# Nine EPEF points for one day: age is at catch, not at processing

> Birds travel and the plant processes them the next day. EPEF divides by age, so that extra day costs nine points over a 39.4-day cycle. The error has a signature: exact conversion with a biased EPEF.

URL: https://raptia.co/en/blog/epef-and-the-day-the-bird-leaves-the-shed
Publicado: 2026-09-22
Actualizado: 2026-09-22

---


## How old is a bird on the day it is processed?

One day older than when it left the shed. The truck loads at night, the bird
reaches the plant before dawn and goes on line that morning. Between the date
the farm records and the date the plant records there is a day, sometimes two.

It looks like minor bookkeeping. It is not: EPEF **divides** by age.

## What does that day cost?

Over a 39.4-day cycle, using 40.4 instead drops the index by roughly 2.5%. On a
flock sitting at 360, that is **nine points**.

| Figure | With catch-day age | With processing date |
|---|---:|---:|
| Age used | 39.4 days | 40.4 days |
| Feed conversion | 1.654 | 1.654 |
| Live weight | 2.519 kg | 2.519 kg |
| Survival | 94.66% | 94.66% |
| **EPEF** | **366** | **357** |

Nine points is a gap that, in many operations, is enough to decide whether a
farm manager is rewarded or not.

## Why is it so hard to see?

Because the error does not touch feed conversion.

<dfn>Feed conversion ratio</dfn> is kilos of feed divided by kilos of live
weight produced. Age enters that division nowhere. So the flock shows an
impeccable conversion — the same one anybody would compute by hand — alongside
an EPEF that does not square with it.

<aside>

**The signature, written down so it gets recognised:** exact conversion with a
biased EPEF can only come from age or survival, because those are the only two
terms that enter one index and not the other. If mortality is also correct, age
is what is left. Nothing else needs auditing.

</aside>

## What if the flock left in two batches?

Then there are two ages, and the right question is not "which one do I use?" but
"which flock am I talking about?".

A thinned flock has birds that left on day 34 and birds that left on day 42. The
EPEF of the whole is neither batch's: it belongs to the complete flock, and the
age that matches it is the mean weighted by kilos — not the last truck's, not
the first's.

Taking the last truck's inflates age and sinks the index. Taking the first's
does the opposite. Both are defensible in a meeting and both are wrong.

## How is it fixed without touching historical data?

The missing figure is usually already there: the daily record for the thinning
day carries the birds removed. The removal date for that batch follows from it
without asking anyone and without depending on what the plant wrote down.

The rule worth writing down, because it gets forgotten:

- **EPEF age is the age on the day the bird leaves the shed.** Not the
  processing date, not the plant's docket, not the invoice.
- **With a staggered catch, it is the mean weighted by kilos** across batches.
- **If the removal date is not in the record, say so.** An index computed from
  the processing date because nothing else was available is an index that must
  be flagged as estimated, not one published alongside the rest.

## What else looks wrong for the same reason?

Any indicator with age in the denominator. **Average daily gain** — final weight
over days — is underestimated in exactly the same proportion. **Cost per shed
day** is spread over one day too many and comes out cheap. And next cycle's
projection, if calibrated against history, inherits the whole bias.

That is the uncomfortable trait of denominator errors: they do not produce an
odd number that draws attention. They produce an entire complex that appears to
perform slightly below what it actually does, consistently, for years. And
against a biased history, the farm always meets target.