory. The month-end models (MEM) mostly stopped working for a while after COVID, but they continued to work in certain currencies and on certain time frames. And their performance has picked up again over the past 12-18 months.

AUDUSD, for example, has worked well, especially in the final half-hour. This chart shows the return of long AUDUSD from 15:30 to 16:00 LDN (i.e., the 30 minutes into WMR, which is normally 10:30 a.m. to 11:00 a.m. NY).

And here is long AUDUSD for the full four hours before WMR. It’s been great lately but was sus for seven years around COVID. When I say “2.5% filter” that means: sum of SPX and NDX returns was > +2.5% for the month.

The chart is not particularly sensitive to what you choose for the parameter. As long as stocks are up, real money needs to sell USD on the last day of the month. As the SPX + NDX monthly return flips negative, they need to buy USD. The hedging behavior is opposite in JPY, so you often see USD/JPY rally into month end while every other USD pair sells off. The three charts I have shown so far (AUD, AUD, JPY) all have ~60% hit rate and positive skew.

In 2022, I wrote a piece titled: RIP Month-End Models. But I spoke too soon. The models have returned to relevance since then. The upshot is that there is still month-end hedging in the direction you would expect, but because it is spread out and has become more sophisticated and discretionary, the signals are not as powerful. The final thirty minutes still deliver the bacon because that is when the most risk-averse benchmark huggers channel their gigantic flows.

For completeness, here are GBPUSD and EURUSD (7 a.m. to 11 a.m. on top, last 30 minutes on bottom). You can see that 2020 to 2023 was a tough period for the models but they are better of late.

If trading was easy, it wouldn’t pay so well.

Even if you are a longer-term FX trader, or a medium-term macro trader, you can find ways to make these huge waves of volume and volatility into month end useful. They can offer great entry points and lower transaction costs if you want to turn large size.

Finally, the flow story for AUD looks great right now. I discussed the dividend thing yesterday and have shown today that AUDUSD is one of the most reliable month end trades. So, you have fundamentals and flow both pushing pretty hard in the same direction for the Oz. RHS.

---

Calendar

Dear next week’s calendar: You have nothing to be ashamed of. You are plenty interesting.

---

Final Thoughts

1. I like to keep things short on Friday.
2. That said, this week’s Friday Speedrun is longer than usual, because fun. I don’t automatically send Friday Speedrun to all am/FX subscribers because I don’t want to send people something they didn’t ask for. If you would like to receive Friday Speedrun, please sign up here.

https://www.spectramarkets.com/sign-up/

3. Crypto still trades like it’s overbought. When RSI goes to 85, the thing usually either consolidates or reverses. My reversal call is looking increasingly stale as we consolidate day after day.

Nanu Nanu.

HT AC

good luck ⇅ be nimble