Set-piece added value

wa_setpieces.core.value blends two things into one number per delivery: the xT added by the delivery itself, and – if it produced a shot, whether straight off the ball or via a second-phase loose ball – how good a chance that shot was. added_value = delivery_xt_added + shot_value.

The shot link comes from Opta’s own assist-chain qualifier (wa_setpieces.core.chains.link_set_piece_shots()), scoped to the shooting team’s own event stream – eventId is only unique within one team’s event stream in this feed (both teams number their events 1, 2, 3, … independently), so resolving it without that scoping can silently attribute a shot to the wrong team’s delivery.

from pathlib import Path

from wa_setpieces import load_matches
from wa_setpieces.core.value import set_piece_added_value, set_piece_value_summary
from wa_setpieces.viz.plots import plot_xt_added_bars
from wa_setpieces.core.xt import XTModel

try:
    _here = Path(__file__).resolve().parent
except NameError:
    _here = Path.cwd()
DATA = _here.parent / "tests" / "data" / "sample_match.json"

events = load_matches(DATA)
model = XTModel.fit(events, x_bins=8, y_bins=6)

Per-delivery breakdown for free kicks:

set_piece_added_value(events, "free_kick", model)

	matchId	eventId	contestantId	playerId	playerName	outcome	delivery_xt_added	shot_value	added_value	is_goal
0	sample_match	6	cxb4hqite921i36gwrezdts7c	ecaofku4htx8if4kk2teq1rrp	R. Adé	1	0.000100	0.0	0.000100	False
1	sample_match	67	f2yd0yzt0om6qhks9gbowu1d6	77t9kl34rhnesp7qz8uqdcga2	S. Callegari	1	-0.000630	0.0	-0.000630	False
2	sample_match	85	f2yd0yzt0om6qhks9gbowu1d6	9g05j8p6gyul0zlnx70rffkve	S. Mina	1	0.001337	0.0	0.001337	False
3	sample_match	87	cxb4hqite921i36gwrezdts7c	dll3x4ua7q15dcb7gbm61jso9	L. Segovia	1	0.000000	0.0	0.000000	False
4	sample_match	108	cxb4hqite921i36gwrezdts7c	cebzrizhi195acmaw8c4cku09	A. Tobar	1	0.002644	0.0	0.002644	False
5	sample_match	121	f2yd0yzt0om6qhks9gbowu1d6	7ww4nc2olcox6pyggzepjth0q	R. Silva	0	0.000000	0.0	0.000000	False
6	sample_match	141	f2yd0yzt0om6qhks9gbowu1d6	77t9kl34rhnesp7qz8uqdcga2	S. Callegari	0	0.000000	0.0	0.000000	False
7	sample_match	171	f2yd0yzt0om6qhks9gbowu1d6	9k61c9b8xjgh6jy6tziw11905	P. Velasco	0	0.000000	0.0	0.000000	False
8	sample_match	208	cxb4hqite921i36gwrezdts7c	dll3x4ua7q15dcb7gbm61jso9	L. Segovia	1	-0.001122	0.0	-0.001122	False
9	sample_match	220	cxb4hqite921i36gwrezdts7c	ecaofku4htx8if4kk2teq1rrp	R. Adé	1	0.000566	0.0	0.000566	False
10	sample_match	305	f2yd0yzt0om6qhks9gbowu1d6	c9htznfxd2gjo5s3klq2tvlqt	D. Armas	0	0.000000	0.0	0.000000	False
11	sample_match	307	cxb4hqite921i36gwrezdts7c	ecaofku4htx8if4kk2teq1rrp	R. Adé	1	-0.001222	0.0	-0.001222	False
12	sample_match	455	f2yd0yzt0om6qhks9gbowu1d6	9g05j8p6gyul0zlnx70rffkve	S. Mina	1	-0.000255	0.0	-0.000255	False
13	sample_match	488	f2yd0yzt0om6qhks9gbowu1d6	d6gphu89283pngbaenacu1cr9	R. Jaramillo	1	0.010445	0.0	0.010445	False
14	sample_match	473	cxb4hqite921i36gwrezdts7c	dll3x4ua7q15dcb7gbm61jso9	L. Segovia	0	0.000000	0.0	0.000000	False
15	sample_match	526	f2yd0yzt0om6qhks9gbowu1d6	czrsu9ep8tzghq0klucpfv6c5	B. Caicedo	1	-0.003492	0.0	-0.003492	False
16	sample_match	572	f2yd0yzt0om6qhks9gbowu1d6	4t4oc5mv6gujh932jz6scuulm	N. Molina	1	-0.000008	0.0	-0.000008	False
17	sample_match	543	cxb4hqite921i36gwrezdts7c	6w4sdzr4l7g9efzx6paxvadat	J. Quintero	1	0.009179	0.0	0.009179	False
18	sample_match	588	cxb4hqite921i36gwrezdts7c	d7d208zfvkuvjaba38rlcjxl	G. Valle	0	0.000000	0.0	0.000000	False
19	sample_match	595	cxb4hqite921i36gwrezdts7c	7dr2jx3u755avbchwnycsq0h5	K. Minda	1	-0.001591	0.0	-0.001591	False
20	sample_match	705	f2yd0yzt0om6qhks9gbowu1d6	d6gphu89283pngbaenacu1cr9	R. Jaramillo	1	-0.001397	0.0	-0.001397	False
21	sample_match	712	f2yd0yzt0om6qhks9gbowu1d6	9mrxht9j08y9w07w3877o2gwk	Ó. Quiñónez	1	-0.000712	0.0	-0.000712	False
22	sample_match	671	cxb4hqite921i36gwrezdts7c	3oyfdo7dnp4taew78zalpjp0q	G. Villamíl	1	-0.008039	0.0	-0.008039	False
23	sample_match	715	cxb4hqite921i36gwrezdts7c	d7d208zfvkuvjaba38rlcjxl	G. Valle	0	0.000000	0.0	0.000000	False
24	sample_match	844	f2yd0yzt0om6qhks9gbowu1d6	7ww4nc2olcox6pyggzepjth0q	R. Silva	1	0.000657	0.0	0.000657	False

Ranked, as a diverging bar chart (green = created value, red = cost it):

free_kick_value = set_piece_added_value(events, "free_kick", model)
fig, ax = plot_xt_added_bars(
    free_kick_value, value_col="added_value", title="Free-kick added value"
)

Rolled up per team:

set_piece_value_summary(events, "corner", model)

	contestantId	deliveries	total_added_value	avg_added_value	goals
0	cxb4hqite921i36gwrezdts7c	2	-0.0192	-0.0096	0
1	f2yd0yzt0om6qhks9gbowu1d6	7	0.0000	0.0000	0