Yesterday at 11:38 UTC this desk published a board and a claim. Four dollar–yen strikes expired at the 14:00 cut — $1.3bn at 156.50, $1.2bn at 158.00, $746mln at 159.50, $1.2bn at 160.25, $4.446bn in all — and we said 73.01 per cent of it sat at prices nothing visible would reach, leaving 158.00 as the only strike where flow should be more than noise. The day has closed and the claim is correct in both halves. Dollar–yen ran 157.84 to 158.39, a 55.0-pip day. The 156.50 block finished 134.0 pips below the low, 159.50 finished 111.0 pips above the high, 160.25 finished 186.0 pips above it, and $3.246bn — 73.01 per cent, to the decimal we published — expired at levels the market never saw. The 158.00 strike was reached. It was reached by 16.0 pips, and the pair closed back above it. So the reachability test passed, and passing it bought you nothing, because the two moves that actually mattered yesterday happened an hour and fifty minutes before the cut and four hours and forty minutes after it.
The arithmetic scores, and three of the four blocks were never in play
Take the scoring plainly, because it is the cheapest part of this and the part a system can automate. The day’s low was 157.84 and the day’s high was 158.39, which is 55.0 pips of range for a Wednesday carrying $4.446bn of expiry. Against that:
- 156.50, $1.3bn, 29.24 per cent of the board. 134.0 pips below the low. In units of the realised day, 2.44 ranges away.
- 158.00, $1.2bn, 26.99 per cent. Traded through, by 16.0 pips, at roughly 12:10 UTC.
- 159.50, $746mln, 16.78 per cent. 111.0 pips above the high — 2.02 ranges.
- 160.25, $1.2bn, 26.99 per cent. 186.0 pips above the high — 3.38 ranges.
We pre-registered 156.50 as 4.3 Asian session ranges away, using the 35.7-pip Tokyo range that was the only measurement available at 11:38. The realised day was 1.54 times that range, so in full-day units the distance was 2.44 rather than 4.3. The estimate was conservative by a factor of about 1.8 and the conclusion survived it comfortably, which is the useful thing about expressing a strike as a multiple rather than a direction: you can be wrong about the denominator by eighty per cent and still be right about whether the thing is reachable.
Sixteen pips is twenty-nine per cent of the day, and it reverted
Here is the part that should change what your schedule does rather than what you think about the market. We said 158.00 was the one level where flow should be more than noise. It was reached. The penetration was 16.0 pips, which is 29.1 per cent of the entire day’s range, and by the close — 158.12 at one publisher, 158.18 at two others — the pair was back above the strike it had broken.
So a rule that fires on “spot reaches a strike carrying $1.2bn” fired correctly and handed you a sixteen-pip excursion that gave itself back. The reachability test was answering the right question and the wrong version of it. What a sizing decision needs is not will spot touch this strike but how far through it does spot have to go before the notional behind it is doing the moving, and on a 55-pip day sixteen pips does not clear that bar. We have no threshold to offer you, because we have one observation. What we can say is that the test we published is binary and the quantity it is standing in for is not, and that the binary version will keep returning true on days like this one.
The expiry window contained neither of yesterday’s two moves
The strongest argument against building a news window around the New York cut is what else happened. The low of 157.84 was set at about 12:10 UTC, as the ten-year yield fell to the 5.07 to 5.08 per cent area — two publishers, one basis point apart on where exactly — which is one hour and fifty minutes before the cut. Then at roughly 18:40 to 19:00 UTC, four hours and forty minutes after the cut, a wire report on Gulf embassy evacuations lifted crude, the ten-year and the dollar together, and dollar–yen printed 158.34. The State Department subsequently narrowed the order to Saudi Arabia alone and most of the move faded.
Note what that leaves. The day’s two highs are 158.38 in the Asian morning at about 00:20 UTC and 158.34 on the embassy headline at about 18:40 — eighteen hours apart, four pips apart, and both roughly 111 pips below the 159.50 strike. Both of the day’s actual catalysts sat outside the expiry window, one on each side of it. A filter that widened around 14:00 UTC yesterday spent its risk budget on the quietest part of the session and had nothing left on either side.
What a 73 per cent board is actually telling your scheduler
The uncomfortable reading is that the board was good information and the wrong input. Every notional on it was published accurately and confirmed in text. The concentration figures we derived were right. The distance arithmetic was right. And the operational content of the whole exercise reduced to one sentence: three of four blocks are decoration, and the fourth is within noise of spot.
That suggests the board belongs in your schedule as a veto rather than a trigger. A board whose mass sits two to three session ranges away is positive evidence that the cut minute does not need special treatment — which is the opposite of how an expiry headline reads and the opposite of what a widened stop at 14:00 does. On yesterday’s numbers the correct action was to leave the window alone and keep the budget for an unscheduled headline, and the unscheduled headline duly arrived at 18:40 and was retracted by a government department within the hour. We are not claiming that generalises from one day. We are claiming that the day it failed on is the day the board looked most informative.
What this does not tell you
We cannot tell you where spot printed at 14:00 UTC itself. Nobody publishes a dollar–yen fix for the cut minute, so “158.00 was reached” means reached during the day, at about 12:10, and not reached at the cut. The 16.0-pip figure is therefore the day’s maximum penetration and not a measurement of anything that happened in the expiry window.
The three publishers we read give three closes six pips apart and three highs five pips apart, so every distance above carries a few pips of error bar. We have used 158.39 as the high because it is the widest of the three, which makes the misses on 159.50 and 160.25 the smallest they can honestly be stated as. We still have no per-level sizes for the retail order book — they remain inside an image — so the band we used yesterday to define “visible” was a set of price levels with no quantities attached. And the entire argument rests on one day with one board. A 73 per cent unreachable board that preceded a violent cut would falsify the veto idea immediately, and we have not looked for one.
