Témata prací (Výběr práce)Témata prací (Výběr práce)(verze: 368)
Detail práce
   Přihlásit přes CAS
Testing the Presence of Adaptive Switching Behavior in Equity Markets
Název práce v češtině: Testování přítomnosti adaptivního chování v akciových trzích
Název v anglickém jazyce: Testing the Presence of Adaptive Switching Behavior in Equity Markets
Klíčová slova: omezená racionalita, adaptivní chování, intensita volby, efektivita trhů
Klíčová slova anglicky: Bounded Rationality, Adaptive Switching, Intensity of Choice, Market Efficiency
Akademický rok vypsání: 2014/2015
Typ práce: diplomová práce
Jazyk práce: angličtina
Ústav: Institut ekonomických studií (23-IES)
Vedoucí / školitel: PhDr. Jiří Kukačka, Ph.D.
Řešitel: Mgr. Filip Staněk - zadáno vedoucím/školitelem
Datum přihlášení: 17.06.2015
Datum zadání: 18.06.2015
Datum a čas obhajoby: 14.09.2016 00:00
Místo konání obhajoby: IES
Datum odevzdání elektronické podoby:22.07.2016
Datum proběhlé obhajoby: 14.09.2016
Oponenti: PhDr. Natálie Švarcová, Ph.D.
Kontrola URKUND:
Seznam odborné literatury
Brock, W., Lakonishok, J., & LeBaron, B. (1992). Simple technical trading rules and the stochastic properties of stock returns. Journal of Finance, 1731-1764.

Sullivan, R., Timmermann, A., & White, H. (1999). Data-snooping, technical trading rule performance, and the bootstrap. Journal of Finance, 1647-1691.

Isakov, D., & Hollistein, M. (1999). Application of simple technical trading rules to Swiss stock prices. Financial Markets and Portfolio Management, Vol. 13, No. 1, pp. 9-26.

Tian, G. G., Wan, G. H., & Guo, M. (2002). Market efficiency and the returns to simple technical trading rules: new evidence from US equity market and Chinese equity markets. Asia-Pacific Financial Markets, 9(3-4), 241-258.

Wong, W. K., Manzur, M., & Chew, B. K. (2003). How rewarding is technical analysis? Evidence from Singapore stock market. Applied Financial Economics, 13(7), 543-551.

Lento, C., & Gradojevic, N. (2011). The profitability of technical trading rules: a combined signal approach. Journal of Applied Business Research (JABR), 23(1).

Bajgrowicz, P., & Scaillet, O. (2012). Technical trading revisited: False discoveries, persistence tests, and transaction costs. Journal of Financial Economics, 106(3), 473-491.

Edwards, R. D., Magee, J., & Bassetti, W. H. C. (2012). Technical analysis of stock trends. CRC Press.

Pring, M. J. (2014). Technical Analysis Explained: The Successful Investor's Guide to Spotting Investment Trends and Turning Points. McGraw Hill Professional.

Cont, R., Kukanov, A., & Stoikov, S. (2014). The price impact of order book events. Journal of financial econometrics, 12(1), 47-88.

Zhou, W. X. (2012). Universal price impact functions of individual trades in an order-driven market. Quantitative Finance, 12(8), 1253-1263.

Almgren, R., Thum, C., Hauptmann, E., & Li, H. (2005). Direct estimation of equity market impact. Risk, 18(7), 58-62.
Předběžná náplň práce v anglickém jazyce
It is almost four decades since the first applications of agent based modeling to financial markets. During that time, the approach was shown to be quite fruitful in replicating many peculiar features of financial time series such as excess volatility, excess kurtosis or autocorrelation of absolute returns, which cannot be easily explained with the standard rational expectations models. An agent based modeling, build on a premise that agents poses knowledge of multiple trading strategies and decide whichever to use based on their past profitability, seems to be quite reasonable and appealing way of depicting financial markets as it does not require unrealistically high cognitive abilities on the side of agents and closely resembles the never-ending search for profitable opportunities described by financial practitioners.
This is however not without costs. Agent based models are usually coupled with many ad hoc assumptions regarding trading strategies and in particular, the way agents select them. The rule translating past profitability of individual strategies into their popularity, often called the switching function, is commonly assumed to be the symmetric logit model determined by the single parameter intensity of choice which is of crucial importance for the dynamics of the model. Estimates of the parameter nevertheless vary and in many cases are not significantly different from zero, implying that agents are inattentive to the past profitability and basically invalidating the whole concept.
To shed some light on the way in which profitability of strategies translates into their popularity, we propose a method which might recover the switching function directly from market data. We test the performance of the method in a controlled environment of an artificial stock market and then apply it to the real stock market data to test various hypothesis regarding the sensibility of assumptions usually imposed on the switching function, namely symmetricity across strategies, stability over time, and the precise functional form.

1. Hypothesis: Market movements conveys information about strategies that were used by traders in a given period which allows for inference about the switching function
2. Hypothesis: The switching function is symmetric across strategies
3. Hypothesis: The switching function is stable over time
4. Hypothesis: The switching function is invariant across different markets
5. Hypothesis: The parameter intensity of choice is significantly different from zero, that is: popularity of a particular strategy is increasing in its own profitability

The method is designed to recover the switching function in an environment commonly assumed in financial agents based models, that is in a market which is populated by some population of strategies whose popularity is determined by their respective performance through the switching function (or equivalently, that the market is populated by agents who switch between the strategies based on their performance). In such environment, it is possible to recover the switching function by focusing only on some sample from the population of strategies (as the whole population is unlikely to be known in the real markets) and comparing orders that would be placed by strategies included in the sample with the price change in the given period.
More precisely, provided that the price impact function is known, we can translate time series of price into time series of excess demand. Since the excess demand is given by the sum of products of relative popularities of individual strategies and their indicator function which attains values {-1,0,1} depending on whether the strategy is currently inactive or active and entering into short or long position, we can treat the sum over the strategies excluded from our sample as a noise and estimate the likely shape of the switching function for strategies included in the sample as both their past profitability and the value of the indicator function is observable in every period.
We test the performance of this method in a controlled environment of an artificial stock market with randomly generated trading strategies trading based on signs of past price changes and a known switching function. Focus will be placed on the nature of the error, the effect of resampling, the misspecification of the price impact function and structural assumptions needed to be imposed on the switching function in order to estimate the model.
Next we apply this method to the real world data. We collect a panel of stock market prices for individual stocks at daily frequency (other frequencies are also plausible, we nevertheless believe that this choice will deliver more precise estimators, see below) and translate these into estimates of daily excess demand using price impact functions. Further, we choose a sample of trading strategies operating on day to day data. This sample can be either randomly generated or based on surveys describing common technical trading rules used by financial practitioners (see the Core Bibliography). The latter might be preferable as the sample of strategies will be likely small compared to the whole population of strategies, implying that price movements originated from changes of profitability within the sample will be negligible compared to the noise term. Including strategies with are known to be used by practitioners might to some degree alleviate this issue. The estimates which we obtain will be used to test assumptions commonly imposed on the switching function (see Hypotheses).

Expected Contribution:
1. Exploring effectiveness of the suggested direct estimation method
2. Testing validity of assumptions commonly imposed on the switching function

1. Introduction
2. Literature Review
3. Theoretical Framework
4. Estimation Method and its Robustness (Monte Carlo Simulations)
5. Estimation
6. Conclusions
Univerzita Karlova | Informační systém UK