A US housing release says starts are running at 1.4 million. There were not 1.4 million new US foundations dug that month. The published seasonally adjusted annual rate expresses the month’s adjusted pace on a yearly scale. That unit matters before the data enter a dollar forecast.
The practical question is how to convert an annualized construction pace without losing track of US housing starts or Canadian dollars per US dollar. Multifamily unit counts can swing when a small number of large buildings break ground. The worked example uses a US dollar account, and the historical workbook gives you the observations and arithmetic to check the argument yourself.
Follow a multifamily project through the survey
When excavation begins on an apartment building with 140 units, all 140 units can enter starts at that stage. A single project can move a regional monthly reading. Counting cranes or construction sites would give another number. One builder starting a tower and another starting one detached home each initiated a building, but their unit contributions differ sharply.
A seasonally adjusted annual rate of 1.452 million is equivalent to an adjusted pace of 121,000 units for a month. Dividing by twelve does not reconstruct the actual unadjusted number that broke ground in that calendar month. For actual counts, use the unadjusted table. Keep the annualized level, the month-on-month percent change and the year-over-year comparison in different fields.
Read the permit pipeline alongside financing costs
Higher mortgage rates can change affordability, lender standards and a builder’s willingness to start. Land availability, weather and permits already in hand can delay a response. A national permit-minus-starts gap is an arithmetic difference between flows. It is not an exact count of a fixed cohort waiting to break ground. To study the pipeline, consult the tables on units authorized but not started.
A surprise increase in US starts could support the dollar through growth expectations, but monthly uncertainty and the Canadian housing cycle also matter for USD/CAD. Compare the housing report’s stated confidence interval with the size of the change. A headline within a wide interval deserves less confident language even if a chart moves sharply at the release.
Which unit did the contractor actually start?
Use two sample projects, one detached home and one apartment block. Track permits, excavation and completion for each, keeping counts of buildings separate from counts of housing units. A dozen sites can produce fewer housing units than one apartment building. That difference is large enough to change a regional monthly percentage.
A builder holding a permit may wait for financing, subcontractors or a change in expected selling prices. Another can begin a previously authorized project while the latest pace of permits falls. Read the authorized-but-not-started series when asking about projects waiting in the pipeline. Two monthly flow totals cannot reconstruct the underlying stock by simple subtraction.
If the report includes a percentage change with a wide confidence interval, record the interval beside the estimate in a replay journal. Even a statistically uncertain change can move markets when it differs from expectations. Distinguish the uncertainty about economic output from the certainty that a particular bid or offer printed at a certain time.
How to identify a usable USD/CAD setup on a chart
Open a five-minute USD/CAD replay and hide everything to the right of your intended decision. This is a hypothetical short case, constructed to show how residential construction fits into an execution record. Use New York local time for the chart, retaining the actual offset for the date. For a US futures contract, also retain the exchange session and contract month. An economic reference period, a release time and a bar timestamp belong in separate fields.
In the first diagram the reference high is 1.37496. The candle that trades above it reaches 1.37572 before price returns into the range. To describe that move as a sweep in ICT terminology, mark the existing level first. If the level only becomes obvious after the reversal, save the observation as a retrospective pattern. An increase in volume or a financial headline cannot reveal whose orders traded there.
The displaced leg reaches beyond 1.37306. Read bars five, six and seven as a sequence. A gap between the outer wicks brackets the interval from 1.37320 to 1.37366. It is a chart definition of an imbalance. Price could fill that interval, pass straight through it or never revisit it. Only the first return that satisfies a rule written in advance can provide a candidate entry for this particular exercise.
How to trade the idea on demo and define failure
For the rehearsal, a sell at 1.37344 follows a return into the marked interval. If the order fails to fill, record an unfilled setup and retain the original timestamp. Moving the order toward the next favorable candle after that candle forms changes the test. Keep the macro or account question separate: Units authorized minus units started this month equals the total backlog. A convincing answer about that issue cannot make a missed entry executable.
The protective exit is 1.37616, beyond the sweep, and the planned target is 1.36664. For this short idea the favorable direction in the pair is lower. A live position can experience partial fills, a spread jump or a gap past the stop. For this drawing only we assume all levels execute exactly as stated. The second chart displays one favorable sequence so that the accounting can be checked. It supplies no evidence about how often such a sequence occurs.
Set an expiry for the entry and cancel the study if the invalidation trades before entry or if the spread makes the gross plan infeasible. If a coarse chart touches stop and target within one bar, lower-frequency OHLC cannot settle the order of events. Use properly timestamped executable quotes if available. Otherwise keep a separate ambiguous sample, and state that rule before counting wins. Confidence intervals around national monthly estimates can be wide. Preserve the release’s own uncertainty language before calling a small change decisive.
Worked example in a US dollar account
| Ticket field | Rehearsal value |
|---|---|
| Instrument | USD/CAD retail spot illustration |
| Direction and size | short | 25,000 base units |
| Entry / stop / target | 1.37344 / 1.37616 / 1.36664 |
| Gross risk / reward | USD 49.41 / USD 124.39 |
| Assumed round-trip charge | USD 3.10 |
Calculate the distance before calculating the money. The entry-to-stop distance is |1.37344 minus 1.37616| = 0.00272, or 27.2 pips. The target distance is |1.36664 minus 1.37344| = 0.00680, or 68.0 pips. Keeping absolute distances here avoids giving a short setup a negative risk amount. The sign of the actual result belongs on the fill record after closing.
At 25,000 base units, the distance to the stop represents 68.00 CAD and the distance to the target represents 170.00 CAD. Convert to a USD result separately at each assumed exit by dividing by 1.37616 or 1.36664 Canadian dollars per US dollar. The conversion is slightly different on the two outcomes. Gross reward divided by gross risk is USD 124.39 / USD 49.41 = 2.52. Written as risk:reward this is 1:2.52.
Assume a round-trip cost of USD 3.10 on either completed outcome. We use executable entry and exit prices, so a spread already inside those prices must not be charged again. The simplified net reward is USD 124.39 minus USD 3.10 = USD 121.29. Net planned loss is USD 49.41 plus USD 3.10 = USD 52.51. Net R:R is 1:2.31 because USD 121.29 / USD 52.51 = 2.31. Financing, slippage and tax are excluded from this short rehearsal.
Check the losing branch with the same care as the winning one. If the exit slips by one additional pip against you, the quote-currency impact on this size is 2.50 CAD before any conversion. A plan denominated in dollars should reserve for an adverse fill or skip a setup that cannot fit the intended budget. Thousands of units at an annual rate become thousands at an adjusted monthly equivalent when divided by twelve.
What the historical dollar data show
What did US housing starts actually do across different US market conditions? We used 120 consecutive usable months from January 2015 through December 2024. The first change uses the preceding observation. The exchange rate is Canadian dollars per US dollar. A positive dollar-strength reading below means the US dollar gained against the other currency, after explicitly inverting USD-per-foreign-currency quotes where necessary.
The indicator is US housing starts, expressed in thousands of units, annual rate. The input series is HOUST. We pair it with the EXCAUS exchange-rate history. The comparison uses later-published observations and common reference periods. In particular it contains no archived consensus forecast. A rise from one reference period to another measures change; it cannot tell you whether traders were surprised on the publication day.
First inspect the middle of the distribution. The median signed change was 0.427 percent. The lower and upper quartile observations were -3.897 and 6.322 respectively, using ordered sample ranks for those quartiles. Extreme observations matter for account resilience, but starting with an extreme can give an inaccurate picture of the movement a reader usually encountered in this particular history.
| Year | Observations | Mean change (percent) | Mean USD change vs CAD (%) |
|---|---|---|---|
| 2015 | 12 | 1.205 | 1.477 |
| 2016 | 12 | 1.220 | -0.207 |
| 2017 | 12 | -0.382 | -0.343 |
| 2018 | 12 | -0.415 | 0.437 |
| 2019 | 12 | 3.151 | -0.163 |
| 2020 | 12 | 1.293 | -0.211 |
| 2021 | 12 | 0.855 | 0.005 |
| 2022 | 12 | -2.048 | 0.508 |
| 2023 | 12 | 1.557 | -0.096 |
| 2024 | 12 | 0.333 | 0.505 |
Across 61 increasing readings, the dollar strengthened in 31 observations. Across 58 decreasing readings, it strengthened in 32. Unchanged readings sit outside both groups. The difference between the two counts should be evaluated against the number in each group. A group with more observations can have more dollar gains even when its fraction of gains is smaller.
For scale, the median dollar move across all months was 0.057%. The middle half of dollar changes ran approximately from -0.707% to 1.279%. These are changes between average quotes. They omit the sequence of prices inside each observation period. They cannot tell you whether a protective stop or profit target would have traded first, or whether either was attainable at the displayed average.
Six historical cases you can reconstruct
The cases below were selected by a fixed rule: extremes of the indicator change, a reading near the median, a small absolute change, an opposing move where available and a recent comparison. When the same date qualifies twice, the next eligible observation fills the slot. This selection deliberately includes unusual periods. Use the full annual table above for perspective before generalizing from a case.
Largest increase | April 2015
Separate single-family units from multifamily units before describing a building boom. A few large foundations can change a monthly count substantially.
To understand this extreme, compare March 2015 with April 2015. The US housing starts reading went from 960.0 to 1,190.0 thousands of units, annual rate. The signed change by the method used here was 23.958 percent. Check the size of the starting level before describing the move. Across the full sample, 119 changes were lower and 0 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.26180 to 1.23370 Canadian dollars per US dollar. That is a -2.227% change in the quote and a -2.227% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. An unusually large reading can be an observation to investigate without being the basis of a repeatable trade.
Carry that interpretation forward one observation as a check, without moving the endpoints. In May 2015 the indicator change was -9.328 percent and the dollar change was -1.305%. The dollar continued in the same direction on this strictly defined comparison. The extra period is included to test persistence. It cannot justify a fill at the earlier average or establish which event moved the market.
Largest decline | April 2020
During a decline check both the uncertainty interval and prior authorized projects. A drop in breaking ground differs from households cancelling home purchase contracts.
For the opposite tail, compare March 2020 with April 2020. The US housing starts reading went from 1,266.0 to 936.0 thousands of units, annual rate. The signed change by the method used here was -26.066 percent. Check the signed difference and the denominator before describing the move. Across the full sample, 0 changes were lower and 119 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.39600 to 1.40480 Canadian dollars per US dollar. That is a 0.630% change in the quote and a 0.630% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. A decline of the same number of points from a much higher level would have a different percentage meaning.
Carry that interpretation forward one observation as a check, without moving the endpoints. In May 2020 the indicator change was 11.004 percent and the dollar change was -0.541%. The dollar did not continue in the same direction on this strictly defined comparison. The extra period is included to test persistence. It cannot justify a fill at the earlier average or establish which event moved the market.
Near the sample median | January 2021
Use the more ordinary case to convert the annualized pace back to an adjusted monthly equivalent. Label that operation so nobody mistakes it for the raw monthly count.
For a more ordinary comparison, compare December 2020 with January 2021. The US housing starts reading went from 1,614.0 to 1,624.0 thousands of units, annual rate. The signed change by the method used here was 0.620 percent. Check how close this result was to the sample midpoint before describing the move. Across the full sample, 60 changes were lower and 59 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.28090 to 1.27250 Canadian dollars per US dollar. That is a -0.656% change in the quote and a -0.656% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. A well-chosen ordinary month is a counterweight to a persuasive account based only on exceptional headlines.
Carry that interpretation forward one observation as a check, without moving the endpoints. In February 2021 the indicator change was -13.916 percent and the dollar change was -0.228%. The dollar continued in the same direction on this strictly defined comparison. The extra period is included to test persistence. It cannot justify a fill at the earlier average or establish which event moved the market.
A quiet reading | November 2020
When total starts barely move, inspect regions with opposing changes. A national average can conceal disruptions from weather and regional funding conditions.
For the small-change case, compare October 2020 with November 2020. The US housing starts reading went from 1,543.0 to 1,543.0 thousands of units, annual rate. The signed change by the method used here was 0.000 percent. Check the precision of the published inputs before describing the move. Across the full sample, 58 changes were lower and 61 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.32180 to 1.30730 Canadian dollars per US dollar. That is a -1.097% change in the quote and a -1.097% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. Near zero, rounding can decide which sign appears on a one-decimal calendar even when the dollar move is economically noticeable.
Carry that interpretation forward one observation as a check, without moving the endpoints. In December 2020 the indicator change was 4.601 percent and the dollar change was -2.019%. The dollar continued in the same direction on this strictly defined comparison. The extra period is included to test persistence. It cannot justify a fill at the earlier average or establish which event moved the market.
Opposite directions | March 2020
A Canadian housing release can shift USD/CAD separately. Distinguish a Canadian-dollar move from evidence about how the US construction pipeline is responding.
To challenge a simple directional interpretation, compare February 2020 with March 2020. The US housing starts reading went from 1,549.0 to 1,266.0 thousands of units, annual rate. The signed change by the method used here was -18.270 percent. Check which currency is actually appreciating before describing the move. Across the full sample, 2 changes were lower and 117 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.32860 to 1.39600 Canadian dollars per US dollar. That is a 5.073% change in the quote and a 5.073% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. A counterexample rules out a one-direction rule while leaving open several possible economic explanations.
Carry that interpretation forward one observation as a check, without moving the endpoints. In April 2020 the indicator change was -26.066 percent and the dollar change was 0.630%. The dollar continued in the same direction on this strictly defined comparison. The extra period is included to test persistence. It cannot justify a fill at the earlier average or establish which event moved the market.
The last observation | December 2024
Save the units of every series. Comparing building counts with dwelling-unit counts is especially misleading in a large multifamily development.
At the edge of the chosen sample, compare November 2024 with December 2024. The US housing starts reading went from 1,300.0 to 1,494.0 thousands of units, annual rate. The signed change by the method used here was 14.923 percent. Check what later revisions may have changed before describing the move. Across the full sample, 113 changes were lower and 6 were higher. This rank is descriptive and was available only after the whole sample existed.
Over those same reference periods, USD/CAD moved from 1.39730 to 1.42470 Canadian dollars per US dollar. That is a 1.961% change in the quote and a 1.961% change in the value of USD against the other currency. The dollar is the base currency, so the two percentages match. A subsequent observation would change the sample statistics without changing what a trader could have known on the earlier date.
This is the last comparison in the defined study window. Leave the following observation blank when reproducing the table. Filling it with an assumed continuation would transform a descriptive observation into a forecast. A reader extending the sample should save another dated copy of both series, append the same frequency and check whether revisions changed the period used here.
How to reproduce and challenge the result
The construction stage and reporting unit
Represent one large apartment block as one project and then as its full number of housing units. Compare each with ten detached homes. The exercise explains how regional unit starts can jump without ten times as many building sites becoming active.
Study starts and permits as related but separate stages. Their difference in any month does not identify individual unstarted houses. The historical starts series is in thousands of units at a seasonally adjusted annual rate.
Build the first row by hand before copying a formula down a spreadsheet. Your source document should identify start or permit, region, single-family or multifamily and annualized rate. In a second column record the observation that would complicate the argument: many more units begin while the number of new projects barely changes. The workbook includes periods with divergent signs so you can test a claim on its least convenient examples. A difference between two averages describes something narrower than a high, low or executable closing quote.
An explicit assumption worth testing is this: “Units authorized minus units started this month equals the total backlog.” Treat that statement as a candidate error. Confidence intervals around national monthly estimates can be wide. Preserve the release’s own uncertainty language before calling a small change decisive. Keep a second copy of the sheet with the assumption corrected, then trace the first column where the comparison changes. A conclusion that depends on this choice needs the assumption in its caption. A result that survives still needs an out-of-sample test before it can support an execution rule.
Use the time zone named America/New_York when reconstructing a US release calendar. Eastern Standard Time and Eastern Daylight Time have different offsets from UTC. The government series date generally identifies an observation period or effective date, while a trading log needs the time information became public. Preserve the date visible on the original release, the timestamp of the quote, and your actual decision time. This keeps a familiar local clock from introducing a one-hour or one-month look-ahead error.
Sources and calculation notes
US Census housing construction definitions. US Census building permits definitions.
Historical inputs: HOUST series and units · EXCAUS exchange-rate history. Data checked on November 14, 2026. The study ends in December 2024 to use completed periods and a common end date across the guides.
Frequently asked questions
Is a permit the same as a building loan?
It is permission from the relevant authority. Financing and other requirements can remain outstanding.
Why can starts exceed permits in a month?
Starts can use permits issued earlier, and the series have differences in timing and coverage.
Do completions give the same trading signal?
They describe another stage of construction. A project completing today may have started under very different financing conditions.
Can I divide the annualized level by twelve?
Yes for an equivalent adjusted monthly pace. Use the unadjusted table for the actual number linked to that month.
Does one strong region establish a national boom?
Check regional weights, multifamily concentration and sampling uncertainty. A national figure can mask substantial regional differences.
For the next practice session, review the news-release filter, liquidity sweeps, market structure shifts, and hindsight-free replay. Use the same dated chart and account currency across the four exercises so you can compare your notes.
