Skip to content

Alpha Dive · Long engine

Sector attribution, measured — not one number pretending to be a portfolio total

The Long engine publishes sector-level alpha in two sleeves, USD and EUR & non-USD, each against its own sector benchmark. It does not publish a portfolio-level aggregate, because no single benchmark or sector weighting has been approved for the book as a whole — see the card below rather than a headline figure.

02 · Coverage

What this sleeve actually observed

Recomputed on every sleeve switch, directly from the source rows — never held over from the other sleeve.

Data coverage
36 months
Aug 2023 – Jul 2026
Sectors positive
5 / 7
Jul 2026 · 2 unobserved
Best sector, latest month
+12.3%
Consumer Discretionary
Worst sector, latest month
−4.5%
Health Care
Missing / error cells
5 / 324
2% of this sleeve, never shown as zero

03 · Portfolio-level alpha

The number this page will not print

Every other section below is a sector- or security-level diagnostic. None of them combine into this.

Methodology pending

Official aggregate alpha

The workbook carries sector-level attribution but has no approved single benchmark and no approved sector-weighting scheme for the Long portfolio as a whole. Without one, no aggregate Long alpha figure exists — and this page will not invent one to fill the gap.

Specifically, none of the following stands in for it: a simple average of security performance, a simple average of sector alphas, a product of sector deltas, or a single benchmark such as the S&P 500 or the EURO STOXX 50 applied across the whole sleeve. Each would smuggle in a weighting or benchmarking choice nobody has approved. Every figure below this card is a sector- or security-level diagnostic — never a portfolio total.

04 · Sector × month

Every sector, every observed month

Click a cell to filter the security contributors table below to that month. Missing readings are hatched, not zero-coloured — scroll horizontally on a narrow screen; the sector column stays put.

SectorAug '23Sep '23Oct '23Nov '23Dec '23Jan '24Feb '24Mar '24Apr '24May '24Jun '24Jul '24Aug '24Sep '24Oct '24Nov '24Dec '24Jan '25Feb '25Mar '25Apr '25May '25Jun '25Jul '25Aug '25Sep '25Oct '25Nov '25Dec '25Jan '26Feb '26Mar '26Apr '26May '26Jun '26Jul '26
Communication Services
Consumer Discretionary
Consumer Staples
Energy
Health Care
Industrials
Information Technology
Materials
Utilities
PositiveNegativeMissing / error — never shown as zero

Coverage is not symmetric between sleeves: 5 of 324 USD cells are unobserved, against 66 of 324 in the EUR & non-USD sleeve. The gaps below are real reporting gaps in a sparser sleeve, not an artefact of this chart.

05 · Latest month

This sleeve’s most recent reading, sector by sector

Diverging around zero. Sectors not observed this month are left off the chart, never drawn as a flat bar at zero.

Jul 2026

Not observed in Jul 2026: Consumer Staples, Utilities — rendered as elsewhere on this page, never as zero, and left out of this chart entirely rather than drawn as a flat bar.

06 · Compounded ranking

Sectors ranked by compounded alpha — with their coverage

A sector compounded over 32 of 36 months and one compounded over all 36 are not the same kind of number. Coverage sits next to every reading for that reason, not in a footnote.

RankSectorCompounded alphaCoverageAvg monthlyPositive months
1Utilities+174.4%
32/36 mo
+3.36%68.8%
2Consumer Staples+33.2%
35/36 mo
+0.91%65.7%
3Energy+26.6%
36/36 mo
+0.72%58.3%
4Communication Services+21.7%
36/36 mo
+0.63%61.1%
5Information Technology+7.8%
36/36 mo
+0.24%52.8%
6Consumer Discretionary+5.9%
36/36 mo
+0.24%50.0%
7Industrials+5.4%
36/36 mo
+0.23%44.4%
8Materials−1.1%
36/36 mo
+0.07%50.0%
9Health Care−15.0%
36/36 mo
−0.42%44.4%

Sector-level diagnostic only; not an official aggregate portfolio alpha. The largest reading here — Utilities at +174.4%, compounded over 32 of 36 months — is that kind of diagnostic: a real compounding effect within one sector's own attribution, not a portfolio return and not comparable to a benchmark-relative total.

07 · Security contributors

Top and bottom holdings, ranked by contribution

Contribution is weight × month-to-date performance, not the raw return — a small position with a large return moved the portfolio less than a large position with a modest one. Filtered by the month picked in the heatmap above, or chosen directly below.

Latest month
Top contributorsWeightPerf. MTDContribution
MSFT US EquityInformation Technology2.8%+20.9%+0.59%
AAPL US EquityInformation Technology2.8%+17.5%+0.49%
VLO US EquityEnergy1.7%+20.2%+0.35%
ADBE US EquityInformation Technology1.4%+22.3%+0.32%
GRMN US EquityConsumer Discretionary1.0%+28.0%+0.28%
ROST US EquityConsumer Discretionary1.4%+18.4%+0.25%
ADSK US EquityInformation Technology1.1%+19.7%+0.21%
COP US EquityEnergy1.6%+12.3%+0.20%
CRM US EquityInformation Technology1.4%+14.1%+0.19%
ROP US EquityInformation Technology1.1%+15.3%+0.16%
Bottom contributorsWeightPerf. MTDContribution
KLAC US EquityInformation Technology1.5%−27.5%−0.42%
AMAT US EquityInformation Technology1.6%−20.0%−0.32%
LRCX US EquityInformation Technology1.3%−21.5%−0.28%
FIX US EquityIndustrials1.4%−8.4%−0.12%
NVT US EquityIndustrials1.0%−11.0%−0.11%
ELV US EquityHealth Care1.2%−4.8%−0.06%
EW US EquityHealth Care1.2%−4.3%−0.05%
MGM US EquityConsumer Discretionary0.7%−7.2%−0.05%
APH US EquityInformation Technology1.8%−2.4%−0.04%
META US EquityCommunication Services1.9%−2.0%−0.04%

56 holdings observed in Jul 2026, USD sleeve. Ranked by contribution (weight × month-to-date performance), not by return.

08 · Methodology & data quality

How to read every number on this page

What sector_alpha measures

Each sector-level reading is the sleeve’s weighted portfolio return for that sector less that sector’s own benchmark return for the month — a per-sector excess return, not a position-level one. It is read directly from the workbook’s own monthly summary sheet (see each row’s source_sheet / source_cell for the exact cell).

Null means missing or an Excel error, never zero

A blank cell or an Excel #N/A / #DIV/0! becomes null in the source JSON. This page renders every null as an em dash and never as 0% — a month the model did not observe is not the same statement as a month in which it matched its benchmark exactly, and interpolating between neighbours would fabricate an observation that never happened.

Two files, one window

The sector-level file (Aug 2023 Jul 2026, 36 months) has no calendar gap. The security-level file covers the same 36 months exactly — this is checked directly against the sector file on every load rather than assumed, because an earlier delivery had the security file skip months the sector file did not. The two files describe the same engine at different levels of detail, not the same window twice: the heatmap and sector ranking below only ever use the sector-level window, and the security contributors table’s month selector is limited to the months the sector file actually has, so a click there always lines up with a real heatmap column.

The EUR & non-USD sleeve is far sparser

Missing cells are not evenly spread across the two sleeves — the KPI row and the heatmap legend both show the current split, and it is worth repeating here: the non-USD bucket carries roughly 13.2x the missing-cell rate of the USD bucket. Treat any shape you see in the EUR sleeve’s heatmap or ranking table with that in mind before reading it as a signal.

One duplicate source row

The security-level file contains 1 duplicate (date, region, ticker) combination 2025-12-01 · USD · PODD US Equity. Both rows are kept and both can appear in the contributors table; the page keys each row by its own source sheet and row number so neither is silently dropped or double-counted into one.

No aggregate Long alpha, on this page or anywhere else

See the “Official aggregate alpha” card above. It is not a placeholder waiting for a number — the methodology genuinely does not exist yet, and this page does not compute a simple average of security performance, a simple average of sector alphas, a product of sector deltas, or a single benchmark comparison to stand in for it. Every other figure here is a sector- or security-level diagnostic.