Synergy evaluation, visualization, and investment determination support system in m & a

Abstract

To provide a visualization / determination method and device for quantifying the synergy of M & A, and for achieving quick and consistent Go / No-Go determination by integrating non-finance and finance. The method of claim 1, further comprising: SOLUTION: The method organizes the Business Performance Indicators (KPIs) in the role of Impact / Ease / Speed and normalizes to 0-5 with piecewise linear mapping to obtain an axis score S _ axis. In consideration of the principal axis, an integrated index S _ index is calculated by convex combination of the average of a plurality of axes and the principal axis score. S _ index> = theta is set as a non-financial gate using the determination threshold value theta estimated by learning, and DeltaNPV·IRR (DeltaNPV _ star corrected by a penalty coefficient as needed) obtained by DCF cooperation is integrated with the financial gate to output a final determination. History metadata such as cohort _ id, version information of Method / ParameterLibrary, source, acquisition time, and overwrite identifier are stamped on the output to ensure reproducibility and auditability.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] This invention relates to synergy evaluation technology in corporate mergers and acquisitions (M&A), and relates to the normalization and visualization of business evaluation criteria KPIs (Key Performance Indicators), investment gate determination and result output, as well as the data structure and version management that support these. [Background technology]

[0002] Traditionally, synergy assessment in M&A deals has relied on ad hoc trial balances and departmental interviews. KPI definitions, units, and metrics vary by assessor and deal, resulting in a lack of comparability across deals. Furthermore, the granularity of high-level synergy hypotheses (e.g., sales growth, cost reduction) during the initial stage of the process often differs from the raw data of individual KPIs observed after the M&A deal is completed, making it difficult to correlate the two. This has led to insufficient continuity and reproducibility in decision-making and performance monitoring throughout the entire process, from pre-DD (pre-due diligence: business feasibility assessment), DD (due diligence), contract signing (establishment of synergy clauses), and post-merger integration (PMI) after the deal is completed. Furthermore, synergy assessment practices often involve a mix of publicly available and internal information, and there is a lack of systems for systematically retaining and tracking the source, acquisition time, processing history, quality, and reliability of each piece of data, resulting in a lack of auditability and explainability. Regarding quantitative evaluation, although efforts to calculate the change in business value converted to present value (Delta NPV (Net Present Value)) and the Internal Rate of Return (IRR) using methods such as the Discount Cash Flow (DCF) method are widely used, general-purpose software / services that gate decisions based on uniform thresholds, including non-financial conditions such as regulatory compliance, dependencies, and data history, and that support and output the final investment decision in a machine-readable format, are not widely available, resulting in operational burdens and variations in judgments even among business company personnel and experts such as financial advisors (FAs). DISCLOSURE OF THE INVENTION [Problem to be solved by the invention]

[0003] Given the above background, the objective of this invention is to establish a unified framework that enables the following throughout the entire process, from pre-DD, DD, contract signing (formulation of synergy protection clauses), to post-deal PMI: (i) establish an index system for synergy evaluation that is comparable across industries, (ii) normalize KPIs to a common scale of 0 to 5 to reduce distortions in visualization, (iii) mechanically determine whether to invest using a unified gate that includes dissynergies and non-financial conditions, (iv) output the result of the determination (PASS or FAIL) to directly support decision-making, (v) ensure auditability by stamping data provenance and version information on the output, and (vi) enable safe and consistent replacement of parameters according to industry, region, size, etc. using cohort identifiers. [Means for solving the problem]

[0004] The present invention, with the configuration described in the claims, normalizes KPIs from 0 to 5 using a piecewise linear mapping (monotonic or unimodal), and visualizes them in a radar diagram (8-axis synergy radar diagram) with a fixed axis order and fixed radius of 0 to 5 (polar coordinate plot with the upper score limit fixed as the upper radius limit). The main axis is identified using impact / ease / speed weighting or a main flag, and investment feasibility is determined using a two-layer gate based on modified Delta NPV and IRR with dissynergy penalties applied, and the result of the determination (PASS or FAIL) is output. The output is stamped with provenance metadata and version information, and parameters are re-estimated and inherited based on cohort identifiers. [Effects of the Invention]

[0005] This invention improves the comparability, explainability, and auditability between deals, enabling consistent decision-making support from visualization to investment decisions. Furthermore, machine-readable output of decision results (PASS / FAIL) facilitates automatic integration with internal approval processes and deal management systems. Furthermore, the synergy evaluation index provided by this invention helps dealers (intermediaries) who do not directly manage M&A transactions, as well as investors and business companies considering investment target companies after M&A, to estimate the business growth impact resulting from synergies. This provides value indicators for companies and businesses in related industries, contributing to the advancement of deal discovery, portfolio selection, and market valuation. Additionally, parameter operation based on provenance metadata marking and cohort identifiers satisfies governance requirements and improves model maintainability, potentially reducing operational variability and workload. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a system diagram of this embodiment. [Figure 2] Figure 2 is a MONO_UP mapping image. [Figure 3] Figure 3 shows a MONO_DOWN mapping image. [Figure 4] Figure 4 shows an image of the BAND_TRI mapping. [Figure 5] Figure 5 is a diagram of the BAND_TRAP mapping. [Figure 6] Figure 6 is a Synergy Index chart for Example 1 (T-Mobile*Sprint). [Figure 7] FIG. 7 is a Synergy Index diagram for Example 2 (IBM*Red Hat). [Figure 8] Figure 8 is the Synergy Index for Example 3 (AB InBev*SABMiller). [Figure 9] Figure 9 is (1 / 2) of the parameter list. [Figure 10] Figure 10 is the parameter list (2 / 2). DETAILED DESCRIPTION OF THE INVENTION

[0007] The descriptions in this specification are preferred embodiments and examples, and the processing order, formula form, mapping type, normalization, gate determination, learning procedure, visualization, etc. can be changed as appropriate without departing from the spirit of the present invention. The system of this embodiment includes an input acquisition unit, an industry information management unit, an evaluation guideline selection unit, a learning module unit, a normalization unit, a score calculation unit, a visualization output unit, a penalty / gate determination unit, a DCF collaboration unit, and a report generation unit. Figure 1 shows a system diagram of this embodiment. Specific aspects of each configuration will be described in detail in turn. [Input acquisition section]

[0008] The headquarters accepts as input the KPI raw data (values, units, currency, evaluation base period) for the M&A target entity, as well as historical metadata such as its source, acquisition time, processing history, etc. After receiving the data, it performs period alignment, currency and unit conversion, seasonal adjustment, outlier treatment, and freshness (age) measurement, and maps it to identifiers in the KPI dictionary. As a result of the preprocessing, the post-preprocessing input value x_i (a numerical value with the unit, currency, and evaluation base period adjusted) for each KPI is determined and stored as a pair with the KPI dictionary ID (kpi_id).

[0009] The headquarters output is at least the following set: { kpi_id, x_i, unit, evaluation reference period, source, acquisition time, processing history, CI_i (0 to 100), CI calculation sub-element u_{i,j}}. Here, x_i is used as input to the [Normalization Section], and CI_i is used only for the weight correction g(CI_i) in the subsequent stage (see [Evaluation Guideline Selection Section]) (it is not used for calculating s_i itself).

[0010] The definition of the sub-elements that make up CI_i is as follows (the normalized scale from 0 to 1 given to each element indicates a larger contribution at the top): ·u_src: Source reliability (rating of official documents, audited reports, primary information, etc.) · u_age: Information freshness (elapsed time since acquisition, consistency with the evaluation reference period) ·u_comp: Consistency / Completeness (fulfillment of required KPIs, presence of missing / inconsistent) · u_cons: consistency (continuity within the series, outliers, consistency of revision history) · u_proc: Processing validity (applicability of units, currency conversion, seasonal adjustment, Winsor / Clamping, etc.) CI_i is calculated by normalizing each sub-element u_{i,j} (j in {src, age, comp, cons, proc}) from 0 to 1, then combining them with the allocation ratio lambda_j (SUM lambda_j=1) using the following formula: CI_i = round(100*SUM_j lambda_j*u_{i,j}). Parameters such as the allocation ratio lambda_j and the time constant T_age are stored in the Parameter Library (PL) for each cohort and updated through learning and re-estimation based on actual data or fine-tuning based on operational experience. Updates are recorded in an auditable manner and managed under version control of Method / PL. In other words, while the mathematics of CI_i are defined by the above equation, the inclusion criteria for each sub-element and optimization of the allocation ratio are determined through empirical adjustment and learning procedures to ensure system stability and evaluation reproducibility.

[0011] In addition, headquarters defines cohorts (industry, region, size, etc.) based on the Global Industry Classification Standard (GICS) classification to account for differences in KPI expression across industries and differences in the conversion rate at which KPIs are reflected in free cash flow (FCF). For each cohort, a dictionary is maintained that maps representative industry-specific KPIs to generic KPI names (e.g., customer acquisition / retention, unit price, conversion rate, utilization rate, processing capacity, inventory turnover, churn rate, etc.). Furthermore, the mapping type (MONO_UP / MONO_DOWN, BAND_TRI / BAND_TRAP) for each KPI and the control point selection procedure based on the aforementioned Theta (threshold candidate set) are managed separately for each cohort. Specifically, cohort_id is defined and named using a combination of industry, size, region, and business model (e.g., TELCO_US_2018, RETAIL_APAC_SMB), and DEFAULT is used if undetermined. This cohort is managed in the [Industry Information Management Section], and the settings are used in the subsequent [Normalization Section] and [Score Calculation Section].

[0012] Furthermore, headquarters will store various initial input values ​​used in other system components in PL, and will automatically update them through learning, or if a proxy or estimated value is used due to a lack of information, a Parameter Override ID (POID) will be assigned to the parameter to mark its history. The PL / Method version ID, POID, application period, reason for overwriting, etc. will be recorded in the audit log and marked on the output. The specified parameters described for each department, including the headquarters, are summarized in the parameter list shown in Figure 6. [Evaluation Mode Settings]

[0013] The input acquisition unit accepts at least the following two evaluation modes depending on the operator's purpose. The mode is propagated to all processing systems as a flag m in {HYP, DSC} and is reflected in the following mandatory KPI requirements, modulation coefficient g(CI) for CI correction, gate operation, and report notes. Here, g(CI) is a monotonically non-decreasing correction coefficient that maps data reliability CI in [0,100] to the range of (0,1] and can be given, for example, as a linear form g(CI) = 1 - eta + eta * (CI / 100) (eta in [0,1], default eta = 1). eta is a coefficient that adjusts the degree of suppression; at eta = 1 it is proportional to the reliability, and at eta = 0 it is a uniform weight. In the hypothesis testing mode (which emphasizes highly reliable data), described below, eta is increased (default 1), and in the exploration mode, eta is suppressed to reduce the impact of uncertain KPIs, among other adjustments. Parameter updates are performed using version control to maintain output consistency. g(CI) is used as a correction value for the KPI weight w_i, which will be explained in the "Evaluation Guideline Selection Section" below. As will also be explained later, it does not affect the KPI normalized value s_i or thresholds (L / T / H, etc.). (i) Hypothesis mode (m=HYP)

[0014] This mode allows the system operator to explicitly input their synergy hypotheses (e.g., increase in average revenue per user (ARPU), net subscriber additions (Net Adds), aggregate run rate, improved API compatibility, etc.) and related assumptions (ramp alpha_t, kappa / tau / omega / c, etc.). A set of essential KPIs is strictly required, and estimation of missing or insufficient information is minimized. If proxy or estimation is unavoidable, the POID, CI reduction (DeltaCI), etc. are inscribed in the history as a mark of arbitrary overwriting of information. In subsequent processing, the suppression effect of g(CI) is kept small, assuming that CI is sufficiently high, and the operation of theta prioritizes the learned theta* (provisional theta0 is used only if theta* does not exist for the cohort_id). Here, theta and theta* are thresholds for determining the business and financial value of the overall index S_index for synergies, which will be explained in the [Score Calculation Section] below.

[0015] Furthermore, the hypothesis verification mode also doubles as a mode in which KPI-level verification data from M&A and business integration projects that have already been implemented is input, and based on that factual information, provisional parameters are accumulated and updated into learned parameters. In other words, by inputting actual data, the system acquires learning material for theta* and g(CI), which contributes to improving the evaluation accuracy of subsequent projects. (ii) Discovery / Scouting mode (m=DSC)

[0016] This mode broadly explores potential synergy structures when combining two (or multiple) companies. Processing begins with minimal basic attributes (industry, size, region, business model, etc.) and readily available KPIs. Missing KPIs are provisionally synthesized using weak contribution (weight suppression using CI-linked g(CI)) or proxy / estimation. The report includes "recommendations for collecting missing KPIs," "potential focus areas," and "potential bottlenecks (regulations, dependencies, IT, etc.)." To avoid jumping to conclusions, the report also includes a fail-safe (theta = theta 0 operation, review required) as necessary. When transitioning to HYP, fulfillment of the missing KPI triggers recalculation, and the differences (S_index, focus areas, gate determination, Delta NPV, etc.) are output for comparison. Here, the primary axis is the output axis on the visualization diagram that corresponds to the synergy that is most important in light of the value hypothesis of the project, and the primary flag is a binary flag (0 / 1) assigned to it. Regardless of the mode, the primary flag is set to 1 for the primary synergy axis (0 for others). Details will be explained in [Visualization Output Section]. [Hypothesis Testing Mode Details]

[0017] This section describes the key points of learning collaboration in the hypothesis testing mode (m=HYP). (a) Receiving learning data: Receives the KPI measurement values ​​and evaluation period of completed projects, the label of success (Pass) / failure (Fail) of synergy creation, and cohort_id. The source, acquisition time, and processing history of the input are recorded as the default provenance metadata. (b) Updating theta*: Search for a threshold using a predefined performance index (Youden's J) for the received data and update theta* for each cohort. The procedure follows the traditional training / validation division, and details on training are described in the [Learning Module] section below. (c) Checking for changes in CI: If the distribution of CI_i has changed significantly compared to the most recent operational settings, it will be subject to review, and if necessary, a decision will be made to provisionally operate theta (theta = theta0) or to re-learn. (d) Record fixing: Records of the scope of application of the updated theta* and g(CI), the period and data used, the version information of the Method / PL, the time of implementation and the person in charge are kept and can be referenced from the report generation section.

[0018] (Note) This section describes the operation of learning collaboration, and the formulas for normalization, synthesis, and gate determination are the same as in other descriptions. (Note) In both modes, provenance (cohort_id, Method / PL version ID, POID, CI[theta*], generation time, hash, etc.) is stamped in a unified manner. Mode differences are limited to CI correction of w_i, required KPI requirements, theta operation policy, and report notes, and the formulas for normalization, composition, and DCF linkage are common. (Note) In this specification, "Method" refers to a set of procedural specifications for the evaluation process. Specific details are described below, but include at least the following: (i) the method for referencing the KPI mapping type and thresholds (interpretation of MONO_UP / DOWN, BAND_TRI / TRAP, and L / T / H / TL / TH), (ii) the use of I / E / S weights W and contributions q_i, the normalization procedure for w_i, and the application method for CI correction g(CI), (iii) the calculation formula for the integrated index S_index and the procedure for identifying the main axis, (iv) theta operation policy (trained theta* / provisional theta0) and the gate judgment procedure, and (v) the procedure for estimating theta* using the performance index (default: Youden's J) in the learning module. The Method is a procedure specification, and the numerical parameters themselves (threshold values, initial weight values, etc.) are stored in the PL. The Method is common across projects, while the PL differs depending on the cohort. Each output is stamped with the Method version ID and PL version ID to ensure reproducibility. [Industry Information Management Department]

[0019] The headquarters identifies the target cohort and determines the cohort_id based on the preprocessed data from the input acquisition unit and the company profile (industry, region, size, business model, etc.). Based on the cohort_id, it reads the Parameter Library (PL) and Method version information and provides recommended initial values ​​for the mapping type (map_type_i) for each KPI, outlier treatment policy, I / E / S criteria weights, penalty and gate related settings, and financial conversion coefficients (kappa, tau, omega, c, alpha_t, etc.) used in the DCF linkage unit. The output is {cohort_id, Method / PL version ID, KPI → mapping setting table, penalty and gate settings, financial parameters}. If an overwrite (POID assignment) occurs, the history of the overwrite is also handed over to subsequent processing.

[0020] Once the cohort_id is determined, the headquarters will read the mapping type (map_type_i) from the PL for each KPI belonging to that cohort. The mapping types are classified as follows: MONO_UP: Monotonically increasing type (e.g. market share, sales growth rate, API compatibility rate, etc.) MONO_DOWN: Monotonically decreasing type where the evaluation decreases as the value increases (e.g., cancellation rate, inventory days, cost rate, etc.) BAND_TRI: Triangular type that is optimal around the target value (e.g., personnel size, utilization rate, inventory level, etc.) BAND_TRAP: Trapezoidal shape that is optimal within a certain range (e.g., quality index, availability rate, delivery adherence rate, etc.)

[0021] The selection of the control points (L, T, H, TL, TH) used to normalize each KPI is based on the previously mentioned Theta (threshold candidate set). In other words, the learning module sweeps the candidate set Theta for the cohort and determines the control points based on evaluation criteria (reproducibility, discrimination performance, stability index, etc.). The Industry Information Management Department provides operational parameters for this learning procedure, such as the search range, granularity, priority, and default points for each cohort, and records and distributes the control points obtained as a result of learning, along with the version information of the Method / PL. As a result, each KPI stores its normalization settings in the dictionary in the form of "{kpi_id, map_type_i, control point selected on Theta (L, T, H, TL, TH)}" and is used in the piecewise linear mapping process in the subsequent [Normalization] stage. The mapping type is selected and the control points are determined based on the learning results of the cohort's past distribution, industry characteristics, and performance data, and are recorded in the audit log with metadata (source, update time, Method / PL version ID, POID, etc.).

[0022] To accommodate differences in the expression of industry-specific KPIs, cohort definitions are managed as combinations of industry, region, size, business model, etc., with reference to the GICS (Global Industry Classification Standard) classification. Each cohort maintains a dictionary of representative industry-specific KPIs and generic KPI names (e.g., customer acquisition / retention, unit price, conversion rate, utilization rate, processing capacity, inventory turnover, churn rate, etc.), ensuring consistency with the mapping type and Theta-based control point selection procedure. The cohort_id is named based on a combination of industry, size, region, business model, etc. (e.g., TELCO_US_2018, RETAIL_APAC_SMB), and DEFAULT is used if undetermined. The cohort determination results, as well as the map_type_i and control point settings based on them, are passed to subsequent sections (e.g., evaluation guideline selection section, normalization section, score calculation section) along with the Method / PL version ID. [Evaluation Guidelines Selection Department]

[0023] Based on the cohort_id and PL / Method version, the headquarters determines the standard weights W=(w_I, w_E, w_S) of I / E / S, which are the impact (scale), ease, and speed of each KPI used within the project, and obtains the contribution composition q_i=(q_iI, q_iE, q_iS) of each KPI from the dictionary. Next, the effective weight (non-normalized) u_i = w_I*q_iI + w_E*q_iE + w_S*q_iS is calculated, multiplied by the reliability correction g(CI_i) to obtain w_i′ = u_i*g(CI_i), and normalized within the synergy evaluation axis by w_i = w_i′ / SUM_k(w_k′). Here, SUM_i w_i=1, q_i>=0 and q_iI+q_iE+q_iS=1 (if not set, it is divided equally into 1 / 3). In addition, since freely changing the weighting of I / E / S for each synergy evaluation axis may result in the focus of the evaluation being dispersed, it is possible to uniquely fix one of I / E / S within a project (e.g., prioritize Ease) and perform the evaluation. The output is { W, q_i, g(CI_i), u_i, fixed indicator selection (I / E / S, unique within the project), theta operation policy (reference learned theta* / provisional theta0)} and passed to the normalization part. [Normalization part]

[0024] The headquarters accepts as input the preprocessed input values ​​x_i (units, currency, evaluation base period, etc. aligned) of each KPI received from the Input Acquisition Unit, and the mapping type (map_type_i) and control points (L, T, H, TL, TH) on Theta presented by the Industry Information Management Unit, and calculates a bounded score s_i (KPI normalized score) between 0 and 5 using a piecewise linear mapping. Values ​​outside the bounds are saturated with a clamp, and Winsorization is applied as necessary. The output is the normalized score sequence {s_i} for each KPI, and an audit record of the mapping type, threshold, and preprocessing policy used, which are handed over to the Score Calculation Unit. Specifically, the control points of a KPI's piecewise linear mapping use the standard terms L, T, H, TL, and TH. When setting initial values, quantiles (e.g., Q10 / Q50 / Q90) may be mapped to L / T / H / TL / TH, but text, diagrams, and output should be consistently written as L / T / H / TL / TH. And the mapping type (map_type_i) is as follows: Monotonically increasing piecewise linear mapping (MONO_UP): A mapping in which the score monotonically increases as the input value increases. Monotonically decreasing piecewise linear mapping (MONO_DOWN): A mapping in which the score monotonically decreases as the input value increases. · Unimodal triangular type (BAND_TRI): A mapping that is maximized at the target value T and is piecewise linear in L / T / H. Trapezoidal type (BAND_TRAP): A mapping that has a band that is full from TL to TH and is piecewise linearly defined by L / TL / TH / H. Note: The output {s_i} in this section is in KPI units, and is later aggregated by axis to obtain S_axis (8 axes: sales, cost, finance, human resources / organization, intellectual property / technology, regulations / licenses, market / competition, and IT / digital). Each axis is treated as a bounded score from 0 to 5, and is visualized as a radar diagram (8-axis synergy radar diagram) with a fixed axis order and fixed radius from 0 to 5 in the [Visualization Output] section described later. As a mapping, clamp(a,b,x) := a (when x < a) x (when a <= x <= b) b (when x > b) year, MONO_UP(L <H): s_i = clamp(0,5, 5*(x_i - L) / (H - L)) (x_i<=L→0, x_i>=H→5) MONO_DOWN(L <H): s_i = clamp(0,5, 5*(H - x_i) / (H - L)) (x_i<=L→5, x_i>=H→0) ·BAND_TRI(L < T < H): s_i = 0 (x_i <= L or x_i >= H) s_i = 5 * (x_i - L) / (T - L) (L < x_i <= T) s_i = 5 * (H - x_i) / (H - T) (T < x_i < H) ·BAND_TRAP(L < TL <= TH < H): s_i = 0 (x_i <= L or x_i >= H) s_i = 5 * (x_i - L) / (TL - L) (L < x_i <= TL) s_i = 5 (TL < x_i <= TH) s_i = 5 * (H - x_i) / (H - TH) (TH < x_i < H) However, for numerical stability, assume L < T < H, L < TL <= TH < H, and set it to satisfy (H - L) * (T - L) * (H - T) * (TL - L) * (H - TH) > 0. The mapping image diagrams of MONO_UP in Figure 2, MONO_DOWN in Figure 3, BAND_TR in Figure 4, and BAND_TRAP in Figure 5 are shown. [Score calculation unit]

[0025] This department receives as input the {s_i} from the normalization department and {W, q_i, g(CI_i), u_i, fixed index selection (any one of I / E / S, unique within the case), theta operation policy (refer to the learned theta* / provisional theta0)} from the evaluation index selection department. Taking the set of KPIs belonging to each synergy axis j as A_j, the axis score S_axis(j) = SUM_{i in A_j} (w_i * s_i) is calculated. Here, let W * q_i = w_I * q_iI + w_E * q_iE + w_S * q_iS, and the I / E / S weighted contribution C(j) = SUM_{i in A_j} { (W * q_i) * s_i} If a fixed index selection is specified (e.g., Ease), the contribution is limited to a* in {I,E,S}. C_{a*}(j) = SUM_{i in A_j} { (w_{a*}*q_{i,a*}) * s_i} Then, the axis with the largest C_{a*}(j) is identified as the key evaluation axis A_primary. If there is no fixed indicator selection, the axis with the largest C(j) is identified as A_primary. The integrated index S_index is S_index = (1-beta)*mean_A{ S_axis(A)} + beta*S_axis(A_primary) where mean_A{ S_axis(A)} is the arithmetic mean over the eight axes. The output is { S_axis (8 axes), A_primary, S_index, contribution Top-K (top K items of {i, (W*q_i)*s_i}), audit information of weight w_i} and is provided to the visualization output section and penalty*gate judgment section. [Learning Module]

[0026] The headquarters will accept data on past cases with labels indicating whether the synergy created in the M&A was successful (Pass) or not (Fail) (KPI value, evaluation period, cohort_id, label). Here, the real label y corresponds to the success or failure of the synergy, and is defined as y=1 for success (Pass) and y=0 for failure (Fail). The headquarters sets multiple candidate thresholds theta for the integrated index S_index obtained by the score calculation unit, and calculates the following judgments and statistics for each theta. The predicted label ŷ(theta) is defined as 1 if S_index >= theta, and 0 otherwise. This creates a confusion matrix from the combination of the actual label y and the predicted label ŷ(theta), with the following elements: TP(y=1 and ŷ=1), FP(y=0 and ŷ=1), FN(y=1 and ŷ=0), TN(y=0 and ŷ=0). Based on the confusion matrix, the following derived indices are calculated (if the denominator is 0, it is set to 0): TPR(sensitivity, Recall)=TP / (TP+FN) FPR (false positive rate)=FP / (FP+TN) Precision=TP / (TP+FP) F1(Harmonic Mean)=2*Precision*TPR / (Precision+TPR) Youden's J=TPR-FPR The headquarters sweeps the candidate set Theta (e.g., 0 to 5 at equal intervals) and finds the threshold theta* that maximizes a predetermined performance index (default is Youden's J). If necessary, the search can be performed with additional indexes such as F1 or constraints (e.g., FPR<=alpha). The output is the trained theta*, the performance index value at that time (e.g., Youden's J, F1, etc.), the applied cohort_id, Method / PL version information, the number of data items used, the search range, execution time, and other historical information. These are used to set the theta operation policy (reference trained theta* / provisional theta0) in the [Evaluation guideline selection section] and the reference value in the [Penalty / Gate judgment section]. [Visualization output part]

[0027] The headquarters accepts {S_axis, A_primary, S_index} and theta (theta0 or theta*) as input and generates an 8-axis radar diagram with a fixed axis order and fixed radius from 0 to 5. The primary axis is highlighted with an annotation / marker, and theta is overlaid as a reference ring. The output is an image object (radar diagram) and provenance metadata (cohort_id, Method / PL version ID, POID, CI[theta*], creation time, hash, etc. (digest value for tamper detection and identity verification, e.g., SHA-256)), which are passed to the report generation unit. In addition, headquarters will refer to the m flag (HYP / DSC) and reflect it in the annotations (hypothesis verification / exploration) and threshold display (theta operation) in the visualization output. The primary axis is determined according to the procedure specified in Identifying Primary Axis. That is, (i) the axis with primary flag = 1 is given top priority, (ii) the axis with the largest I / E / S contribution C(a) = SUM_{i in a}{(W*q_i)*s_i} is adopted, and (iii) in the event of a tie, a fixed axis order is used to resolve the tie. The determined primary axis is highlighted in the 8-axis synergy radar diagram (radar diagram), explanation generation, and annotations. [DCF Cooperation Department]

[0028] Headquarters takes {kappa, tau, omega, c, alpha_t} and scale parameters (S_ann, etc.) as inputs and calculates DeltaRev_t, FCF_oper_t, DeltaWC_t, CAPEX_t sequentially over the period t=1...T, obtaining DeltaNPV and IRR from the DeltaFCF_t series. If necessary, it applies the total penalty coefficient P_total to output the corrected DeltaNPV*. The output is {DeltaFCF_t series, DeltaNPV, DeltaNPV*, IRR, history of parameters used}, which is passed to the penalty / gate judgment unit and report generation unit. <parameters> kappa: Sales → FCF conversion rate (pre-tax coefficient including expense adjustments, 0<=kappa<=1) tau: tax rate (0<=tau<=1) omega: working capital ratio (0<=omega) c: CAPEX ratio (0<=c) S_ann: Steady-state annual incremental sales (annual amount at full ramp) r: Discount rate (WACC) alpha_t: Ramp rate (0 to 1) for year t, given by the ramp-up function alpha_t=f(t;phi) (the function form is selected depending on the project). P_total: Total penalty coefficient (0<=P_total<=1, depreciation coefficient for feasibility, regulations, etc. calculated separately) <Ramp function example> Linear (upper bound): alpha_t = min(1, a*t) (where a is the scaling factor) Step: Update the level at a given point in time and eventually converge to 1 S-curve (logistic): alpha_t = 1 / (1 + exp(-k*(t - t0))) (k>0: slope, t0: center) Exponential convergence: alpha_t = 1 - exp(-lambda*t)(lambda>0) Table driven: Determine alpha_t from a given time series table and use interpolation as needed In either case, the range is 0<=alpha_t<=1. In this example, alpha_t = min(1, a*t) is used. <Calculation procedure (DeltaNPV, DeltaNPV*)> DeltaRev_t = alpha_t * S_ann FCF_oper_t = kappa(1-tau) * DeltaRev_t DeltaWC_t = omega * DeltaRev_t CAPEX_t = c * DeltaRev_t DeltaFCF_t = FCF_oper_t - DeltaWC_t - CAPEX_t Discount factor: Disc_t = (1+r)^{-t}, therefore DeltaNPV = SUM_{t=1}^T (DeltaFCF_t * Disc_t) Penalty application: DeltaNPV* = P_total * DeltaNPV (if there are separate cost deductions, subtract them from DeltaNPV before applying P_total) IRR definition: NPV(r_IRR) = SUM_{t=1}^T (DeltaFCF_t / (1+r_IRR)^t) = 0. Let r_IRR be the IRR (obtained by numerical solution). <Supplementary information>

[0029] The timing is based on the end-of-year valuation (end-of-period cash). If approximation of the occurrence during the period is required, a semi-annual adjustment, etc. will be separately set. - Amount units should be consistent within the project (e.g., million yen). The parameter history records the function form and coefficients of {kappa, tau, omega, c, alpha_t}, S_ann, r, P_total}, period T, evaluation date and time, Method / PL version, etc. [Penalties and Gate Judgment Section]

[0030] Headquarters will use {S_index, theta* (or theta0), DeltaNPV, DeltaNPV*, IRR} as input and evaluate non-financial gates (S_index >= theta*) and financial gates (IRR >= r [default: WACC] and DeltaNPV* >= 0) using a unified procedure. For both comparisons, boundaries containing "=" are considered to be successful. If theta* has not been learned or is not adopted, the provisional value theta0 will be used. The output is the judgment result (PASS / FAIL), the reason for the result (sign of the difference S_index-theta*, IRR-r, DeltaNPV*, breakdown of penalty factors), and historical metadata. This is passed to the report generator. In addition, headquarters will refer to the m flag (HYP / DSC) and reflect it in the gate judgment threshold operation (theta* / theta0) and the judgment basis notes. <Supplementary information>

[0031] ·r is the discount rate for the project (default is WACC), and the value set at the time of evaluation is used. Delta NPV* uses the value after applying penalties defined in [DCF Linkage Section]. · Units and currencies will be unified within the project, and comparative amounts will be evaluated in the same units. If necessary, it is possible to allow provisional judgments (e.g., HYP) only for non-financial gates, and limit financial gates to confirmed data (e.g., DSC). [Report Generation]

[0032] The headquarters will integrate the outputs from the visualization output section, penalty / gate decision section, and DCF linkage section to generate (i) machine-readable (standard formats such as JSON / XML) and (ii) human-readable (PDF / dashboard) data. At a minimum, the final decision (PASS / FAIL), S_index, theta and difference, Top-K major contributors, used mapping / control points, missing / surrogate / estimated treatment, DeltaNPV / DeltaNPV* / IRR, cohort_id, Method / PL version ID, POID, CI[theta*], evaluation reference period, generation time, hash, etc. will be co-stamped and output. Here, the Top-K major contributors refer to the top K items (default K=10, all items within K will be displayed if the same ranking applies) extracted in descending order by contribution value v_i (default: v_i=(W*q_i)*s_i) for all KPIs. If the top K by axis is also listed, this will be clearly stated. In addition, headquarters will refer to the m flag (HYP / DSC) and indicate the mode (hypothesis verification / exploration) in the report output, recommend collection, and add notes requiring review. (Notes on Examples) The numerical values ​​described in this section are simulated values ​​used to explain the operation procedures of the invention and do not represent actual measured values ​​for a specific case. They can be easily replaced with publicly available information or internal company information under NDA. [Example]

[0033] (Telecommunications sector: T-Mobile US * Sprint) This example focuses on a major merger case in the telecommunications industry (the business merger between T-Mobile US and Sprint). The M&A was announced on April 29, 2018, and completed on April 1, 2020. The transaction was valued at approximately USD 26.5 billion at the time of announcement and was conducted as an all-stock exchange. The purpose of the merger is to accelerate the nationwide rollout of 5G networks and improve network efficiency by integrating the base stations and spectrum assets of both companies. According to public information, the merger is expected to generate synergies of more than US$43 billion over the five years following the merger. In this system, key KPIs (net customer growth rate, ARPU, network equipment efficiency, frequency coverage, etc.) are extracted based on publicly available information, and the input acquisition unit obtains the preprocessed input value x_i. The normalization unit processes this data according to the mapping type map_type_i and the control points (L, T, H, TL, TH) on Theta set by the Industry Information Management Unit, and calculates the normalized score s_i for each KPI. The calculated {s_i} is weighted by I / E / S in the score calculation section to find S_axis, and then passed through the integrated score S_index to the gate judgment section to obtain a PASS / FAIL evaluation.

[0034] (1) Cohort cohort_id:TELCO_US_2020(Industry=Telecommunications, Region=US, Size=Large, Model=MNO) (2) Table 1 shows an example of KPI setting and normalization. [Table 1] (3) Axis generation and S_index calculation S_index=3.58 (simulated), theta*=3.25 → PASS. Weighted by I / E / S; ḡ(CI)=0.85. (4) Handling of missing values, surrogate values, and estimated values Secondary dimensions (organization, regulation, etc.) are treated as proxies or estimates using POIDs, and DeltaCI (confidence change) is annotated for each KPI. (5) Report generation The 8 synergy axes, top K contribution values, control points, Method / PL ID, POID, and CI notes are output. Figure 6 shows the radar diagram (T-Mobile*Sprint) Synergy Index for Example 1. [Example]

[0035] (Software Sector: IBM * Red Hat) This example focuses on a technology acquisition case in the software industry (IBM's acquisition of Red Hat). The M&A was announced on October 28, 2018, and completed on July 9, 2019. The transaction was valued at approximately US$34 billion, making it the largest acquisition in the software industry at the time. The objective of the agreement is to establish a hybrid / multi-cloud strategy based on open source technology, strengthening IBM's competitive advantage in the cloud field. It was made clear that the two companies will jointly develop and deploy enterprise cloud infrastructure while maintaining Red Hat's independence and neutrality. In this system, the KPIs extracted are OSS contribution, customer retention rate, product license sales composition ratio, subscription revenue ratio, etc. For these KPIs, the preprocessed input value x_i is obtained, and s_i is calculated based on map_type_i and the control point on Theta. Next, the S_axis (each axis including technology synergy, customer base synergy, brand affinity, etc.) is aggregated and quantified as an integrated score, S_index. The gate judgment section applies the financial indicators Delta NPV·IRR and the weighting adjustment g(CI_i) based on CI_i, and outputs the results in one of three categories: PASS / FAIL / Needs further consideration.

[0036] (1) Cohort cohort_id:SOFTWARE_US_2019 (industry=Software, region=US, size=Large, model=Subscription) (2) Table 2 shows an example of KPI setting and normalization. [Table 2] (3) Axis generation and S_index calculation S_index=4.00, theta*=3.25 → PASS. Weights w={0.5,0.3,0.2}, ḡ(CI)=0.95. (4) Treatment of proxy values Insufficient decomposition of NRR → surrogate KPI (ARR_end / ARR_start) used with POID; DeltaCI=-0.03. (5) Report generation The 8 synergy axes, top K results, control points, Method / PL, POID, and CI notes are output, and the differences between the cohorts are compared with Example 1. (6) DCF correlation output typical value In this example, the parameters obtained in items (1) through (5) were applied to the DCF linkage module, which automatically generated the discounted cash flow series, adjusted NPV, IRR, and final gate decision. Table X shows a typical output of the financial linkage results. Table 3 shows a representative example of the outputs associated with DCF. [Table 3] Table 4 shows a summary of the financial decisions ("PASS if (IRR >= r and DeltaNPV* >= 0)) [Table 4] This confirmed that the proposed DCF linkage module generates project-level financial indicators in a manner consistent with non-financial scoring and integrates them into the integrated gate assessment. Figure 7 shows the radar diagram of Example 2 (IBM*Red Hat) Synergy Index. [Example]

[0037] (Consumer Goods Sector:AB InBev * SABMiller) This example focuses on a portfolio expansion consolidation case in the consumer goods and beverage industry (AB InBev's acquisition of SABMiller). This M&A was agreed upon in 2015 and completed on October 10, 2016, with a total transaction value of approximately US$103 billion. The objectives of the transaction were to expand the brand portfolio in major global markets and optimize the supply chain, and after the integration, one of the world's largest beer companies, known as "Megabrew," was formed. In this system, the KPIs input are sales volume growth rate, average selling price, cost rate, regional revenue dispersion, brand asset score, etc. These pre-processed input values ​​x_i are normalized based on map_type_i and the control points on Theta to obtain s_i. Next, S_axis is calculated by taking the weighted average of I / E / S, and the integrated score S_index is calculated. The DCF linkage part calculates DeltaNPV and IRR from the DeltaFCF_t series, and the gate judgment part judges PASS / FAIL against the threshold theta*. (1) Cohort

[0038] cohort_id:BEVERAGE_GLOBAL_2016(Industry=Beverage, Region=Global, Size=Mega, Model=FMCG) (2) Table 5 shows an example of KPI setting and normalization. [Table 5] (3) Axial integration and calculation of S_index S_index = 2.71 (simulated value), theta* = 3.25 → verdict: FAIL. Weights w = {0.10, 0.10, 0.25, 0.35, 0.20}, mean confidence interval correction coefficient g_bar(CI) = 1.00. (4) Financial collaboration and gate determination The DCF module calculates DeltaNPV and IRR and applies the penalty coefficient P_total. The judgment condition is PASS if (S_index >= theta* and IRR >= r and DeltaNPV* >= 0). (5) Report generation Output includes synergy index, value bridge, CI weighted results, Method / PL information, and POID notes. Figure 8 shows the radar diagram of the Synergy Index for Example 3 (AB InBev*SABMiller). The parameters used in the embodiment are summarized in FIGS. 9 and 10.

Claims

1. By computer, (a) obtaining KPI values ​​for the target company, and normalizing the KPI values ​​to multiple synergy axes as bounded scores with predetermined lower and upper limits using a monotonic piecewise linear mapping; (b) identifying a primary axis from the axis scores; (c) Generate a radar diagram (8-axis synergy radar diagram) with a predetermined axis order and fixed radius (polar coordinate plot with the upper score limit fixed as the upper radius limit), overlay a reference ring, and highlight the main axis with a marker to output it. A method for visualizing M&A synergies, including:

2. 2. The method of claim 1, wherein the lower and upper bounded scores are either 0 and 5, 0 and 10, or 0 and 100, respectively.

3. In the method described in claim 1 or 2, the synergy axis includes at least one of the following axes: sales synergy, cost synergy, financial synergy, human resources / organization, intellectual property / technology, regulations / licenses, market / competition, and IT / digital, and may also include axes with names that are substantially synonymous with each of these axes.

4. In the method according to any one of claims 1 to 3, the monotonic piecewise linear mapping is selected from shape types including a bandwidth-optimized unimodal type in addition to the index direction (upward / downward), the shape type is at least one of ascending, descending, triangular, and trapezoidal, the mapping has at least three or more control points (thresholds), and supports outlier processing (clipping or winsorization).

5. In the method according to any one of claims 1 to 4, the identification of the main axis is performed based on an integrated score obtained by weighting and combining Impact, Ease, and Speed, or a main axis flag set by a user or a system.

6. The method of any one of claims 1 to 5, further comprising imprinting the output with provenance metadata comprising at least one of version control information, source information, confidence, generation time, and tamper-detection hash, or a combination thereof, wherein the version control information includes at least one identifier or version number for identifying a change history of a method, library, parameter set, etc.

7. In the method of claim 1, the reference ring represents a decision threshold, which is initially set to a provisional value theta_0 and is replaced by the decision threshold theta* after learning, where theta* is estimated by maximizing a performance index (at least Youden's J or F1) for past cases labeled as success or failure.

8. In the method of claim 5, the weight w_i of each KPI is determined by multiplying the reference weight W of Impact / Ease / Speed ​​by a correction coefficient g(CI_i) corresponding to the data reliability CI_i of each KPI, and normalizing it within the axis.

9. In the method according to claim 5 or 8, the integrated score S_index is calculated by convex combination of an arithmetic average S_avg of the multi-axis scores and a main axis score S_main, S_index = (1-beta)*S_avg + beta*S_main where beta is a coefficient satisfying 0<=beta<=0.5 and is determined by at least one of the following: (i) a predetermined default value, (ii) a value estimated by maximizing a performance index (at least correlation coefficient, AUC, or Youden's J) for past cases labeled with success / failure, or (iii) a user-specified value.

10. In the method according to any one of claims 1 to 9, at least a part of the non-financial gate is performed by comparing the integrated index S_index with a decision threshold theta* (S_index >= theta*), and theta* is estimated by the learning process according to claim 7.

11. The method according to claim 6, wherein the provenance metadata includes a parameter overwrite identifier (POID) at the time of overwrite, a reason for overwrite, an application period, a creator, and a creation time.

12. 12. The method according to any one of claims 1 to 11, The method further includes a step of generating a DeltaFCF series based on financial conversion parameters including at least one of the incremental sales to FCF conversion rate kappa, tax rate, realized ramp alpha, working capital ratio omega, CAPEX ratio c, evaluation base period, and discount rate, and calculating at least one of the discounted present value (DeltaNPV) or internal rate of return (IRR) from the DeltaFCF series.

13. In the method of claim 12, when at least one of non-compliance with regulations, non-established dependencies, cannibalization, excessive integration costs, insufficient cultural fit, and insufficient data reliability is detected, a non-negative penalty amount associated with each event is subtracted from Delta NPV to calculate a revised Delta NPV, and the revised Delta NPV is used for subsequent judgments and output.

14. In the method of claim 12 or 13, the gate judgment is performed by configuring at least one of the following gates: a first layer: financial gate (IRR >= cost of capital (WACC) and Total Delta NPV or Adjusted Delta NPV >= 0) and a second layer: non-financial gate (thresholds for regulations, dependencies, and data history), or a combination thereof, and if all gates are met in the judgment, it is judged as PASS, and if any gate is not met, it is judged as FAIL, and the judgment result (PASS or FAIL) is output.

15. In the method according to any one of claims 12 to 14, the library parameters are automatically or manually re-estimated in response to the arrival of market data or test implementation data, and the old version is replaced with the new version and the impact difference is logged.

16. In the method of any one of claims 12 to 15, evaluation is performed on multiple candidate target companies under the same set of evaluation parameters, the top K cases (K is a predetermined natural number) are selected from the gate-passed cases based on Delta NPV (or a composite indicator), and each selected case is classified into a focus-specific cluster based on the focus axis and presented.

17. A program that causes a computer to execute the method according to claim 1.

18. An information processing device that executes the method according to claim 1.

19. A data structure for reference by the M&A synergy evaluation process, The data structure comprises (a) a cohort identifier, (b) a KPI definition record (name, unit, indicator direction / shape type / control point / outlier treatment), (c) weights, penalties, and gate parameters, (d) financial conversion parameters, and (e) history and version control metadata (at least one or a combination thereof), and by referencing the data structure, the following actions can be performed: normalization → focus identification → (optionally) weighting → visualization, and the DeltaFCF / DeltaNPV / IRR calculation and gate judgment described in claims 12 to 14. The cohort_id identifier includes one or more combinations of industry classification, region classification, size classification, business model classification, market segment, accounting standard / currency, evaluation reference period, growth stage, and transaction type. The data structure stores, in association with the identifier, direction / shape, control points, outlier treatment, weights, penalties, and gate thresholds, financial parameters, quantiles and standardization criteria, number of references and confidence indicators, and library version ID, and is characterized in that it has a resolution rule based on parent-child inheritance in a hierarchical or composite form.

Images