Headline Position (EOD)

It answers what a book's or a portfolio's headline Greeks were at the close, on the market data that the day is booked against. The decision it feeds is the sign-off: the market risk function accepts the day's risk position, or it sends the desk back for a correction before the numbers stand.

The report exists beside its intraday sibling because the two answer different questions. The intraday definition says what risk looks like now, on an indicative body, and a later run replaces it. This definition says what the risk was at the close, on the authorised body, and no later run of the intraday definition replaces it. A reader who wants a figure to quote or to hedge against reads the intraday one. A reader who wants the figure that stands reads this one.

Cycle eod. Frequency scheduled and on-demand. The definition is seeded with the expression 0 18 * * 1-5, which fires it at 18:00 London time on weekdays, and a person may ask for one at any time. It is not a member of the ten-report Market Risk EOD set, so nothing orders it after the market data cut.

The concurrency policy is skip: a firing that arrives while a run of the same definition is in flight is dropped, because a second close report of the same day answers nothing the first one did not.

• Trades selected by scope. The scope shape is book or portfolio, and the seeded definition reads the entity's Global Portfolio.
• Market data for the close. The run selects the end of day cut, and it values the portfolio at the weekday before the run date.
• The risk report configuration the definition binds, Valuation and sensitivities (EOD) - Global Portfolio. It states the market data type eod and the base currency of the entity.
• The pricing engine the definition binds, EOD Official, which is the set of engines whose numbers are booked.

The run produces the engine's own output archive, which holds the report's CSV artefacts, and it records where that archive landed against the run instance. It also records the cut the run selected, with the cut's identity, label, kind and instant.

The archive is the only shape the figures take today. The rows a reader would query are a later phase, designed on Report post-processing and not built.

The set is the configuration's. The measures below are the ones the business catalogue states this report carries; their definitions are on Risk reporting and are not repeated here.

The engine aggregates by book, by portfolio, by currency pair, by Greek and by unit hedge. The aggregation is the producing engine's, not the reader's.

The run binds two analytics, so the archive carries three files a reader uses.

nvp.csv, written by the npv analytic:

sensitivity.csv, written by the sensitivity analytic, one row for each trade and risk factor:

scenario.csv, written by the same analytic, holds the value of each trade in each shifted market. Its columns are TradeId, Factor, Up/Down, Base NPV, ShiftSize_1, ShiftSize_2, Scenario NPV, Difference and Base Currency. A reader uses it to check a Delta figure against the two values it came from.

The npv analytic states the base currency and the file it writes. The sensitivity analytic states the two documents it reads and the three thresholds the run uses. The fragment is trimmed from external/ore/examples/MarketRisk/Input/ore_sensi.xml.

<Analytic type="npv">
  <Parameter name="active">Y</Parameter>
  <Parameter name="baseCurrency">GBP</Parameter>
  <Parameter name="outputFileName">npv.csv</Parameter>
</Analytic>
<Analytic type="sensitivity">
  <Parameter name="active">Y</Parameter>
  <Parameter name="marketConfigFile">simulation.xml</Parameter>
  <Parameter name="sensitivityConfigFile">sensitivity.xml</Parameter>
  <Parameter name="scenarioOutputFile">scenario.csv</Parameter>
  <Parameter name="sensitivityOutputFile">sensitivity.csv</Parameter>
  <Parameter name="outputSensitivityThreshold">0.000001</Parameter>
  <Parameter name="recalibrateModels">Y</Parameter>
  <Parameter name="parSensitivity">Y</Parameter>
</Analytic>

The run document that carries this fragment is generated, not stored. The service builds it from the risk report configuration, the analytics and the engine settings the definition binds. The run settings row states the market data type eod, so the document's asofDate is the weekday before the run date.

A definition of this report states the rows below. The values are the seeded ones. The mechanism is on ore parameter.

A person reads this report once a day, after the desk has closed.

1. Open the run of Headline Position (EOD) for the entity and the day. Read the run instance first. It names the cut it selected, and the cut's label and instant say which body the figures came from.
2. Take npv.csv and select the rows whose NettingSet is the book under review. Sum NPV(Base) over those rows. That total is the present value of the book at the close, and it is the figure the P&L measures are taken against.
3. Compare that total with the same total from the previous day's run of the same definition. The difference is the daily P&L. Compare with the month end and the year end totals for the MTD and YTD figures.
4. Take sensitivity.csv and group by Factor_1. Sum Delta within each factor. Read the result as money per shift of that factor, and note that the shift size is stated in ShiftSize_1 and is already inside the number.
5. Sort the factor groups by absolute total. The largest groups are the risks the sign-off is really about, and they say which deeper report a reader opens next: a curve ladder, a vol ladder or a spot ladder.
6. Where a figure looks wrong, take the matching row from scenario.csv and check Difference against the Delta of the same trade and factor. The two agree to the printed precision when the run is sound.
7. Sign off, or send the book back. The run instance holds the figures that were signed off, and a later run of the intraday definition never replaces it.

The report states what its definition and its configuration state. Four limits follow, and a reader must know them.

• The cut is reproducible and the body behind it is not. The run records the cut and its instant, and the market data body that the cut fixes is not materialised. So a rerun tomorrow selects a different cut and returns different figures. The record says which body was signed off, and it cannot hand the body back.
• The dataset holds curve quotes only, and no volatility market data. A volatility factor therefore has nothing to shift, and no volatility sensitivity can be computed until the store holds that data. This is a fact about the data and not about the report.
• The result store is not built, so the figures live in the archive. Nothing queries them, and a reader opens the CSVs by hand.
• The seeded definition runs on EOD Official engines. A bump report on analytic sensitivities ideally binds an engine set whose numbers are built for bumping; the ACME dataset carries such a set for London alone, named Risk Sensitivities. This definition does not bind it.

No substitute is defined for this report, so a test run of it executes every phase: it gathers trades and market data and assembles its own bundle. The report is exercised by the full-pipeline proof, not by a smoke test. The cut resolution it relies on is exercised by Light the intraday reports on the LIVE cut, which proves the same chain for the live selection.

The machinery that proves a run without gathering anything is the Sample bundle smoke test.

• Reports, the inventory this page is listed in.
• Headline Position, the intraday sibling, on the live cut.
• Risk reporting, the classification taxonomy and the measures catalogue.
• The cut is the body a run prices from, what a cut holds and what it does not.
• Report execution pipeline, the phases every run goes through.
• Cuts, the structure note for the cut this report reads.
• Light the intraday reports on the LIVE cut, the task that seeded the cut selections.

• Report families — the structure note that orders this cluster, and where to read this page in it.
• Open Source Risk Engine, the valuation engine whose sensitivity analytic produces the figures.