Değişen pazar yapısı ve rekabet koşulları şirketleri, yeni çözümler geliştirme zorunluluğu ile karşı karşıya bırakmıştır. Birleşmeler, şirketlerin yeni çözüm arayışları sonucu gündeme gelmiş buluşlardan bir tanesidir. Faaliyetlerin daha etkin yürütülmesi, faaliyet sinerjisi ve finansal sinerji elde edilmesi, yönetim etkinliğinde artış, piyasa payı, ürün geliştirme ve dağıtım sistemindeki ilerlemeler, marka, patent birleşmelerin başlıca nedenlerini oluşturur. Birleşmelerin doğru bir strateji olabilmesi için, birleşilecek veya satın alınacak şirket seçiminin, çok iyi analiz edilmesi gerektiği, bu makalede önemle vurgulanmıştır. Birleşmeler sağlıklı yapılması durumunda anlamlı olacaktır. Bu da birleşme sürecinde doğru tekniklerin kullanılması anlamını taşır. Makale kapsamında, şirketleri birleşmeye iten nedenler ve birleşen firmalarda performans artışının gerçekleştiği, stratejik planlamanın şirket birleşmeleri ile olan ilintisi anlatılmış, birleşme süreci yapay sinir ağları ile analiz edilmiş, karar vericiye sunulmak üzere, birleşme için bir yapay sinir ağı modeli oluşturulmuş, modelin oluşturulmasında MATLAB yazılımı kullanılmıştır. Neden sinir ağları sorusunun cevabı ise, onun teori gereksinimi esnektir, araştırma yaklaşımı kuralcı değildir ve bilinmeyeni sunuş şekli bulanık küme tabanlıdır ve en önemli özelliği gerçek dünya problemlerine uygulanabilir olmasıdır. Türkiye’de şirket birleşmeleri henüz gelişme aşamasındadır ve tam anlaşılamamıştır. Yöntemleri konusunda yatırımcılar yeterli bilgiye sahip değildir. Makalenin amacı birleşmenin şirketlerin büyümesi için bir yöntem olduğu, bu sürecin çok iyi analiz edilmesi gerektigi ve en doğru şirketi bulmak için sinir ağları modeli kullanılmasi önerilmiş, Türkiye’nin bir eksikliğine katkı sağlanması amaçlanmıştır.

Today companies have been engaged in a new pursuit for adapting themselves to the changing market and competition conditions. Mergers are among the trends which have gained wider interest, especially in recent years. In order for the mergers to be a true strategy, the choice of company which will be merged or acquired has to be perfectly analyzed. This process was analyzed via artificial neural networks. Predictor variables selection is important factor about quality of the prediction model to get success. Predictor variables that used as inputs on the neural network model are specified on the basis of six hypotheses being frequently suggested in academic and popular financial literature. These are inefficient management hypothesis, growth resource mismatch hypothesis, industry disturbance hypothesis, size hypothesis, market-to-book hypothesis, priceearning hypothesis. 12 predictor variables that support these hypothesis are average excess return (AER), return on equity (ROE), average turnover (AVTURNOV), average growth (AVGROWTH), average liquidity (AVLIQUID), average leverage (AVLEV), GRDUMMY to see AVGROWTH, AVLIQUID, AVLEV combination, IDUMMY to see sector impact on merger and acquisition, total net book value of asset (SIZE), market-to-book (MTB), price and earnings (PE), average payout (AVPAYOUT). Based on inefficient management hypothesis, inefficiently managed firms are often acquired. AER, ROE, AVTURNOVER are chosen predictor variables to measure firm’s management’s performance. Growth-research mismatch hypothesis, firms with high growth and low resources or with low growth and high resources are likely to be acquired. The combination of AVGROWTH, AVLIQUID and AVLEV is to form GRDUMMY variable that shows growth-research mismatch impact. IDUMMY variable will be selected based on industry hypothesis to cluster industry impact on merger and acquisition. Based on size hypothesis, smaller firms are more likely to be acquired than larger firms. Size will be selected as a variable as well. Market-to-book hypothesis, the companies have got low market-tobook ratios are likely to be targets and same logic deals with price-earnings hypothesis. For these 12 variables, the data were obtained from COMPUSTAT. Most of the data items for variables were averaged over three to four years prior to observation year. Merger and acquisition process has been analyzed by the artificial neural network. To train the multilayer network to predict the company merger and acquisition, back propagation algorithm has been used. The advantages of back propagation algorithm is provided with a set of examples of proper network behavior where an input to the network and corresponding target output. Working approach of algorithm is to adjust the network parameters in order to minimize the mean square error. The first part of the article deals with the stimulators of the merger, the performance increase in merged companies, and the connection of strategic planning with company mergers. In the second part, artificial neural networks, the method used in the merger and acquisition process, is investigated in scope and structure. The reason for handling the artificial neural networks is that their requirement for a theory is flexible, their research approach is not prescriptive, their presentation of the unknown is fuzzy based, and most importantly, its adaptability to the real world problems. It is a considerably difficult process to determine the layer number and number of nodes on these layers that are optimum for acquiring the best neural network model. Several combinations of hidden layers and nodes are tried before reaching the satisfactory model. This process takes a long time and the optimal network is produced after many trials. The activation acquiring process cannot be realized without the computer support. In this process MATLAB 6.5 is utilized which is explained in the third part. In Turkey company mergers have not reached their maturity and they can not be completely understood. Investors do not have sufficient information on its methods. This article aims to support the view that a merger is a way for a company to grow, and to contribute to a better understanding in Turkey by making use of neural network models for identifying the best company to acquire.