<<  Обеспечение финансовой устойчивости торговой компании в условиях кризиса Издательство Дрофа Представляет  >>
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Programming Neural Networks and Fuzzy Systems in FOREX Trading
Basic course info: Accessibility of course materials
Basic course info: Accessibility of course materials
Basic terms of Learning Algorithms (LA)
Basic terms of Learning Algorithms (LA)
Neurons in Biology
Neurons in Biology
Neurons in Biology
Neurons in Biology
Neurons in Biology
Neurons in Biology
Neurons in Biology
Neurons in Biology
Comparison of Neuron with Silicon-based Hardware
Comparison of Neuron with Silicon-based Hardware
Comparison of Neuron with Silicon-based Hardware
Comparison of Neuron with Silicon-based Hardware
Biologic Neuron and its Mathematical Model
Biologic Neuron and its Mathematical Model
Картинки из презентации «Programming Neural Networks and Fuzzy Systems in FOREX Trading» к уроку экономики на тему «Компании»

Автор: Balazs Kovacs, Dr. Gabor Pauler. Чтобы познакомиться с картинкой полного размера, нажмите на её эскиз. Чтобы можно было использовать все картинки для урока экономики, скачайте бесплатно презентацию «Programming Neural Networks and Fuzzy Systems in FOREX Trading.ppt» со всеми картинками в zip-архиве размером 854 КБ.

Programming Neural Networks and Fuzzy Systems in FOREX Trading

содержание презентации «Programming Neural Networks and Fuzzy Systems in FOREX Trading.ppt»
Сл Текст Сл Текст
1Programming Neural Networks and Fuzzy 12function (Eg. Sin(x)) with a suitably
Systems in FOREX Trading. Presentation 0 paramete-red higher order polynom in a
Bal?zs Kov?cs (Terminator 2), PhD Student given range Fourier Transform
Faculty of Economics, University of P?cs (Fourier-transzform?ci?): a complex
E-mail: Dr. nonlinar function (eg. Stock price, sound
Gabor Pauler, Associate Professor wave, etc.) is assembled as weighted sum
Department of Information Technology of Sin(x) type functions with different
Faculty of Science, University of P?cs wavelenght and phase. Evaluation of
E-mail: analytic methods: ? They have relatively
2Content of the Presentation. Basic low Computational requirement
course info Purpose of the Course Course (Sz?mol?sig?ny) ? They require high level
Agenda Accessibility of course materials analytic mathematical knowledge ? They are
FOREX Trading Bot Building Contest not Modular (Modul?ris): modeling any
Requirements, Grading and Consultation additional local „bumps” or „steps” will
Introduction Basic terms of Stock result exponentially more complex global
Market/FOREX Fractal Theory of formulation Therefore, we will not deal
Stock/Currency Pair Prices Basic terms of with analytic approximation methods in
Distribution Free Estimators (DFE) Basic this course. Instead of them, we will use
terms of Rule-Based Systems (RBS) Basic Rule Based Systems, RBS (Szab?lyalap?
terms of Learning Algorithms (LA) Basic rendszerek).
terms of Artificial Neural Networks (ANN) 13Basic terms of Rule-Based Systems
Biologic analogy: Neurons in Human brain (RBS) 1. Rule-Based Distribution Free
Comparison with silicon-based hardware Estimators (szab?ly-alap?
Biologic Neuron and its Mathematical Model eloszl?sf?ggetlen becsl?si rendszerek)
Neural references. approximate Control Function (Vez?rl?si
3Basic course info: Purpose of the f?ggv?ny) ( ? ) among Input-Output
Course. Nowadays there are wide range of Variables (I/O v?ltoz?k) of Decision Space
fancy FOREX „course providers” which (D?nt?si t?r) with the help of Rule Basis
promise that you can be a millionaire (Szab?lyb?zis) containing k=1..l finite
within 4 weeks, without any serious set of rk Rules(Szab?ly): They Associate
economic and mathematic training, just (Egym?shoz rendel) vi, vj values, or [vil,
completing 1 week rapid course where you viu] intervals of xi i=1..n input and yo
learn drawing curves to charts visually. o=1..O output variables They have
By contrast, we do not teach how you can Linguistic(Nyelvi) representation: IF
be rapidly millionaire with FOREX. Instead InputVar1 = Intreval AND InputVar2 =
of it we teach how to avoid loosing Interval AND.. THEN OutputVar = Interval
everything you have very rapidly with They have Graphic (Grafikus)
FOREX: Participants will be trained representation: multi-dimensional
besides basics of FOREX and how to use Hyperbars (Hipert?glatest) in decision
Meta Trader 4 (MT4) FOREX Platform: space (we denote them Yellow ?) Rules of a
Recognizing their psychologic limits and rule basis can be Mutually Exclusive
assembling customized trading strategies (k?lcs?n?sen egym?st kiz?r?ak): they have
accordingly, The MQL programming language no Intersection (Metszet) = Common subset
of MT4 to create your own indicators, (K?z?s r?szhalmaz) in decision space.
Theoretical basics of Artificial Neural Alternatively, they can be Overlapping
Networks and Fuzzy Systems, Using the (?tlapol?ak) All rules of the basis has
Joone open-source Neural Shell under GNU mx(rk) Validity (?rv?nyes-s?gi) value,
license, programming it in Java and set up which shows whether the rule is Valid/
its data link with MT4. To motivate gifted Fires (T?zel) (Red ?) at a given x vector
students, we organize a FOREX Trading Bot (Green O) of input variables: x ? rk If
Contest paralel with our course, where there is only one rule in the base to fire
team 3-4 students can tune their software at any x input vector then rule basis is
to reach maximum amount of return from Non-Contradictive (Ellentmond?smentes),
limited amount of investment, within else Contradictive (?nel-lentmond?).
limited time frame making limited number 14Basic terms of Rule-Based Systems
of trades As a unique option Trading Bots (RBS) 2. Effective approximation of
with extremely high computational continous functions would require large
requirement can be run at the new number of rules in the base to get
University of Pecs Supercomputer in C++ rea-sonable Resolution (Felbont?s), eating
environment The course has English up resources To avoid this, rules can have
language course material, even if it wk?[0,1] importance weights. Estimated
presented in Hungarian: To close out otput yx* is computed as weighted sum of
simple-minded Gamblers (Szerencsej?t?kos) output values of firing rules. This is
Because – even if you have trade platforms called Interpolation (Interpol?ci?) among
and courses in Hungarian - almost every rules: yx* = Sk wk ? mx(rk) ? yk (0.1)
additional resource in FOREX you really Interpolation enables to model continous
need (eg. indicator source codes, economic control func-tions with less rules more
analysises, user guides of trading bots) effectively. It has 2 methods: Bayesian
are most freshly available only in Probability (Bayes-i val?sz?n?s?g) rules:
English! So the main outcome of this It uses mutually exclusive, Crisp (?les)
course is not being a millionaire in 4 rule base Where multiple rules can fire
weeks (which is unrealistic at FOREX binary mx(rk)?{0,1} for a given x input
anyway) but to develope proficiency using vector But simutaneous firing rules Occour
Artificial Neural Networks and Fuzzy (Bek?vetkez) only with a pk?[0,1]
Systems in a difficult simulated Probability weight (Val?sz?n?s?gi s?ly),
battleground called FOREX. And that where sum of their probabilities is 1
knowledge can result getting better paid creating Probability distribution
positions in many areas of engineeering or (Val?sz?n?s?geloszl?s): Sk pk ? mx(rk) = 1
business. (0.2) ? It is supported by Probability
4Basic course info: Course Agenda. theory (Elm?let) ? It requires data about
Legenda: Presented by Dr. Gabor Pauler. probabilities of relatively large number
Presented by Dr. Gabor Pauler. Presented of mutually exclusive rules, which is
by Dr. Gabor Pauler. Presented by Bal?zs unrealistic to get in the practice Fuzzy
Kov?cs. Presented by Bal?zs Kov?cs. Rule Inference (Fuzzy szab?ly
Presented by Bal?zs Kov?cs. Week. k?vetkeztet?s): Rule basis has overlapping
Presentation. Quiz. Grade%. Practice. Home rules: Boundary (Hat?r) of Support (Tart?)
assignment. Grade%. 0. Introduction, of one rule are in the middle of support
grading, Neural basics1. -. - . Forex of neighboured rules mx(rk)?[0,1] validity
basics 1. Install MT4, create account. of a rule can change continously: It is 1
3%. 1. Neural basics 2, Learning methods: in the middle of support and 0 at boundary
Hebb, Delta, Backpropagation1. Neural (we denote it with yellow shading),
basics 1. 3% . MT4 GUI. Basic trade in forming not crisp/fuzzy rules: they occour
MT4. 3%. 2. Backpropagation2. Learning certainly but their validity is
methods. 3%. Basic indicators. Use of uncertain/changing gradually wk weights do
indicators. 3%. 3. Joone GUI. not form probability distribution ?
Backpropagation. 3%. Indicator programming Theoretically it is less sound method ?
in MQL. MQL programming. 3%. 4. Time But can model complex nonlinear continous
series forecasting networks and their functions using much less rules/weights to
representation in Joone. Joone GUI. 3%. tune.
Compound indicators in MT4. Compound 15Basic terms of Learning Algorithms
indicators. 3%. 5. Trading strategies in (LA). Manual definition of several
MT4. Time series nets. 3%. Connecting MT4 thousand rules and their weights with the
with Joone. MT4-Joone connection. 3%. 6. help of experts is expensive, slow Thats
Rule based systems: Crisp inference CRT. why Expert System Shells, ESS (Szak?rt?i
Trading strategy. 3%. CRT in SPSS. CRT in rendszer shell) - using manual Bayesian
SPSS. 3%. 7. Rule based systems: Fuzzy probabilistic rule bases - failed to
basics. CRT. 3%. Fuzzy inference, multi become the mainstream of Artificial
valued results. Stock Futures. 3%. 8. Intelligence, AI (Mesters?ges
FuzzyTech1. Fuzzy basics. 3%. FuzzyTech2. Intelligencia) So we need Learning
Breasts. 3%. 9. Special topologies: RBF, Algorithms(Tanul? Algoritmus) which can
ART, Kohonen. FuzzyTech. 3%. Topology set up rules and their weights
diagrams. Character design. 3%. 10. automati-cally form an X,Y Sample database
Neurofuzzy, FAM. Spec topology. 3%. FAM in (Minta adat-b?zis) of pre-viously Observed
FuzzyTech. FAM. 3%. 11. Text mining (Megfigyelt) j=1..m xj ,yj vectors of xi
basics. FAM. 3%. Text mining in SPSS. Text i=1..n input/yo o=1..O output vars. They
mining. 3%. 12. Text mining topologies. have 2 groups: Classification and
Text mining. 3%. Text mining in Joone. -. Regression Trees, CRT (Klasszifik?ci?s ?s
-. regresszi?s f?k) algorithms: They can
5Basic course info: Accessibility of estimate only discrete valued (Diszkr?t
course materials. We can’t use TAB here! ?rt?k?) output variables from
All course materials are available at continous/discrete inputs (Eg. Estimate
PTE-TTK Szent?gothai Szakkoll?gium Bankrupcy/Survival of a company from its
website: financial rates) Building Decision tree (D?nt?si fa) of connected crisp Bayesian
rex/ in form of PowerPoint presentations probability rules Trying to set up rule
and practices These are NOT conventional boundary values at each input variable,
„three sentences/slide” projectable which separate best output values ? Low
presenta-tions, but almost full-text computational reqirement ? Can use only
materials with: Linked-in case study crisp hyperbar rules, which are
materials Step-by step animated software ineffective modelling complex nonlinear
usage usable at computer lab However it is Transversal (?tl?s) control functions
highly recommended for stu- dents to print Artificial Neural Networks, ANN
them out in handout format and taking (Mesters?ges Neur?lis H?l?zatok): they can
notes to slides, as questions in quiz may estimate continous/discrete outputs from
be represented from oral comments of tutor continous/discrete inputs Building kind of
also All course materials are in English „implicte fuzzy rules”, without liguistic
to cap- ture Business English But represntation and direct acces by user
presentations are in Hungarian, and we From random initial boundaries and rule
have Hungarian Notes MT4 GUI can be both. weights They can model complex nonlinear,
6Basic course info: FOREX Trading Bot transversal control functions (Eg.
Building Contest. To motivate gifted Recognizing a letter „N” from dots of ink
students, we organize a FOREX Trading Bot scanned in a picture) effectively At a
Contest paralel with our course, where price of difficult parametering and
team 3-4 students can tune their software brutally high computational requirement.
to reach. Rules of the contest are: 16Content of the Presentation. Basic
Server: FxPro MT4 Base currency: USD course info Purpose of the Course Course
Maximum Leverage: 1:50 Demo account Agenda Accessibility of course materials
capital: 5000 USD FOREX Trading Bot Building Contest Requirements, Grading and Consultation
Platform: FxPro MT4 Client Terminal Introduction Basic terms of Stock Market/FOREX Fractal Theory of
/client-terminal Operating system of Stock/Currency Pair Prices Basic terms of
trading bot: Windows 2000, XP, Vista, Distribution Free Estimators (DFE) Basic
Windows 7 Time range: 2011.11.10. 8:00:01 terms of Rule-Based Systems (RBS) Basic
- 2011.12.08. 7:59:59 Trading hours: terms of Learning Algorithms (LA) Basic
whenever markets are open Currency pairs: terms of Artificial Neural Networks (ANN)
all possible pairs of EUR, USD, GBP, CHF, Biologic analogy: Neurons in Human brain
JPY, CAD, AUD, NZD can be traded in demo Comparison with silicon-based hardware
account Who can participate: registered Biologic Neuron and its Mathematical Model
students of current course Performance Neural references.
benchmarks: Passively managed static 17Neurons in Biology. Human brain
currency portfolio, Tutors demo account contains 1011 Neurons (Idegsejt) connected
Maximal number of modifications on a with 1016 Synapses (Szinapszis) organized
trading bot: 5 Minimal number of trades in Hemi-spheres (F?lteke) > Cortexes
completed by bot: 10 (without closing) (K?reg) > Layers (Mez?) > Blocks
Using any foreign code in bots without Unisolated short Dend-rits(?g) transmit
referencing it will result in immedaiate inco-ming electric signals at 2.3m/s to
exclusion from contest Opened positions Cell membra-ne(Sejtfal)of neuron
will be closed at the end of time range by col-lecting electric charge At certain mV
tutors Winner team will be the one with potential Treshold (Hat?r?rt?k), neuron
the highest balance at the closing emits electric signal by its Signal
Identical balances among more teams will function (Jelz?si f?ggv?ny), which is
result in deuce Relative result% = Team tranmitted at 90m/s on long synapses
balance/Tutor benchmark balance of teams covered with isolator Myelin (Mielin)
compared to benchmarks can be published in jumping over the Ranvier-gaps (R?s)
university media/certificates. Excited Sytaptic termi-nals (V?gbunk?)
7Basic course info: Requirements, emit Neurotransmitter(Inger?-let?tviv?)
Grading and Consultation. Mid-semester molecules (Eg. Acetilcolin, Opiats)
requirements: Max. 10 ? 3points = 30 Opening ion channels on other neurons
points from simple 5-question quizes mem-brane making them ac-cumulat electric
written at the beginning of presentations charge.
where students are evaluated individually 18Comparison of Neuron with
Quizes are from the last presentation and Silicon-based Hardware. In the electron
practice Missed quizes can be substituted microscope image above we can see a neuron
by one extra 6 point quiz ad the end of laid on leads of a modern microchip.
semester Max. 10 ? 3points = 30 points Neuron is 10-12 times bigger than
from home assignments evaluated at project condensers and transistors of basic logic
team-level. Teams are free to reallocate gates, however it can perform such a
their home assignment points internally to non-linear computing function, which
proportionate it to contribution of their requires hundreeds of basic logic gates in
members! Home assignments are due to the a math cooprocessor. Moreover, neurons
beginning of next practice Missed home require much less energy and produce much
assignments cannot be replaced after less heat than silicon-based chips.
deadline as they are group assignments Max Currently a 100TByte blade-supercomputer
40 team points from trading bot contest = compa-rable in storage capacity with human
40 ? Relative result% Grading of brain - but still inferior in speed, as
individual students: 0-29points:Reject brain can share work among 1011 simple
signing course(0), 30-49points: Fail(1), processors instead of 103 more difficult
50-59points: Sufficit(2), 60-69points: ones - consumes 2-3 m3 space, 380V
Medium(3), 70-79points: Good(4), industrial current and cooling capacity of
80-points: Excellent(5) In case of a supermarket Human brain consumes 1500cm3
Fail(1), there are 2 possibilities for volume even storing oxygene and glucose
correction at oral exam from course for 15-20 secs of work, and requires 5-10
material of presentations to get credit Watts of power input and cooling.
Consultations: Tutors will provide 19Biologic Neuron and its Mathematical
consultation at Department of Informatics, Model. Fuctions of a neuron in ANN
PTE-TTK, at times prearranged at Mathematical Model: Non-volatile Memory or (Permanens mem?ria): ji synapses Results: connecting j=1..m neurons with i=1..n
Students can track their mid-semester neurons in the network during t=0..T time
results at periods transmit sjt?R signals of jth neuron in tth period with changing wjit
rex/ExamForex/. Intensity/ Weight(S?ly). Teaching/
8Content of the Presentation. Basic Training(Tan?t?s) of net means changing
course info Purpose of the Course Course the initially random wji0?R weights. All
Agenda Accessibility of course materials information learnt is stored as synaptic
FOREX Trading Bot Building Contest weights Volatile Memory (R?vid t?v?
Requirements, Grading and Consultation mem?ria): a neuron aggregates wjit?sjt
Introduction Basic terms of Stock weighted signals of incoming synapses into
Market/FOREX Fractal Theory of a xit Membrane value (Membr?n ?rt?k) in
Stock/Currency Pair Prices Basic terms of the Activation Process (aktiv?ci?s
Distribution Free Estimators (DFE) Basic folya-mat), additionally they Passively
terms of Rule-Based Systems (RBS) Basic decay (Passz?v lecseng?s) membran value by
terms of Learning Algorithms (LA) Basic (1-di) Decay Rate (Lecseng?si r?ta) to
terms of Artificial Neural Networks (ANN) keep membrane value within [li, ui]
Biologic analogy: Neurons in Human brain Lower/Upper bounds (Als?/Fels? Korl?t) and
Comparison with silicon-based hardware smooth (Sim?t) its changes in time. There
Biologic Neuron and its Mathematical Model are 2 methods of membrane value
Neural references. aggregation: Additive (Addit?v):
9Basic terms of Stock Market/FOREX 1. xit=di(Sj(wjit?sjt)/Sj(wjit))+
The Stock Exchange (r?szv?nyt?zsde) is a +(1-di)?xit-1 i=1..n, j=1..m, t=1..T (0.3)
Non-profit Company (non-profit t?rsas?g ), Multiplicative (Multiplikat?v):
what is Exclusive (Kiz?r?lagos) trading xit=diPj(sjtwjit)(1/ Sj(wjit))+
place of stocks of Publicly Quoted +(1-di)?xit-1, i=1..n, j=1..m, t=1..T
(t?zsd?re bevezetett, ny?lv?nosan (0.4) Aggregated membran value emits
?rfolyam-jegyzett) Companies signal by mono-tonic increasing (Monoton
(r?szv?nyt?rsas?gok). These are larger, n?vekv?) signal function with ai inflexion
stabile firms complying strict Accounting point as signal treshold and bi slope: sit
(Sz?mviteli) rules. Stocks of smaller = 1/ (1+e-bi?(xit-ai)), i=1..n, t=1..T
companies not quoted publicly are traded (0.5).
at Over The Counter (OTC) market. 20Content of the Presentation. Basic
Macroeconomic (Makrogazdas?gi) function of course info Purpose of the Course Course
stock exchange is Effective Allocation of Agenda Accessibility of course materials
Investment Resources (hat?konyan ossza el FOREX Trading Bot Building Contest
a v?llalkoz?sok k?zt a beruh?z?si Requirements, Grading and Consultation
er?forr?sokat) allocating more money to Introduction Basic terms of Stock
more profitable companies with larger Market/FOREX Fractal Theory of
growth in a public, open competition. Stock/Currency Pair Prices Basic terms of
Microoeconomic (Mikrogazdas?gi, v?llalati Distribution Free Estimators (DFE) Basic
szint?) functions of Stocks/Equity terms of Rule-Based Systems (RBS) Basic
(R?sz-v?nyt?ke): It helps Raise Funds terms of Learning Algorithms (LA) Basic
(T?k?t gy?jt) necessary for operating a terms of Artificial Neural Networks (ANN)
company and: Represents proportional Biologic analogy: Neurons in Human brain
Ownership/share (tulajdonr?sze) in a Comparison with silicon-based hardware
company, giving the right to Vote (Szavaz) Biologic Neuron and its Mathematical Model
in Board of Directors (igazgat?tan?cs) Neural references.
governing it, except: non-voting stocks 21References 1. Hungarian language
Profitable companies pay Dividend course notes: Notes Neural networks
(osztal?k) of profit for that, but it is biologic analogy:
not guaranteed, except: if it is Preferred
Stock (els?bbs?gi/aranyr?szv?ny): always Neural networks chatroom:
pays dividend, but cannot be sold and
usually does not have vote Ordinary stocks s/6007/Neural.htm GNU-licensed neural
can be Sold (eladhat?) on the stock software: Source code libraries in C++,
exchange any time at Spot Stock Price without install utility: SNNS:
(aktu?lis ?rfolyam), or at Futures
(hat?rid?s ?rfolyam) except: if the NNS/ (+install and user guide)
company has Pre-emptive Option (el?v?teli
jog), to block Hostile Takeover (t?mad?
c?l? r?szv?nyfelv?s?rl?s) by competitor
firms If we Bought (vett?k) or Underwrite
(Lejegyezt?k) a stock in the past, and c-plus-p.html
there was Hausse, Bull (?rfolyamemelked?s)
we can earn Yield (?rfolyam-nyeres?g). If itchell/ftp/faces.html
there was Baisse, Bear (cs?kken?s) then we
Loose (veszt). ts_library.html
10Basic terms of Stock Market/FOREX 2.
Equity is the most profitable but most aiparts.htm
risky tool of Investment Portfolio
Management (T?ke befektet?si portfoli? /
menedzsment): You can buy and hold (Long)
stocks of profitable and less risky
companies (Blue Chips) to make profit from
dividend or price increase, and liquidate ository/ai/areas/neural/systems/cascor/
stocks (Short) of bad companies to avoid
loss. It can use 3 basic techniques: Hedge ository/ai/areas/neural/systems/qprop/
(Fedezeti ?gylet): to short/long a stock
whose price tendencyously moves against a ository/ai/areas/neural/systems/rcc/.
price of another stock or Currency 22References 2. GNU-licensed neural
(Valuta) longed/shorted to eliminate risk software: Source code libraries in Java:
of loss from adverse price movement. Less Java Neural Networks by Jochen Fr?lich:
risky, less profitable. Arbitrage
(Arbitr?zs): short/long a stock very ehl/diplom/e-index.html (Java Class,
rapidly (1day-some hours) to make profit Internet applet about Kohonen-nets, free,
from minor price fluctuations. Medium no GUI, Tutorial in HTML)
risk, medium profitable. Speculation
(Spekul?ci?): open a short/long Position
(Poz?ci?) for longer time frame against jfroehl/diplom/e-index.html
the price Expectations (V?rakoz?s) of the
whole market, and try to influence them
with tricks to rapidly change their .html
expectations. Very risky, very profitable.
Actors of stock exchange: Broker (br?ker): /
does not own stock just trades it by
Comission (Megb?z?s) of the owner for a de/ini/PEOPLE/loos
Fee (D?j), Dealer (d?ler): can own stocks
Underwriter (undervr?jter): can buy all de/ini/VDM/research/gsn/DemoGNG/GNG.html
stocks of a new company for re-sale. From
broker to underwriter they have more .html
rights to perform difficult and risky
trades, but they have to comply more and l
more strict accounting and stock exchange
rules FOREX, FOReign EXchange eural/laboratory/laboratory.html
(Devizat?zsde) differs from ordinary stock
exchange 2 ways: Instead of trading stocks
against one Currency (Valuta) eg. (Sell
IBM?for USD), several Foreign Exchanges
(Deviza, valut?ra sz?l? sz?mlak?vetel?s) gence/complexe/neuron/mlp.html Biologic
are traded against each other in Currency modelling software: Neuron:
Pairs (Valutap?rok): eg. USD?EUR, GBP?CHF, (free,
JPY?EUR, etc. There are only brokers GUI, Win XP install, Tutorial in HTML)
called FOREX companies/providers trading Genesis:
with someone else’s money, who want to (free,
hedge, arbitrage or speculate. GUI, Win XP install, Tutorial in HTML)
11Fractal Theory of Stock/Currency Pair PDP++:
Prices. Both at Stock Exchange/FOREX there
is strong Information Asimmetry P++.html (C++ source code library, GUI,
(Inform?ci? aszimmetria): most investors Win XP install, Tutorial in HTML).
do not have any direct information about: 23References 3. Decision support
Changing technology level and marketing software: JNNS:
efficiency of a company (denoted with
green) Plans of Governments (Korm?ny) and oftware/JavaNNS (Simplified SNNS in Java,
Central Banks (K?zponti Bank) of 2 GUI, Win XP install, Tutorial in PDF)
countries determining at most price of a JOONE: (Java, GUI,
given stock/currency pair long term They Win XP install, Tutorial in PDF)
have to decide allocation of their money Commercial neural decision support
among stocks/currencies from partial software: NeuroSolutions:
information and their expectations, so
they tend to fall in selling/buying panic (60 days shareware, no save, GUI, Win XP
at sudden big changes. Therefore, both install, Excel Add-in, Excel Wizard,
Stock Exchange/FOREX are strictly MATLAB modul, Tutorial in PDF Medical,
controlled markets with many safety rules. automotive appliacations) NeurOK:
But this will result in a Stepped
(L?pcs?s) price (denoted with red) update (Excel Add-in, C forr?sk?d, XML-es
behavior: Without strong external impulse fel?let, Win XP install, financial
brokers tend to build „dream worlds” applications) EasyNN:
setting up prices by their expectations (30 days
ignoring slow and small changes of reality shareware, GUI, Win XP install, Tutorial
(eg. In „.com boom” of 2000s, small in HTML, financial forecasting
internet-based companies were worth more applications) ALNFit Pro:
than General Electric and other industrial
giants) But when the difference between .shtml (30 days shareware, GUI, Win XP
them gets to big, they update price in install, Tutorial in PDF, p?nz?gyi
smaller-bigger sudden steps, instead of el?rejelz?si applications) Trajan:
continous change As prices are influenced
by many different lenght cycles (eg. 1 loads.htm (30 days shareware, GUI, Win XP
year:seasons .. 1day:daily close), sudden install, Tutorial in HTML, no real
steps are Embedded (Be?gyaz) into each application) AINet:
other at several levels, it creates (1
Fractal (Frakt?l)-type structures: price days shareware, GUI, Win 95 install,
steps in time have self-similar details Tutorial in PDF, nincs m?g val?s
embedded into each other It makes Price alkalmaz?sa) NeNet:
Forecasting (?rel?rejelz?s) necessary for
trading extremely difficult Function wnload.html (performance limited
Estimation (F?ggv?ny becsl?si) problem: shareware, GUI, Win 95 install, Tutorial
Prices of stocks/currency pairs are in HTML, SOM networks oriented) Add-Ons
influenced by numerous parameters creating for Statistical Packages: Statistica
complex multivariate (Sokv?l- toz?s) Neural Networks:
functions Price data is very Noisy (Zajos)
dis- torted by random disturbances, so nce/download.html (no shareware, GUI, Win
Sto- chastic (Sztochasztikus) function XP install, Tutorial in MPEG).
esti- mation is necessary from a Sample 24References 4. Add-Ons for MATLAB:
(Minta) of prices Sometimes it is hard to Matlab Neural Toolbox:
assemble any function from future price
expecta- tions collected from different / (No shareware) SOM ToolBox:
informati- on sources: spot price( ) can
adapt to reality in more alternative ownload/ (Matlab 5, free, GUI, Tutorial in
fractal path. PDF) FastICA:
12Basic terms of Distribution Free
Estimatiors (DFE). Distribution Free code/dlcode.shtml (Matlab 7, free, GUI,
Estimators (Eloszl?sf?ggetlen becsl?si Tutorial in PDF) NetLab:
rendszer) can estimate output of a
complex, multivariate function from (Matlab 5, free, GUI, Tutorial in PDF)
inputs. Functional transformation is NNSysID:
estimated from previously observed
(Megfigyelt) Sample (Minta) of noisy sid.html (Matlab 7, free, GUI, Tutorial in
input-output values, and it does not make PDF) Excel Add-Ins in Financial
any assumptions on Probability Forecasting: NeuroShell:
Distribution (Val?sz?n?s?gi Eloszl?s) of (no shareware)
sample. It means that the function can be NeuroXL: (no
reasonably complex. There are Analytic shareware) Comparison of 50 commercial
(F?ggv?nytani) methods of Approxi-mating licensed neural software:
(K?zel?t) complex functions: Taylor Series
(Taylor-sor): it approximates a non-linear ural_network_tools.htm.
Programming Neural Networks and Fuzzy Systems in FOREX Trading.ppt
cсылка на страницу

Programming Neural Networks and Fuzzy Systems in FOREX Trading

другие презентации на тему «Programming Neural Networks and Fuzzy Systems in FOREX Trading»

«ООО предприятие Магистраль» - Организационная структура Общества. Заключение. Расчетные показатели скоринговой модели. Оценка деятельности предприятия на примере ООО «Магистраль». Цель работы ? изучение теоретических аспектов антикризисного регулирования современного предприятия на примере ООО «Магистраль». ООО «Магистраль» можно отнести к категории предприятий со средним финансовым состоянием, слабость отдельных финансовых показателей, нормальная финансовая устойчивость, близкое к оптимальному, достаточно платежеспособно и рентабельно.

«Тормозная жидкость» - 4. Разработка компании Тосол-Синтез. Характеристики покупателей новой тормозной жидкости ROSDOT. Фирменная транспортная коробка фасовка продукта 455 и 910 гр. 9. Дорожные испытания продолжение. Мультибренды. Ценовое позиционирование тормозных жидкостей dot4. 2. Новые требования к тормозным жидкостям для систем ABS.

«Производство гранул» - Сушка и производство гранул. От 500 до 1500 кв.М. И выше достаточно навеса высота 6-8 м. Биотопливные заводы. Стандарты – не более чем база для начала переговоров. Сравнение ценовых характеристик энергоносителей. Выбор технологии и конфигурация оборудования зависят от следующих факторов: Переменные затраты.

«ЮБиКей групп» - Креатив. Дизайн. Адресная программа. На табличках обыгрываются копирайты, вызывающие неоднозначные, но сильные эмоции. Чем все кончилось. Магазин. Рекламная компания ЮБиКей групп. УМЕЛЕЦ – магазин бензо- и электроинструмента. Копирайты. BTL – кампания по открытию нового магазина. Механика.

«Развитие компании» - Дальнейший рост прибыльности обществ электросвязи. Внедрение разработанной целевой операционной архитектуры МРК. Корпоративная реструктуризация – важный этап реорганизации отрасли. Выгоды. Результаты работы Связьинвеста в 2002 году. Финансово-экономическая служба. Реформа системы управления. Централизация финансовых функций в МРК.

«Capstone» - Grupo Plaza (Аргентина). Номинальный Электрический КПД С65. Выпущен демонстрационный образец класса 3/4* грузовой автомобиль с турбиной С30. Тяговый двигатель. Экобус – экологичный транспорт. CARB автобусы. Сертификация. * - Класс 3/4 - класс грузовых автомобилей с полной массой 4,5 тонн до 7,3 тонн.


40 презентаций о компаниях


125 тем
Картинки > Презентации по экономике > Компании > Programming Neural Networks and Fuzzy Systems in FOREX Trading