MP29 hockeystick candidate

MP29: All-index Hockeystick Candidate

Status: sensitivity case in the current jmMSE analysis. MP29 preserves the earlier +20 annual-increase specification and has not been adopted. This page is part of SC14-MSE-2026-RC1 and uses the common 500-draw refinement.

1 Identity

Component	Current definition
Name in the `jmMSE` analysis	`tun29`, labelled MP29
Stock/index indicator	`cpues.ind`
Indices	`Chile_AcousN`, `Chile_CPUE`, `Peru_Artis`, `Peru_Ind`, `Offshore_CPUE`
Reference years	2019-2023
Recent-data window	3 years
How the indices are combined	Mean (`combine = "mean"`, `metric = "mean"`)
Harvest control rule	Hockeystick (`hockeystick.hcr`); multi-stock testing uses an equivalent corrected function
Output	Catch
Catch allocation	Split among fisheries using their average catch proportions over the last five years (`split.is`)

2 Harvest control rule parameters

Parameter	Value
Minimum catch, `min`	270
Limit metric, `lim`	0.1
Trigger metric	2.125
Target catch	2000
Initial catch for constraints	1385
Maximum annual decrease multiplier, `dlow`	0.85
Maximum annual increase multiplier, `dupp`	1.20

The trigger is the value obtained during tuning and reported in the July presentation. Before formal use it should be copied from the approved CMP record, not recalculated by repeating the tuning exercise.

3 Rule

Let $I_t$ be the annual mean of the standardized index information calculated by the specified method. Before applying the annual change limit, the catch recommendation $C_t^*$ is

\[ C_t^* = \begin{cases} 270, & I_t \le 0.1,\\ 270 + (I_t-0.1)\dfrac{2000-270}{2.125-0.1}, & 0.1 < I_t < 2.125,\\ 2000, & I_t \ge 2.125. \end{cases} \]

The annual change rule then prevents the recommendation from falling below 85% or rising above 120% of the previous recommendation. For the first application year, the current calculation uses 1385 as the previous value.

4 What the current analysis shows

The current analysis retains MP29 and MP32 as annual-limit sensitivities to focal MP43 and MP44 across the reference and alternative operating models. MP29’s realized mean dynamic Kobe-green probability was 0.6098 over 2041–2050. Final adoption still requires approved annual data rules and a formal decision.

Primary sources: model_tune.R, model_robustness.R, and the July WUR presentation.

5 Annual application requirements still to resolve

authoritative provider and annual cutoff for each index;
precise index standardization performed by cpues.ind and the software version used;
treatment of missing or revised indices;
catch units and rounding;
approved catch-allocation update rule; and
exceptional-circumstances review and decision process.

--- title: "MP29 hockeystick candidate" --- {{< include ../_includes/sprfmo-banner.qmd >}} <div class="paper-title">MP29: All-index Hockeystick Candidate</div> ::: {.status-note} **Status:** sensitivity case in the current `jmMSE` analysis. MP29 preserves the earlier +20 annual-increase specification and has not been adopted. This page is part of `SC14-MSE-2026-RC1` and uses the common 500-draw refinement. ::: ## Identity | Component | Current definition | |---|---| | Name in the `jmMSE` analysis | `tun29`, labelled MP29 | | Stock/index indicator | `cpues.ind` | | Indices | `Chile_AcousN`, `Chile_CPUE`, `Peru_Artis`, `Peru_Ind`, `Offshore_CPUE` | | Reference years | 2019-2023 | | Recent-data window | 3 years | | How the indices are combined | Mean (`combine = "mean"`, `metric = "mean"`) | | Harvest control rule | Hockeystick (`hockeystick.hcr`); multi-stock testing uses an equivalent corrected function | | Output | Catch | | Catch allocation | Split among fisheries using their average catch proportions over the last five years (`split.is`) | ## Harvest control rule parameters | Parameter | Value | |---|---:| | Minimum catch, `min` | 270 | | Limit metric, `lim` | 0.1 | | Trigger metric | 2.125 | | Target catch | 2000 | | Initial catch for constraints | 1385 | | Maximum annual decrease multiplier, `dlow` | 0.85 | | Maximum annual increase multiplier, `dupp` | 1.20 | The trigger is the value obtained during tuning and reported in the July presentation. Before formal use it should be copied from the approved CMP record, not recalculated by repeating the tuning exercise. ## Rule Let $I_t$ be the annual mean of the standardized index information calculated by the specified method. Before applying the annual change limit, the catch recommendation $C_t^*$ is $$ C_t^* = \begin{cases} 270, & I_t \le 0.1,\\ 270 + (I_t-0.1)\dfrac{2000-270}{2.125-0.1}, & 0.1 < I_t < 2.125,\\ 2000, & I_t \ge 2.125. \end{cases} $$ The annual change rule then prevents the recommendation from falling below 85% or rising above 120% of the previous recommendation. For the first application year, the current calculation uses 1385 as the previous value. ## What the current analysis shows The current analysis retains MP29 and MP32 as annual-limit sensitivities to focal MP43 and MP44 across the reference and alternative operating models. MP29's realized mean dynamic Kobe-green probability was 0.6098 over 2041--2050. Final adoption still requires approved annual data rules and a formal decision. Primary sources: [`model_tune.R`](https://github.com/SPRFMO/jmMSE/blob/main/model_tune.R), [`model_robustness.R`](https://github.com/SPRFMO/jmMSE/blob/main/model_robustness.R), and the [July WUR presentation](../evidence/presentation_20260725.pdf). ## Annual application requirements still to resolve - authoritative provider and annual cutoff for each index; - precise index standardization performed by `cpues.ind` and the software version used; - treatment of missing or revised indices; - catch units and rounding; - approved catch-allocation update rule; and - exceptional-circumstances review and decision process.