A tactical review method and system for game ball possession analysis and multivariate evaluation

Abstract

The present application relates to a kind of competition ball possession analysis and multivariate evaluation tactical review method and system. Including the following steps data acquisition, competition stage mark, micro-sequence segmentation, feature extraction, tactical value evaluation, review report generation. The module between each step of the present application, by competition stage mark and micro-sequence segmentation, the continuous ball possession process is decomposed into quantifiable tactical unit, the tactical value of each micro-sequence is calculated by multidimensional feature fusion, realizes the quantitative evaluation of ball possession process tactical value, outputs tactical value quantitative index, generates targeted training suggestion and tactical adjustment scheme based on diagnostic result, these suggestions can be directly converted into training plan and tactical arrangement.

Description

Technical Field

[0001] This invention relates to the field of sports data analysis technology, and in particular to a tactical review method and system for ball control analysis and multivariate evaluation in a match. Background Technology

[0002] At present, tactical analysis of football matches is an important means to improve a team's competitive level. Currently, the main methods of football match analysis include: (1) Macroeconomic analysis method based on statistical data This method provides an overall assessment of a match by statistically analyzing macro-level indicators such as possession, number of shots, and pass completion rate. However, this method is relatively coarse-grained and cannot delve into the tactical details of possession, making it difficult to identify key tactical turning points.

[0003] (2) Video-based manual analysis method This method relies on coaches and analysts manually annotating tactical events and player performances by watching game videos. This method is highly subjective, inefficient, and makes it difficult to quantify the tactical value of ball possession.

[0004] (3) Machine learning-based competition result prediction method This method uses historical match data to train machine learning models to predict match outcomes. However, it is primarily used for pre-match predictions and, based on statistical data rather than real-time match events, cannot be used for post-match tactical analysis.

[0005] (4) Deep learning-based competition situation assessment method Deep learning models are used to assess the game situation. However, this method mainly focuses on real-time situation assessment and outputs indicators such as goal probability, lacking quantitative analysis of the tactical value of ball possession and failing to provide actionable review suggestions.

[0006] In summary, existing football match analysis technologies have the following technical problems: (1) The tactical value of ball control cannot be quantified. Current technology can only measure macro-level indicators such as possession rate and number of passes, but it cannot assess the tactical value of a single possession session. For example, possession during backfield passing and possession during threatening attacks have the same statistical weight in current technology, but their actual tactical value differs greatly.

[0007] (2) Key tactical turning points are difficult to identify Current technology lacks the ability to segment and analyze the intentions behind ball possession, making it difficult to identify key turning points (such as transitions from defense to offense, initiation of an attack, and threatening attacks). This makes it difficult for coaches to quickly pinpoint crucial moments in a match during post-match analysis.

[0008] (3) Lack of actionable review suggestions Current technologies primarily output statistical data or probabilistic indicators, lacking analysis and recommendations tailored to specific tactical scenarios. Coaches struggle to translate these outputs into concrete training plans and tactical adjustments. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a tactical review method and system for ball control analysis and multivariate evaluation. By quantifying the tactical value of the ball control process, it distinguishes the tactical contributions of different ball control scenarios; identifies key tactical turning points and locates key moments in the game; and generates actionable review suggestions to guide training plans and tactical adjustments.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A tactical review method for analyzing ball possession and evaluating multiple variables in a match includes: S1, Data Acquisition: Acquire match event data and identify ball possession event sequences. The match event data includes passing events, shooting events, interception events, foul events, etc. S2, Match Phase Labeling: Perform match phase enhancement processing on the ball possession event sequence, and label the ball possession into multiple phases based on the spatial position and tactical intent of the ball possession: backcourt organization phase, advancement phase, and final attack phase; generate phase-aware feature representations based on the phase labels; S3, Micro-sequence segmentation: Based on the ball control intention transfer point and stage boundary, the ball control event sequence is segmented into several micro-sequences; S4, Feature Extraction: Extract tactical features for each micro-sequence, including spatial features, temporal features, event features, and phase features; S5, Tactical Value Assessment: Based on the aforementioned tactical characteristics, assess the tactical value of each micro-sequence, including offensive threat assessment, tactical execution assessment, and tactical value quantification; S6, Post-mortem Report Generation: Based on the aforementioned tactical value quantification indicators, generate a tactical post-mortem report, including key micro-sequence identification, tactical problem diagnosis, and improvement suggestions.

[0011] A system for a tactical review method for ball possession analysis and multivariate evaluation as described above includes: The data acquisition module is used to acquire match event data and identify ball control event sequences; The game phase annotation module is used to perform game phase enhancement processing on the ball control event sequence, including: annotating ball control into multiple phases according to the spatial position and tactical intention of ball control, and generating phase-aware feature representations based on the phase annotations; The micro-sequence segmentation module is used to segment a ball control event sequence into several micro-sequences based on the ball control intention transfer point or stage boundary; The feature extraction module is used to extract tactical features from each micro-sequence; A tactical value assessment module is used to assess the tactical value of each micro-sequence based on the tactical characteristics. The debriefing report generation module is used to generate a tactical debriefing report based on the tactical value.

[0012] Furthermore, the competition phase annotation module includes: The phase division unit is used to mark each ball possession event as the backcourt organization phase, the advancement phase, or the final attack phase based on the spatial coordinates and tactical intentions of the ball possession event. The stage-aware feature generation unit is used to fuse stage annotations with event features to generate stage-aware feature representations.

[0013] Furthermore, the stage division unit adopts the following stage division rules: When the spatial area is the backfield and the tactical intention is organization, it is marked as the backfield organization phase; When the spatial area is the backfield and the tactical intention is to advance, or when the spatial area is the midfield, it is marked as the advancing phase; When the spatial area is the frontcourt and the tactical intention is to attack, it is marked as the final attack phase.

[0014] Furthermore, the micro-sequence segmentation module includes: The intention transfer point determination unit is used to determine the intention transfer point based on the angle of change in direction of continuous passing events. The stage boundary determination unit is used to determine the stage boundary based on whether the stage label of adjacent events changes. The event combination determination unit is used to determine the segmentation trigger conditions based on a specific event sequence pattern.

[0015] Furthermore, the intention transfer point determination unit determines the intention transfer point by: when the angle of change of the passing direction is greater than or equal to a preset angle threshold, or when the ratio of the lengths of adjacent passing vectors is greater than a preset ratio threshold, it is determined to be an intention transfer point.

[0016] Furthermore, the feature extraction module includes: The spatial feature extraction unit is used to extract the penetration rate of the three attacking zones, the attack ratio on the flanks / centrals, and the number of touches in key areas. The time feature extraction unit is used to extract the duration of ball possession, the time of passing decisions, and the speed of offensive organization. The event feature extraction unit is used to extract pass success rate, number of breakthroughs, shooting-related actions, and expected goal value. The phase feature extraction unit is used to extract phase advancement success rate, number of threat events within a phase, and phase transition efficiency.

[0017] Furthermore, the tactical value assessment module includes: The offensive threat index calculation unit is used to calculate the offensive threat index by taking into account spatial position, defensive pressure and shooting probability. The tactical execution score calculation unit is used to evaluate the degree of matching between actual execution and tactical objectives and to calculate the tactical execution score. The value quantification unit is used to integrate the offensive threat index and tactical execution score to generate a micro-sequence of tactical value quantification indicators.

[0018] Furthermore, the attack threat index calculation unit calculates the attack threat index using the following formula: T_attack=w_1×T_space+w_2×T_defense+w_3×T_shot; Where T_space is the spatial threat component based on distance to the goal and shooting angle, T_defense is the defensive threat component based on defensive pressure, T_shot is the shooting probability component based on expected goal value, and w_1, w_2, and w_3 are weighting coefficients.

[0019] Furthermore, the debriefing report generation module includes: The key micro-sequence identification unit is used to identify micro-sequences with tactical value higher than a threshold and mark them as key tactical nodes; The tactical problem diagnosis unit is used to analyze micro-sequences whose tactical value is lower than expected and to diagnose the types of tactical execution problems. An improvement suggestion generation unit is used to generate training suggestions, tactical adjustment suggestions, or personal technical suggestions based on diagnostic results; The tactical problem diagnosis unit classifies tactical execution problems according to the following rules: When the passing success rate is below 0.7, it is diagnosed as a passing error problem; When the speed of the offensive organization is below the speed threshold, it is diagnosed as a slow advance problem; When the threat creation score is below 0.5, it is diagnosed as a threat creation deficiency problem; When the success rate of a phase advancement is below 0.6, it is diagnosed as a phase transition failure. When the directional efficiency is below 0.8, it is diagnosed as an execution deviation problem.

[0020] Compared with the prior art, the advantages of the present invention are as follows: (1) The tactical value of ball control can be quantified. By annotating the game phases and segmenting micro-sequences, the continuous ball control process is decomposed into quantifiable tactical units. The tactical value of each micro-sequence is calculated through multi-dimensional feature fusion, enabling a quantitative assessment of the tactical value of the ball control process.

[0021] (2) Key tactical turning points can be identified By automatically identifying points of ball possession intention transfer and stage boundaries, the system can automatically mark key tactical turning points in a match. This greatly improves the efficiency of post-match analysis and helps coaches quickly pinpoint crucial moments in the game.

[0022] (3) The debriefing suggestions are actionable. The system not only outputs quantitative indicators of tactical value, but also generates targeted training suggestions and tactical adjustment plans based on the diagnostic results. These suggestions can be directly translated into training plans and tactical deployments, making them highly practical. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0024] The specific embodiments of the present invention will be described below: A tactical review method for analyzing ball possession and evaluating multiple variables in a match includes: S1, Data Acquisition: Acquire match event data and identify ball possession event sequences. The match event data includes passing events, shooting events, interception events, foul events, etc. S2, Match Phase Labeling: Perform match phase enhancement processing on the ball possession event sequence, and label the ball possession into multiple phases based on the spatial position and tactical intent of the ball possession: backcourt organization phase, advancement phase, and final attack phase; generate phase-aware feature representations based on the phase labels; S3, Micro-sequence segmentation: Based on the ball control intention transfer point and stage boundary, the ball control event sequence is segmented into several micro-sequences; S4, Feature Extraction: Extract tactical features for each micro-sequence, including spatial features, temporal features, event features, and phase features; S5, Tactical Value Assessment: Based on the aforementioned tactical characteristics, assess the tactical value of each micro-sequence, including offensive threat assessment, tactical execution assessment, and tactical value quantification; S6, Post-mortem Report Generation: Based on the aforementioned tactical value quantification indicators, generate a tactical post-mortem report, including key micro-sequence identification, tactical problem diagnosis, and improvement suggestions.

[0025] Description of the working principle of this invention: A system for a tactical review method for ball possession analysis and multivariate evaluation as described above includes: The data acquisition module is used to acquire match event data and identify ball control event sequences; The game phase annotation module is used to perform game phase enhancement processing on the ball control event sequence, including: annotating ball control into multiple phases according to the spatial position and tactical intention of ball control, and generating phase-aware feature representations based on the phase annotations; The stage division function is as follows: For the i-th event E_i in the ball control event sequence, its stage label S_i is determined by the following function: S_i=f_stage(P_i,Z_i,T_i); in: P_i=(x_i,y_i) represents the spatial coordinates of event E_i, x_i∈[0,100] represents the percentage of the court length, and y_i∈[0,100] represents the percentage of the court width; Z_i∈{backfield, midfield, frontfield} is a spatial region; T_i∈{organization, advancement, attack} represents tactical intent; Phase division rules: S_i= { During the backcourt organization phase, when Z_i = backcourt and T_i = organization During the advancement phase, when (Z_i = backfield and T_i = advancement) or Z_i = midfield In the final offensive phase, when Z_i = forward and T_i = attacking... } Spatial region determination function: Z_i=f_zone(x_i)= { In the backcourt, when 0 ≤ x_i < 33.3 In the midfield, when 33.3 ≤ x_i < 66.7 In the frontcourt, when 66.7 ≤ x_i ≤ 100 } Tactical Intent Determination Function: T_i=f_tactic(E_i,E_{i-1},context); Tactical intent is determined by a combination of the following factors: Pass direction change angle: θ_i=arctan((y_i-y_{i-1}) / (x_i-x_{i-1})); Advancement speed: v_i=√((x_i-x_{i-1})²+(y_i-y_{i-1})²) / Δt_i; Opponent defensive pressure: P_defense=f_defense(opponent position distribution); Tactical intention determination rules: T_i= { organization, when v_i<v_threshold and P_defense<P_threshold advancement, when v_i≥v_threshold and θ_i<θ_threshold attack, when P_defense≥P_threshold or (Z_i=front court and shooting intention) } wherein the threshold parameters are: v_threshold=5m / s (advancement speed threshold); θ_threshold=60° (direction change threshold); P_threshold=0.5 (defensive pressure threshold); Phase-aware feature representation: For a ball possession sequence including n events, the phase-aware feature representation is: F_stage=[S_1,S_2,...,S_n]⊕[E_1,E_2,...,E_n]; wherein: S_i∈{1,2,3} is the phase code (1=backfield organization, 2=advancement, 3=final attack); E_i is an event feature vector; ⊕ is a feature concatenation operation; Phase transition matrix: M_transition= [p_11,p_12,p_13], [p_21,p_22,p_23], [p_31,p_32,p_33] wherein p_ij is the probability of transitioning from phase i to phase j, which is obtained through statistics of historical data.

[0026] ​​The micro-sequence segmentation module is used to segment a ball control event sequence into several micro-sequences based on the ball control intention transfer point or stage boundary; The formula for determining the intention transfer point is as follows: For two consecutive passing events E_i and E_{i+1}, the angle of change in the passing direction is: Δθ_i=arccos( (v_i·v_{i+1}) / (||v_i||×||v_{i+1}||) ) in: v_i=(x_i-x_{i-1},y_i-y_{i-1}) is the vector of the i-th pass; v_{i+1}=(x_{i+1}-x_i,y_{i+1}-y_i) is the vector of the (i+1)th pass; Intent transfer point determination: Intent_Transfer_i= { True, when |Δθ_i|≥θ_threshold or ||v_{i+1}|| / ||v_i||>r_threshold False, otherwise } in: θ_threshold = 60° (direction change threshold); r_threshold=2.0 (propulsion speed ratio threshold); Phase boundary determination formula: For two consecutive events E_i and E_{i+1}, the stage boundary is determined as follows: Phase_Boundary_i= { True, when S_i ≠ S_{i+1} False, otherwise } Where S_i and S_{i+1} are the stage labels for events E_i and E_{i+1}, respectively.

[0027] Event combination rule formula: Determining a specific event sequence: Event_Combination_i= { True, when (E_i.type="cross" and E_{i+1}.type="shot") or (E_i.type="through pass" and E_{i+1}.type="shot") or (E_i.type="breakthrough" and E_{i+1}.type="pass") False, otherwise } Micro-sequence segmentation algorithm: For a ball control sequence containing n events, the micro-sequence segmentation algorithm is as follows: Algorithm: MicroSequence_Segmentation; Input: A sequence of ball control events E=[E_1,E_2,...,E_n]; Output: A set of micro-sequences MS = {MS_1, MS_2, ..., MS_m}; 1. Initialization: start=1, k=1; 2. For i=1 to n-1, execute: a. If Intent_Transfer_i=True, or Phase_Boundary_i=True, or Event_Combination_i=True: -MS_k=[E_start,E_{start+1},...,E_i] -start=i+1 -k=k+1 3. MS_k=[E_start,...,E_n] / / The last micro-sequence 4. Return MS={MS_1,MS_2,...,MS_k} Microsequence feature representation: For the k-th micro-sequence MS_k, its characteristic is represented as: MS_k=(E_{start_k},E_{start_k+1},...,E_{end_k},F_{seq_k}); Where F_{seq_k} is the sequence feature vector of the micro-sequence: F_{seq_k}=[N_pass,T_duration,D_progress,C_transition,V_avg]; Formulas for calculating each component: Number of passes: N_pass = end_k - start_k; Duration: T_duration=t_{end_k}-t_{start_k}; Where t_i is the time when event E_i occurs.

[0028] Propulsion distance: D_progress=√((x_{end_k}-x_{start_k})²+(y_{end_k}-y_{start_k})²); Phase transition type: C_transition= { 1. When S_{start_k}=1 and S_{end_k}=2 / / Backstage → Advance 2. When S_{start_k}=2 and S_{end_k}=3 / / Advance → Final Attack 3. When S_{start_k}=1 and S_{end_k}=3 / / Backcourt → Final Attack (Fast Counterattack) 0, otherwise / / no stage transition } Average propulsion speed: V_avg=D_progress / T_duration; The feature extraction module is used to extract tactical features from each micro-sequence; Used to extract tactical features for each micro-sequence, including: Spatial characteristics: penetration rate in the three attacking zones, attack ratio on the wings / central channels, number of touches in key areas; Time characteristics: duration of ball possession, time for passing decisions, and speed of offensive organization; Event characteristics: pass completion rate, number of dribbles, and shooting-related actions; Phase characteristics: phase progress success rate, number of threat events within the phase, and phase transition efficiency.

[0029] Spatial characteristics: Penetration rate in the three attack zones: P_penetration=N_final_zone / N_total; in: N_final_zone is the number of events in the microsequence that occur in the first three zones (x>66.7); N_total is the total number of events in the micro-sequence; Offensive ratio on the wings / middle: R_wing = N_wing / N_total; R_center=N_center / N_total=1-R_wing; in: N_wing is the number of events that occur in the edge region (y<20 or y>80); N_center is the number of events that occurred in the middle lane region (20≤y≤80); Number of touches in key areas: N_key_area=ΣI(E_i∈Key_Area); in: Key_Area is the critical area (the restricted area and its edge, where x>80 and 15). <y<85) I(·) is an indicator function that takes the value 1 when event E_i occurs in the critical region, and 0 otherwise. Time characteristics: Duration of possession: T_possession=t_end-t_start Passing decision time: T_decision=(1 / N_pass)×Σ(t_{i+1}-t_i); Where N_pass represents the number of passes in the micro-sequence.

[0030] Offensive organization speed: V_organization=D_progress / T_possession; Event characteristics: Passing accuracy: R_pass_success=N_success / N_attempt; in: N_success represents the number of successful passes; N_attempt represents the total number of pass attempts; Number of breakthroughs: N_breakthrough=ΣI(E_i.type="breakthrough" or (E_i.type="through and penetrate the defense")); Shooting-related actions: A_shot=(N_shot,xG,Shot_Quality); in: N_shot represents the number of shots on goal; xG represents the expected goal value; Shot_Quality is the shot quality rating; Expected Goals (xG) Calculation: xG=P(Goal|Distance, Angle, Body Part, Assist Type, Defensive Pressure) Simplified formula: xG=1 / (1+exp(-(β_0+β_1×d+β_2×θ+β_3×P_defense))); in: d is the distance (in meters) from the shooting point to the goal. θ is the shooting angle (degrees); P_defense is the defensive pressure index; β_0,β_1,β_2,β_3 are model parameters; Stage characteristics: Phase progress success rate: P_progress=N_success_progress / N_attempt_progress in: N_success_progress represents the number of times a player has successfully progressed to the next stage or scored a goal. N_attempt_progress represents the total number of attempts to advance; Number of threat events within the phase: N_threat=ΣI(E_i∈Threat_Events) Threat_Events includes: Shoot; Key pass (leading to a shot); Break through the defenses; A threatening touch of the ball inside the penalty area; Phase transition efficiency: η_transition=N_successful_transition / T_transition in: N_successful_transition represents the number of successful transitions; T_transition is the time taken for the stage transition; A tactical value assessment module, used to assess the tactical value of each micro-sequence based on the tactical characteristics, includes: Offensive threat assessment: Based on spatial location, defensive pressure, and shooting probability, an offensive threat index is calculated; Tactical execution evaluation: Calculate the tactical execution score based on the degree of alignment between actual execution and tactical objectives; Value quantification: By integrating the threat index and execution score, a tactical value quantification index for micro-sequences is generated.

[0031] Offensive threat index calculation: The offensive threat index T_attack takes into account spatial positioning, defensive pressure, and shooting probability. T_attack=w_1×T_space+w_2×T_defense+w_3×T_shot; in: w_1+w_2+w_3=1, the weighting coefficients are adjusted according to the tactical system; T_space represents the space threat component; T_defense represents the defensive pressure component; T_shot is the shot probability component; Spatial threat component T_space: T_space=exp(-α×d_goal / 100)×(sin(θ_shot / 2))^β in: d_goal is the distance to the goal (in meters); θ_shot is the shooting angle (in radians); α = 2.0, distance attenuation coefficient; β=1.5, angular sensitivity coefficient; The defensive pressure component T_defense: T_defense=1 / (1+exp(γ×(P_defense-P_threshold))); in: P_defense is the defensive pressure index; P_threshold=0.5, the defensive pressure threshold; γ = 4.0, pressure sensitivity coefficient; Calculation of the defensive pressure index P_defense: P_defense=(1 / 3)×(N_defender_near+D_defense_avg+V_defense); in: N_defender_near=min(N_defender_within_5m / 3,1) is the normalized value of the number of close defenders; D_defense_avg = d_avg / 10 is the normalized value of the average defensive distance (d_avg is in meters); V_defense = v_defense_max / 8 is the normalized value of the maximum speed of the defending player (v_defense_max is in m / s); Shot probability component T_shot: T_shot = xG / xG_max; in: xG is the expected goal value (using the xG calculation formula described above); xG_max=0.8, the maximum expected number of goals (used for normalization); Tactical execution score calculation: The tactical execution score (E_execution) assesses the degree of alignment between actual execution and tactical objectives. E_execution=w_pass×E_pass+w_progress×E_progress+w_threat×E_threat; in: w_pass+w_progress+w_threat=1; E_pass is the pass execution score; E_progress represents the progress score; E_threat creates a score for threats; Pass execution rating E_pass: E_pass=R_pass_success×Q_pass_avg; in: R_pass_success = N_pass_success / N_pass_attempt represents the pass success rate; Q_pass_avg=(1 / N_pass)×ΣQ_pass_i is the average pass quality; Pass quality Q_pass_i calculation: Q_pass_i=w_1×Q_pressure+w_2×Q_progression+w_3×Q_risk; in: Q_pressure = 1 / (1 + P_defense_before) represents the pressure relief mass; Q_progression = Δx_pass / 30 is the mass of the advance (Δx_pass is the distance advanced towards the goal); Q_risk = 1 - P_loss represents the quality of risk control (P_loss is the probability of passing error). Progress score E_progress: E_progress=(D_progress / D_target)×η_direction; in: D_progress represents the actual distance advanced; D_target is the target advance distance (based on tactical objectives); η_direction=cos(θ_actual-θ_target) is the directional efficiency (θ is the direction angle); Threat Creation Rating E_threat: E_threat=N_threat_created / N_threat_expected; in: N_threat_created represents the actual number of threats created; N_threat_expected is the number of expected threat creations (based on tactical objectives); Tactical value quantification formula: The tactical value T_value^k of the micro-sequence k is calculated by fusing the offensive threat index and the tactical execution score: T_value^k=T_attack^k×E_execution^k×w_stage^k; in: T_attack^k is the attack threat index of the micro-sequence k; E_execution^k is the tactical execution score for micro-sequence k; w_stage^k is the stage weight, which is determined based on the stage in which the micro-sequence is located; Stage weight w_stage: w_stage= { 0.5, when S = backcourt organization stage 1.0, when S = advancement stage 1.2, When S = final attack phase } Overall tactical value of ball possession: T_possession=Σ_{k=1}^{m}T_value^k; Where m is the number of micro-sequences during ball control.

[0032] Micro-sequence tactical value normalization: T_normalized^k=T_value^k / T_max; Where T_max=1.0 is the maximum tactical value (theoretical value).

[0033] The debriefing report generation module is used to generate a tactical debriefing report based on the tactical value, including: Key micro-sequence identification: Identify micro-sequences with tactical value exceeding a threshold and mark them as key tactical nodes; Tactical problem diagnosis: Analyze micro-sequences whose tactical value is lower than expected to diagnose tactical execution problems; Improvement suggestion generation: Based on the diagnostic results, generate targeted training suggestions and tactical adjustment plans.

[0034] Key micro-sequence identification: Key tactical node determination: Key_Sequence_k= { True, when T_value^k≥T_threshold and T_attack^k≥T_attack_threshold False, otherwise } in: T_threshold = 0.5 (tactical value threshold); T_attack_threshold=0.6 (Attack threat threshold); Important tactical event judgment: Important_Event_k= { True, when E_{end_k}.type∈{shoot, key pass, dribble} and T_value^k≥0.3 False, otherwise } Tactical problem diagnosis: Tactical value lower than expected: Underperformance_k= { True, when T_value^k <T_expected×0.7 False, otherwise } Where T_expected represents the expected tactical value based on the tactical objective.

[0035] Tactical execution problem classification: Problem_Type_k= { "Passing error", when R_pass_success < 0.7 "Slow progress", when V_organization <V_threshold "Insufficient threat creation" occurs when E_threat < 0.5 "Phase transition failed" when P_progress < 0.6 "Execution deviation", when η_direction < 0.8 } Improvement suggestions generated: Training suggestion generation: Training_Suggestion=f_train(Problem_Type_k,Severity_k); Severity level: Severity_k= { High, when T_value^k < 0.3 In the case where 0.3 ≤ T_value^k < 0.5 Low, when T_value^k≥0.5 } Tactical adjustment suggestions generated: Tactical_Adjustment=f_tactical(T_attack^k,E_execution^k,Context); Personal technical suggestion generation: Technical_Suggestion=f_technical(Player_ID,Skill_Deficiency); Among them, skill deficiency identification: Skill_Deficiency=argmin_{skill}Score_player_skill; Furthermore, the competition phase annotation module includes: The phase division unit is used to mark each ball possession event as the backcourt organization phase, the advancement phase, or the final attack phase based on the spatial coordinates and tactical intentions of the ball possession event. The stage-aware feature generation unit is used to fuse stage annotations with event features to generate stage-aware feature representations.

[0036] Furthermore, the stage division unit adopts the following stage division rules: When the spatial area is the backfield and the tactical intention is organization, it is marked as the backfield organization phase; When the spatial area is the backfield and the tactical intention is to advance, or when the spatial area is the midfield, it is marked as the advancing phase; When the spatial area is the frontcourt and the tactical intention is to attack, it is marked as the final attack phase.

[0037] Furthermore, the micro-sequence segmentation module includes: The intention transfer point determination unit is used to determine the intention transfer point based on the angle of change in direction of continuous passing events. The stage boundary determination unit is used to determine the stage boundary based on whether the stage label of adjacent events changes. The event combination determination unit is used to determine the segmentation trigger conditions based on a specific event sequence pattern.

[0038] Furthermore, the intention transfer point determination unit determines the intention transfer point by: when the angle of change of the passing direction is greater than or equal to a preset angle threshold, or when the ratio of the lengths of adjacent passing vectors is greater than a preset ratio threshold, it is determined to be an intention transfer point.

[0039] Furthermore, the feature extraction module includes: The spatial feature extraction unit is used to extract the penetration rate of the three attacking zones, the attack ratio on the flanks / centrals, and the number of touches in key areas. The time feature extraction unit is used to extract the duration of ball possession, the time of passing decisions, and the speed of offensive organization. The event feature extraction unit is used to extract pass success rate, number of breakthroughs, shooting-related actions, and expected goal value. The phase feature extraction unit is used to extract phase advancement success rate, number of threat events within a phase, and phase transition efficiency.

[0040] Furthermore, the tactical value assessment module includes: The offensive threat index calculation unit is used to calculate the offensive threat index by taking into account spatial position, defensive pressure and shooting probability. The tactical execution score calculation unit is used to evaluate the degree of matching between actual execution and tactical objectives and to calculate the tactical execution score. The value quantification unit is used to integrate the offensive threat index and tactical execution score to generate a micro-sequence of tactical value quantification indicators.

[0041] Furthermore, the attack threat index calculation unit calculates the attack threat index using the following formula: T_attack=w_1×T_space+w_2×T_defense+w_3×T_shot; Where T_space is the spatial threat component based on distance to the goal and shooting angle, T_defense is the defensive threat component based on defensive pressure, T_shot is the shooting probability component based on expected goal value, and w_1, w_2, and w_3 are weighting coefficients.

[0042] Furthermore, the debriefing report generation module includes: The key micro-sequence identification unit is used to identify micro-sequences with tactical value higher than a threshold and mark them as key tactical nodes; The tactical problem diagnosis unit is used to analyze micro-sequences whose tactical value is lower than expected and to diagnose the types of tactical execution problems. An improvement suggestion generation unit is used to generate training suggestions, tactical adjustment suggestions, or personal technical suggestions based on diagnostic results; The tactical problem diagnosis unit classifies tactical execution problems according to the following rules: When the passing success rate is below 0.7, it is diagnosed as a passing error problem; When the speed of the offensive organization is below the speed threshold, it is diagnosed as a slow advance problem; When the threat creation score is below 0.5, it is diagnosed as a threat creation deficiency problem; When the success rate of a phase advancement is below 0.6, it is diagnosed as a phase transition failure. When the directional efficiency is below 0.8, it is diagnosed as an execution deviation problem. Example

[0043] Taking an English Premier League match as an example, the specific workflow of the system of the present invention will be explained.

[0044] Match Background: Manchester City hosted Liverpool, with a final score of 2-1. This invention's system performs a post-match tactical review and analysis of this match.

[0045] Step 1: Data Collection The system obtains match event data through the API interface, including: All passing events (passing time, passing player, receiving player, starting coordinates of passing, ending coordinates of passing, passing result); Shooting event (shooting time, shooting player, shooting position, shooting result); Stealing incidents, fouls, etc.; Step 2: Marking the Competition Phase The system marks each possession stage. Let's take a Manchester City possession as an example: Possession begins: Goalkeeper kicks off (coordinates: 20, 50) → Backfield organization phase; First pass: Goalkeeper passes short to defender (coordinates: 30, 50) → Backfield organization phase; Second pass: The defender passes the ball across to the defensive midfielder (coordinates: 40, 45) → Advance phase (pass crosses the midfield line); Third pass: The defensive midfielder plays a through ball to the striker (coordinates: 80, 30) → final attack phase (entering the attacking third zone); 4th pass: Forward shoots → Final attacking phase; Possession ended: Shot saved; The system labels the sequence for this possession generation phase as: [backfield, backfield, advance, advance, final, final]; Step 3: Microsequence Segmentation Based on the point of ball possession intention transfer and the stage boundary, the system divides the above ball possession into 3 micro-sequences: Microsequence 1 (MS1): Backfield organization → Propulsion phase; Includes the following events: Goalkeeper kicks off → Defender passes across → Defensive midfielder receives the ball; Break-through trigger point: The ball crosses the midfield line and enters the advancing phase; Characteristics: 3 passes, 8 seconds of possession, 40 meters of advance; Micro-sequence 2 (MS2): Advancement phase → Final attack phase; Event included: through ball from defensive midfielder → forward receives the ball; Division trigger point: Entering the third attack zone, transitioning to the final attack phase; Characteristics: 2 passes, 3 seconds of possession, 35 meters of advance, through ball to break through the defense; Micro Sequence 3 (MS3): A shot during the final offensive phase; Includes the event: a striker's shot; Trigger point for segmentation: Shooting action; Characteristics: Shot location (18 meters from the goal, 30 degrees), shot result (saved). Step 4: Feature extraction; The system extracts tactical features for each micro-sequence: Microsequence 1 (MS1) characteristics: Spatial characteristics: 0% penetration rate in the three attack zones (no entry), 100% attack rate in the middle lane; Time characteristics: Ball possession duration 8 seconds, passing decision time 2.7 seconds / time; Event characteristics: 100% pass success rate (3 / 3), 0 dribbles; Phase characteristics: 100% success rate in phase advancement, 0 threat events during the phase; Microsequence 2 (MS2) characteristics: Spatial characteristics: 100% penetration rate in the three attacking zones, 100% attacking proportion through the middle, and 1 touch in key areas; Time characteristics: Ball possession duration is 3 seconds, and passing decision time is 1.5 seconds per pass; Event characteristics: 100% pass success rate (2 / 2), 1 breakthrough (through ball breaking through the defense); Phase characteristics: 100% success rate in phase advancement; 1 threat event during the phase. Microsequence 3 (MS3) characteristics: Spatial characteristics: The shooting position is 18 meters away from the goal, at an angle of 30 degrees, and the shooting area is the edge of the penalty area; Time characteristics: Ball control duration is 0.5 seconds; Event characteristics: 1 shot action, shot result (saved), expected goal value (xG) 0.12; Stage characteristics: The final stage is 100% complete, and it translates into a shot on goal; Step 5: Tactical value assessment; The system evaluates the tactical value of each micro-sequence based on tactical characteristics: Micro Sequence 1 (MS1) Tactical Value Assessment: Offensive threat assessment: Spatial position far from the goal (60 meters away), low defensive pressure, shooting probability 0% → Offensive threat index 0.15; Tactical execution assessment: Successfully transferred the ball from the backfield to the midfield, and successfully advanced the stage → Tactical execution score 0.75; Tactical value quantification: Offensive threat index 0.15 × tactical execution score 0.75 × stage weight 0.5 = 0.056; Microsequence 2 (MS2) Tactical Value Assessment: Offensive threat assessment: Entering the attacking third, a through ball breaks through the defense, and the probability of a shot is 30% → Offensive threat index 0.65; Tactical execution assessment: The through ball successfully penetrated the defense, giving the striker a shooting opportunity → Tactical execution score 0.85; Tactical value quantification: Offensive threat index 0.65 × tactical execution score 0.85 × stage weight 1.0 = 0.553; Micro Sequence 3 (MS3) Tactical Value Assessment: Offensive threat assessment: Shots from the edge of the penalty area, expected goal value 0.12 → offensive threat index 0.45; Tactical execution assessment: Shot completed, but saved → Tactical execution score 0.60; Tactical value quantification: Offensive threat index 0.45 × tactical execution score 0.60 × stage weight 1.2 = 0.324; Total tactical value of ball possession: 0.056 + 0.553 + 0.324 = 0.933; Step 6: Generate the review report The system generates tactical review reports based on tactical value quantification indicators: Key micro-sequence identification: Microsequence 2 (MS2) has a tactical value of 0.553, which is higher than the threshold of 0.5, and is therefore marked as a key tactical node. Micro-sequence 3 (MS3) has a tactical value of 0.324. Although it does not reach the critical node threshold, it is a shot micro-sequence and is marked as an important tactical event. Tactical problem diagnosis: The Micro Sequence 1 (MS1) has a tactical value of only 0.056, slow phase progression (from the backcourt to the midcourt in 8 seconds), and long passing decision time (2.7 seconds / pass). It is recommended to speed up the pace of ball distribution from the backcourt. The shot from Micro Sequence 3 (MS3) was saved due to a narrow shooting angle (30 degrees). It is recommended to seek a better shooting angle or choose to pass the ball. Improvement suggestions generated: Training suggestions: To address the issue of slow ball movement from the backfield, it is recommended to increase training for quick transitions, requiring players to make passing decisions within 2 seconds; Tactical adjustment suggestions: In view of the high threat of through balls, it is recommended to increase the use of such tactical combinations, and at the same time train the forwards to grasp the timing of their runs; Personal technical advice: Regarding the issue of choosing the shooting angle, it is recommended that forwards improve their observation skills before shooting and learn to adjust their body position to seek a better angle.

[0046] Accuracy of tactical value assessment: Accuracy_TV=N_correct / N_total; Where N_correct is the number of microsequences that are correctly assessed for tactical value, and N_total is the total number of microsequences.

[0047] Key turning point identification rate: Recall_KT=N_detected_key / N_actual_key; Where N_detected_key is the number of key inflection points identified by the system, and N_actual_key is the actual number of key inflection points.

[0048] The adoption rate of the review suggestions: Adoption_RA=N_adopted / N_generated; Where N_adopted is the number of suggestions adopted by the coach, and N_generated is the total number of suggestions generated.

[0049] Overall performance rating: Score_overall=w_1×Accuracy_TV+w_2×Recall_KT+w_3×Adoption_RA; Where w_1+w_2+w_3=1 is the weight coefficient.

[0050] The beneficial effects of this invention are as follows: (1) The tactical value of ball control cannot be quantified. Current technology can only measure macro-level indicators such as possession rate and number of passes, but it cannot assess the tactical value of a single possession session. For example, possession during backfield passing and possession during threatening attacks have the same statistical weight in current technology, but their actual tactical value differs greatly.

[0051] (2) Key tactical turning points are difficult to identify Current technology lacks the ability to segment and analyze the intentions behind ball possession, making it difficult to identify key turning points (such as transitions from defense to offense, initiation of an attack, and threatening attacks). This makes it difficult for coaches to quickly pinpoint crucial moments in a match during post-match analysis.

[0052] (3) Lack of actionable review suggestions Current technologies primarily output statistical data or probabilistic indicators, lacking analysis and recommendations tailored to specific tactical scenarios. Coaches struggle to translate these outputs into concrete training plans and tactical adjustments.

[0053] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A tactical review method for ball possession analysis and multivariate evaluation in a match, characterized by: include: S1, Data Acquisition: Acquire match event data and identify ball possession event sequences. The match event data includes passing events, shooting events, interception events, foul events, etc. S2, Match Phase Labeling: Perform match phase enhancement processing on the ball possession event sequence, and label the ball possession into multiple phases based on the spatial position and tactical intention of the ball possession: backcourt organization phase, advancement phase, and final attack phase; Based on the stage annotations, a stage-aware feature representation is generated; S3, Micro-sequence segmentation: Based on the ball control intention transfer point and stage boundary, the ball control event sequence is segmented into several micro-sequences; S4, Feature Extraction: Extract tactical features for each micro-sequence, including spatial features, temporal features, event features, and phase features; S5, Tactical Value Assessment: Based on the aforementioned tactical characteristics, assess the tactical value of each micro-sequence, including offensive threat assessment, tactical execution assessment, and tactical value quantification; S6, Post-mortem Report Generation: Based on the aforementioned tactical value quantification indicators, generate a tactical post-mortem report, including key micro-sequence identification, tactical problem diagnosis, and improvement suggestions.

2. The tactical review system for match ball control analysis and multivariate evaluation according to claim 1, characterized in that: A tactical review method for implementing the ball control analysis and multivariate evaluation method as described in claim 1 includes: The data acquisition module is used to acquire match event data and identify ball control event sequences; The game phase annotation module is used to perform game phase enhancement processing on the ball control event sequence, including: annotating ball control into multiple phases according to the spatial position and tactical intention of ball control, and generating phase-aware feature representations based on the phase annotations; The micro-sequence segmentation module is used to segment a ball control event sequence into several micro-sequences based on the ball control intention transfer point or stage boundary; The feature extraction module is used to extract tactical features from each micro-sequence; A tactical value assessment module is used to assess the tactical value of each micro-sequence based on the tactical characteristics. The debriefing report generation module is used to generate a tactical debriefing report based on the tactical value.

3. The tactical review system for match ball control analysis and multivariate evaluation according to claim 2, characterized in that: The competition phase annotation module includes: The phase division unit is used to mark each ball possession event as the backcourt organization phase, the advancement phase, or the final attack phase based on the spatial coordinates and tactical intentions of the ball possession event. The stage-aware feature generation unit is used to fuse stage annotations with event features to generate stage-aware feature representations.

4. The tactical review system for match ball control analysis and multivariate evaluation according to claim 3, characterized in that: The stage division unit adopts the following stage division rules: When the spatial area is the backfield and the tactical intention is organization, it is marked as the backfield organization phase; When the spatial area is the backfield and the tactical intention is to advance, or when the spatial area is the midfield, it is marked as the advancing phase; When the spatial area is the frontcourt and the tactical intention is to attack, it is marked as the final attack phase.

5. The tactical review system for match ball control analysis and multivariate evaluation according to claim 4, characterized in that: The micro-sequence segmentation module includes: The intention transfer point determination unit is used to determine the intention transfer point based on the angle of change in direction of continuous passing events. The stage boundary determination unit is used to determine the stage boundary based on whether the stage label of adjacent events changes. The event combination determination unit is used to determine the segmentation trigger conditions based on a specific event sequence pattern.

6. The tactical review system for match ball control analysis and multivariate evaluation according to claim 5, characterized in that: The intent transfer point determination unit determines intent transfer points as follows: when the angle of change of the passing direction is greater than or equal to a preset angle threshold, or when the ratio of the lengths of adjacent passing vectors is greater than a preset ratio threshold, it is determined to be an intent transfer point.

7. The tactical review system for match ball control analysis and multivariate evaluation according to claim 6, characterized in that: The feature extraction module includes: The spatial feature extraction unit is used to extract the penetration rate of the three attacking zones, the attack ratio on the flanks / centrals, and the number of touches in key areas. The time feature extraction unit is used to extract the duration of ball possession, the time of passing decisions, and the speed of offensive organization. The event feature extraction unit is used to extract pass success rate, number of breakthroughs, shooting-related actions, and expected goal value. The phase feature extraction unit is used to extract phase advancement success rate, number of threat events within a phase, and phase transition efficiency.

8. The tactical review system for match ball control analysis and multivariate evaluation according to claim 7, characterized in that: The tactical value assessment module includes: The offensive threat index calculation unit is used to calculate the offensive threat index by taking into account spatial position, defensive pressure and shooting probability. The tactical execution score calculation unit is used to evaluate the degree of matching between actual execution and tactical objectives and to calculate the tactical execution score. The value quantification unit is used to integrate the offensive threat index and tactical execution score to generate a micro-sequence of tactical value quantification indicators.

9. The tactical review system for match ball control analysis and multivariate evaluation according to claim 8, characterized in that: The attack threat index calculation unit calculates the attack threat index using the following formula: T_attack=w_1×T_space+w_2×T_defense+w_3×T_shot; Where T_space is the spatial threat component based on distance to the goal and shooting angle, T_defense is the defensive threat component based on defensive pressure, T_shot is the shooting probability component based on expected goal value, and w_1, w_2, and w_3 are weighting coefficients.

10. The tactical review system for match ball control analysis and multivariate evaluation according to claim 9, characterized in that: The debriefing report generation module includes: The key micro-sequence identification unit is used to identify micro-sequences with tactical value higher than a threshold and mark them as key tactical nodes; The tactical problem diagnosis unit is used to analyze micro-sequences whose tactical value is lower than expected and to diagnose the types of tactical execution problems. An improvement suggestion generation unit is used to generate training suggestions, tactical adjustment suggestions, or personal technical suggestions based on diagnostic results; The tactical problem diagnosis unit classifies tactical execution problems according to the following rules: When the passing success rate is below 0.7, it is diagnosed as a passing error problem; When the speed of the offensive organization is below the speed threshold, it is diagnosed as a slow advance problem; When the threat creation score is below 0.5, it is diagnosed as a threat creation deficiency problem; When the success rate of a phase advancement is below 0.6, it is diagnosed as a phase transition failure. When the directional efficiency is below 0.8, it is diagnosed as an execution deviation problem.