the week ahead — august 3 to 7. jobs week.
friday, 12:30 UTC (15:30 мск): US non-farm payrolls, the single most-watched data release in fx. wednesday: the ADP private payrolls report, which the market habitually — and mistakenly — treats as a preview.
this week's briefs build up to friday from the measurement side:
monday — the jobs data family: four numbers, one labor market, and what each instrument actually measures.
tuesday — revisions: why the number you trade on friday is a draft.
wednesday — the two surveys inside one report, and why they can disagree.
thursday — consensus: how the market's expectation forms, and why the error moves prices rather than the number.
by friday's print you'll know where the number comes from. for what to do about it — our asia desk runs the event live all week: @equilon_mike.
friday, 12:30 UTC (15:30 мск): US non-farm payrolls, the single most-watched data release in fx. wednesday: the ADP private payrolls report, which the market habitually — and mistakenly — treats as a preview.
this week's briefs build up to friday from the measurement side:
monday — the jobs data family: four numbers, one labor market, and what each instrument actually measures.
tuesday — revisions: why the number you trade on friday is a draft.
wednesday — the two surveys inside one report, and why they can disagree.
thursday — consensus: how the market's expectation forms, and why the error moves prices rather than the number.
by friday's print you'll know where the number comes from. for what to do about it — our asia desk runs the event live all week: @equilon_mike.
the jobs data family. four instruments, one labor market.
non-farm payrolls (BLS, first friday): the benchmark. jobs added, unemployment rate, wage growth — the release the fed's employment mandate is effectively graded on. moves fx more than any other data point outside central bank meetings.
ADP (private, wednesday before NFP): built from payroll-processing records. different sample, different method — the single-month correlation with NFP is poor, which the market rediscovers monthly, two days after forgetting it.
initial claims (weekly, thursdays): the highest-frequency read. noisy week to week, but the four-week trend catches labor-market turns earlier than anything monthly.
JOLTS (openings, quits — one month lagged): the structural read. the quits rate is the honest confidence indicator: people don't leave jobs they're afraid of losing.
the ranking by market impact: NFP by a wide margin, then claims when they trend, then ADP on surprise only, then JOLTS for economists.
one market. four rulers. never assume they agree.
non-farm payrolls (BLS, first friday): the benchmark. jobs added, unemployment rate, wage growth — the release the fed's employment mandate is effectively graded on. moves fx more than any other data point outside central bank meetings.
ADP (private, wednesday before NFP): built from payroll-processing records. different sample, different method — the single-month correlation with NFP is poor, which the market rediscovers monthly, two days after forgetting it.
initial claims (weekly, thursdays): the highest-frequency read. noisy week to week, but the four-week trend catches labor-market turns earlier than anything monthly.
JOLTS (openings, quits — one month lagged): the structural read. the quits rate is the honest confidence indicator: people don't leave jobs they're afraid of losing.
the ranking by market impact: NFP by a wide margin, then claims when they trend, then ADP on surprise only, then JOLTS for economists.
one market. four rulers. never assume they agree.
why the jobs number is always a draft.
the payrolls figure is built from a survey of businesses — and businesses respond late. the first print is calculated from a partial sample; as more responses arrive over the following two months, the number is revised, sometimes heavily.
on top of that sits the birth-death adjustment: a statistical model estimating jobs at businesses too new to be surveyed and losses at ones that quietly closed. in stable times the model works; at turning points it systematically misses — it assumes the recent past continues.
once a year, everything is benchmarked against tax records — the closest thing to a full count — and entire years get redrawn at once.
none of this is scandal. it's what measuring a 160-million-person labor market in three weeks costs. but it has a market consequence: desks read the revisions to the previous months as carefully as the headline. a strong print with heavy downward revisions is a weaker report than the headline suggests — and price, increasingly often, reacts to page two.
the payrolls figure is built from a survey of businesses — and businesses respond late. the first print is calculated from a partial sample; as more responses arrive over the following two months, the number is revised, sometimes heavily.
on top of that sits the birth-death adjustment: a statistical model estimating jobs at businesses too new to be surveyed and losses at ones that quietly closed. in stable times the model works; at turning points it systematically misses — it assumes the recent past continues.
once a year, everything is benchmarked against tax records — the closest thing to a full count — and entire years get redrawn at once.
none of this is scandal. it's what measuring a 160-million-person labor market in three weeks costs. but it has a market consequence: desks read the revisions to the previous months as carefully as the headline. a strong print with heavy downward revisions is a weaker report than the headline suggests — and price, increasingly often, reacts to page two.
inside the jobs report: two different surveys wearing one press release.
the establishment survey asks businesses: hundreds of thousands of worksites report how many people are on payroll, hours worked, and earnings. this produces the headline payrolls number and the wage figures.
the household survey asks people: roughly sixty thousand households report who is working, looking, or out of the labor force. this produces the unemployment rate and the participation rate.
why they diverge: the household survey catches self-employment and informal work the establishment survey can't see; the establishment survey double-counts multiple jobholders; and the two samples have very different sizes, so the household numbers are noisier month to month.
the market trades the establishment side first — payrolls and wages hit the algorithms. but the fed's mandate is written against the household side: maximum employment is about people, not payroll records. when the two argue for several months, the argument itself is the story.
this afternoon (15:15 мск): ADP. the preview that isn't — our asia desk has the practitioner's version of that warning today.
the establishment survey asks businesses: hundreds of thousands of worksites report how many people are on payroll, hours worked, and earnings. this produces the headline payrolls number and the wage figures.
the household survey asks people: roughly sixty thousand households report who is working, looking, or out of the labor force. this produces the unemployment rate and the participation rate.
why they diverge: the household survey catches self-employment and informal work the establishment survey can't see; the establishment survey double-counts multiple jobholders; and the two samples have very different sizes, so the household numbers are noisier month to month.
the market trades the establishment side first — payrolls and wages hit the algorithms. but the fed's mandate is written against the household side: maximum employment is about people, not payroll records. when the two argue for several months, the argument itself is the story.
this afternoon (15:15 мск): ADP. the preview that isn't — our asia desk has the practitioner's version of that warning today.
where "consensus" comes from, and why the error is the event.
before every major release, news services survey several dozen bank and fund economists. the median of their forecasts becomes "consensus" — the number in every calendar app. around it sits a range, often a wide one, that the headline number will be compared against.
two subtleties worth knowing.
the whisper number: what positioned traders actually expect, as opposed to what economists published. when the whisper drifts away from the official consensus — after a strong ADP, say — a print can "beat consensus" and still disappoint the market that had quietly repriced higher.
the reaction function: the same surprise means different things in different regimes. a hot jobs number is dollar-positive when the market fears cuts are premature, and dollar-negative when it fears the economy is overheating into inflation. the sign of the reaction tells you what the market currently cares about — often the most useful information of the release.
data doesn't move markets. surprise does — measured against positioning, not against the calendar app.
before every major release, news services survey several dozen bank and fund economists. the median of their forecasts becomes "consensus" — the number in every calendar app. around it sits a range, often a wide one, that the headline number will be compared against.
two subtleties worth knowing.
the whisper number: what positioned traders actually expect, as opposed to what economists published. when the whisper drifts away from the official consensus — after a strong ADP, say — a print can "beat consensus" and still disappoint the market that had quietly repriced higher.
the reaction function: the same surprise means different things in different regimes. a hot jobs number is dollar-positive when the market fears cuts are premature, and dollar-negative when it fears the economy is overheating into inflation. the sign of the reaction tells you what the market currently cares about — often the most useful information of the release.
data doesn't move markets. surprise does — measured against positioning, not against the calendar app.
from the asia desk — NFP edition.
Mike's plan for the print went up this morning, and the detail worth noticing is that it was written in advance — the whole point, in his words, is that a plan made before the event can't be improvised during it. flat into the number, spectator for the first fifteen minutes, attention on the 15-to-90-minute window where genuine repositioning shows itself.
and after the print, his grading sheet: did the first move survive ninety minutes. did page two — revisions and wages — agree with the headline. what does monday do with it. three questions that work whatever the number says.
if this week's briefs here were the anatomy of the jobs report, tonight on his channel is the surgery. watching a major event flat, with a written framework, remains the cheapest education available in fx.
@equilon_mike.
Mike's plan for the print went up this morning, and the detail worth noticing is that it was written in advance — the whole point, in his words, is that a plan made before the event can't be improvised during it. flat into the number, spectator for the first fifteen minutes, attention on the 15-to-90-minute window where genuine repositioning shows itself.
and after the print, his grading sheet: did the first move survive ninety minutes. did page two — revisions and wages — agree with the headline. what does monday do with it. three questions that work whatever the number says.
if this week's briefs here were the anatomy of the jobs report, tonight on his channel is the surgery. watching a major event flat, with a written framework, remains the cheapest education available in fx.
@equilon_mike.
Q&A: "is august even worth trading for a beginner?"
the honest answer: mostly no — and that makes it one of the most valuable months of the year.
what august is bad for: learning from your fills. thin books distort everything — winners run further than they should, stops fill worse than they should, and both teach the wrong lesson. a strategy validated in august tape has not been validated.
what august is good for: learning the machinery. events still fire on schedule — jobs week just ran, CPI comes next week, the fed's minutes after that — but the market processes them slowly enough to watch the gears turn. this cycle's briefs are built for exactly that use.
a reasonable august curriculum: watch each scheduled event flat and write down what the first hour did versus the second day. journal as if you were trading, without the positions. read the foundations — the archive of this channel from may is structured for exactly that.
september brings the liquidity back. traders who spent august watching tend to meet it better prepared than traders who spent august paying for entertainment.
the honest answer: mostly no — and that makes it one of the most valuable months of the year.
what august is bad for: learning from your fills. thin books distort everything — winners run further than they should, stops fill worse than they should, and both teach the wrong lesson. a strategy validated in august tape has not been validated.
what august is good for: learning the machinery. events still fire on schedule — jobs week just ran, CPI comes next week, the fed's minutes after that — but the market processes them slowly enough to watch the gears turn. this cycle's briefs are built for exactly that use.
a reasonable august curriculum: watch each scheduled event flat and write down what the first hour did versus the second day. journal as if you were trading, without the positions. read the foundations — the archive of this channel from may is structured for exactly that.
september brings the liquidity back. traders who spent august watching tend to meet it better prepared than traders who spent august paying for entertainment.
the week ahead — august 10 to 14. inflation week.
wednesday, 12:30 UTC (15:30 мск): US CPI for july — the release that steers rate expectations more directly than any other input between fed meetings. thursday: PPI, the pipeline version. friday: retail sales.
this week's briefs are the measurement story behind wednesday:
monday — the CPI basket: what's actually in it, and why shelter dominates.
tuesday — core inflation: the logic of excluding food and energy, properly explained.
wednesday — release mechanics: the lockup, the machine-readable feed, and why the first second belongs to algorithms.
thursday — PPI: inflation's supply chain, and when it matters.
one number, measured one way, moving trillions. by wednesday's release you'll know exactly what kind of number it is. for the trading side of the same night — @equilon_mike runs it live.
wednesday, 12:30 UTC (15:30 мск): US CPI for july — the release that steers rate expectations more directly than any other input between fed meetings. thursday: PPI, the pipeline version. friday: retail sales.
this week's briefs are the measurement story behind wednesday:
monday — the CPI basket: what's actually in it, and why shelter dominates.
tuesday — core inflation: the logic of excluding food and energy, properly explained.
wednesday — release mechanics: the lockup, the machine-readable feed, and why the first second belongs to algorithms.
thursday — PPI: inflation's supply chain, and when it matters.
one number, measured one way, moving trillions. by wednesday's release you'll know exactly what kind of number it is. for the trading side of the same night — @equilon_mike runs it live.
what's actually inside the CPI.
every month, government price collectors record tens of thousands of price quotes — goods and services, in stores and online, across dozens of cities. each item is weighted by how much of household spending it represents, from consumer expenditure surveys.
the weight that dominates everything: shelter, roughly a third of the basket. and most of it isn't even observed rent — it's owners' equivalent rent, an estimate of what homeowners would pay to rent their own homes. an imputed price, for a transaction that never happens, carrying the largest weight in the world's most-watched inflation number.
shelter also moves slowly by construction: leases reprice once a year, so measured rent inflation lags the actual rental market by many months. a meaningful share of any CPI print was, in effect, determined last year.
none of this makes CPI wrong. it makes it specific — a particular basket, measured a particular way, with known lags. wednesday, when the number lands, you'll know what kind of object landed.
every month, government price collectors record tens of thousands of price quotes — goods and services, in stores and online, across dozens of cities. each item is weighted by how much of household spending it represents, from consumer expenditure surveys.
the weight that dominates everything: shelter, roughly a third of the basket. and most of it isn't even observed rent — it's owners' equivalent rent, an estimate of what homeowners would pay to rent their own homes. an imputed price, for a transaction that never happens, carrying the largest weight in the world's most-watched inflation number.
shelter also moves slowly by construction: leases reprice once a year, so measured rent inflation lags the actual rental market by many months. a meaningful share of any CPI print was, in effect, determined last year.
none of this makes CPI wrong. it makes it specific — a particular basket, measured a particular way, with known lags. wednesday, when the number lands, you'll know what kind of object landed.
the logic of core inflation, properly.
the exclusion of food and energy is the most misunderstood convention in macro. it is not a claim that food and fuel don't matter to households — they matter most of all. it is a claim about signal.
energy prices are set globally: a pipeline headline or an OPEC decision moves them overnight, and monetary policy in washington can do nothing about either. food carries weather. both are volatile and mean-reverting — they spike and retrace on timescales shorter than policy operates on.
policy, meanwhile, works with lags measured in quarters. a central bank steering by headline would be forever chasing noise it cannot influence with a tool that arrives late. so the target's operational focus falls on the persistent components — services, rents, wages feeding into prices. the sticky stuff. that's core.
the practical translation for wednesday's print: headline is the household's reality; core is the committee's steering input. the market prices the committee. when the two lines argue, price follows core — not because the market is heartless, but because the fed is what it's pricing.
the exclusion of food and energy is the most misunderstood convention in macro. it is not a claim that food and fuel don't matter to households — they matter most of all. it is a claim about signal.
energy prices are set globally: a pipeline headline or an OPEC decision moves them overnight, and monetary policy in washington can do nothing about either. food carries weather. both are volatile and mean-reverting — they spike and retrace on timescales shorter than policy operates on.
policy, meanwhile, works with lags measured in quarters. a central bank steering by headline would be forever chasing noise it cannot influence with a tool that arrives late. so the target's operational focus falls on the persistent components — services, rents, wages feeding into prices. the sticky stuff. that's core.
the practical translation for wednesday's print: headline is the household's reality; core is the committee's steering input. the market prices the committee. when the two lines argue, price follows core — not because the market is heartless, but because the fed is what it's pricing.