NeuroTennis

[ZaletoFF]

Greetings to all.

I present created by me for forecasting tennis a neural network.
Money-Management: Fixed bet (1 %).
Bookmaker: PinnacleSports.
Starting bank: 10000$.
Format of bets: Tour_Name...Players...Probabilities...Bet...Coef.. .Summ

I shall tell a little about my neural network. For training it I used the data more than for 9 years for WTA and 6 years for ATP. Unfortunately (it is usual for any neural network) I can name only small part of input parameters, was age, the birthplace, a place of a tournament, surface, age, any statistics... I cannot name other parameters...it is normal for any neural network... I use Kelly-Criteria for placing value bets.

The previous results are accessible here.


Previous results:

So, it is made more than 1333 bets in sum WTA+ATP.
Let's look at statistics from the moment of the beginning of experiment on BA.

OVERALL.
Starting bank: 10000$
Current bank: 15042.9$
Total staked: 133900$
Number of bets: 1339
Winning bets: 772
Losing bets: 503
Return bets: 64
Avr. wins odds: 1.709
Strike rate: 60.55%
Yield: +3.77%
Profit: +5042.9$

May: +1885,90$
June: +1708,40$
July: +1068,90$
August: +2194,50$
September: +497,10$
October: -2411,40$
November: +99,5$

Thanks to all, who visit my topc, next season is being better!

Last update 19.11.2006

You have become interested in mine NeuralNetwork? You have offers for me?
ICQ: 397-9-793

[ZaletoFF]

US Open fou...(1) Amelie Mauresmo (FRA) - (WC) Serena Williams (USA)...{25}...{75}...[2] ...1,735...1

US Open fou...(23) Anna Chakvetadze (RUS) - (27) Tatiana Golovin (FRA)...{51}...{49}...[1] ...2,1...1

US Open fou...(17) Andy Murray (GBR) - (7) Nikolay Davydenko (RUS)...{67}...{33}...[1] ...1,84...1

US Open fou...(15) Lleyton Hewitt (AUS) - (25) Richard Gasquet (FRA)...{59}...{41}...[1] ...2...1

[ZaletoFF]

US Open fou...(12) Tomas Berdych (CZE) - (5) James Blake (USA)...{52}...{48}...[1] ...2,5...1

[Mr Intensity]

to PL and good luck mate

[ZaletoFF]

Originally Posted by Mr Intensity

to PL and good luck mate

Sure, thanks.

[ZaletoFF]

US Open fou...(7) Patty Schnyder (SUI) - (10) Lindsay Davenport (USA)...{31}...{69}...[2] ...1,549...1

[ZaletoFF]

Stats will be updated every 100 bets, or mounthly.

[libi71]

Good luck with your system.

It would be helpful if you could post the results regularly (i.e. which bets were won/lost). Would make the thread more interesting.

Updating stats once a month is ok IMHO.

[bez007]

sorry new to tennis betting and i dont understand your tips, can you explain how to read your selections please.

thank you and good luck

dave

[ZaletoFF]

libi71, thank you

I shall update win/lost record after ending of each round. Also, after ending of each round I will show: won bets, lost bets, void bets, avr. coef., profit. I shall update stats, graphs after each 100 bets or mounthly (detailed analysis)...

Thanks for your opinion

[ZaletoFF]

Originally Posted by bez007

sorry new to tennis betting and i dont understand your tips, can you explain how to read your selections please.

thank you and good luck

dave

bez007, yes, I can. Look an example.

ForestHills 12...Anna Smashnova (ISR) - Maria-Elena Camerin (ITA)...{56}...{44}...[1] ...1,98...1

"ForestHills 12..." - place of tour and round (in this example round is 1/2).
"Anna Smashnova (ISR) - Maria-Elena Camerin (ITA)..." - players.
"{56}...{44}..." - probabilities to win match by Anna Smashnova (ISR) and Maria-Elena Camerin (ITA), which my neural network show to me.
"[1] ..." - bet to first player (Anna Smashnova (ISR)). This selection is because if use Kelly Kriteria: 1.98*56/100-1=0.1088 > 0.
"1.98..." - PinnacleSports coef.
"1" - summ of stake (1% or 100$ for the time of experiment).

Thanks for your opinion

[bez007]

thank you, i'm now following your selections lets hope for some good wins

[ZaletoFF]

US Open 14...(19) Jelena Jankovic (SCG) - (4) Elena Dementieva (RUS)...{28}...{72}...[2] ...1,775...1

[ZaletoFF]

US Open sec...(8) Marcos Baghdatis (CYP) - Andre Agassi (USA)...{39}...{61}...[2] ...2,51...1...1,51
US Open sec...Mikhail Youzhny (RUS) - Nicolas Massu (CHI)...{48}...{52}...[2] ...2,24...1...-1
US Open sec...Jiri Novak (CZE) - Paradorn Srichaphan (THA)...{57}...{43}...[1] ...2,68...1...1,68
US Open sec...Vince Spadea (USA) - (29) Jonas Bjorkman (SWE)...{62}...{38}...[1] ...1,671...1...0,67
US Open sec...(21) Gaston Gaudio (ARG) - (WC) Samuel Querrey (USA)...{68}...{32}...[1] ...2,18...1...1,18
US Open sec...Carlos Moya (ESP) - Paul-Henri Mathieu (FRA)...{60}...{40}...[1] ...1,676...1...0,68
US Open sec...(4) David Nalbandian (ARG) - Marat Safin (RUS)...{44}...{56}...[2] ...2,22...1...1,22
US Open sec...Alessio di Mauro (ITA) - (17) Andy Murray (GBR)...{6}...{94}...[2] ...1,074...1...0,07
US Open sec...(q) Lukasz Kubot (POL) - (q) Noam Okun (ISR)...{17}...{83}...[2] ...1,8...1...-1
US Open thi...Andre Agassi (USA) - (q) Benjamin Becker (GER)...{77}...{23}...[1] ...1,315...1...-1
US Open thi...(15) Lleyton Hewitt (AUS) - (20) Novak Djokovic (SCG)...{73}...{27}...[1] ...1,685...1...0,68
US Open thi...(12) Tomas Berdych (CZE) - (23) Dmitry Tursunov (RUS)...{57}...{43}...[1] ...1,769...1...0,77
US Open thi...(10) Fernando Gonzalez (CHI) - (17) Andy Murray (GBR)...{28}...{72}...[2] ...1,84...1...0,84
US Open thi...(21) Gaston Gaudio (ARG) - Marc Gicquel (FRA)...{62}...{38}...[1] ...1,621...1...-1
US Open sec...Aravane Rezai (FRA) - Lucie Safarova (CZE)...{62}...{38}...[1] ...2,89...1...1,89
US Open sec...(q) Varvara Lepchenko (UZB) - (26) Marion Bartoli (FRA)...{19}...{81}...[2] ...1,322...1...0,32
US Open sec...Jelena Kostanic (CRO) - (10) Lindsay Davenport (USA)...{4}...{96}...[2] ...1,08...1...0,08
US Open sec...Mara Santangelo (ITA) - Sybille Bammer (AUT)...{68}...{32}...[1] ...1,588...1...0,59
US Open sec...(17) Daniela Hantuchova (SVK) - (WC) Serena Williams (USA)...{17}...{83}...[2] ...1,242...1...0,24
US Open sec...Jarmila Gajdosova (SVK) - Viktoria Kutuzova (UKR)...{67}...{33}...[1] ...1,909...1...0,91
US Open sec...(q) Yulia Fedossova (FRA) - Kaia Kanepi (EST)...{54}...{46}...[1] ...2,2...1...-1
US Open sec...(24) Na Li (CHN) - Eleni Daniilidou (GRE)...{68}...{32}...[1] ...1,524...1...0,52
US Open sec...(27) Tatiana Golovin (FRA) - (q) Agnieszka Radwanska (POL)...{49}...{51}...[2] ...3,3...1...-1
US Open sec...Virginia Ruano Pascual (ESP) - (5) Nadia Petrova (RUS)...{14}...{86}...[2] ...1,217...1...0,22
US Open thi...(WC) Serena Williams (USA) - (16) Ana Ivanovic (SCG)...{63}...{37}...[1] ...1,667...1...0,67
US Open thi...Kaia Kanepi (EST) - Virginie Razzano (FRA)...{33}...{67}...[2] ...1,493...1...0,49
US Open thi...(24) Na Li (CHN) - (13) Mary Pierce (FRA)...{35}...{65}...[2] ...1,637...1...-1
US Open thi...(27) Tatiana Golovin (FRA) - (5) Nadia Petrova (RUS)...{26}...{74}...[2] ...1,532...1...-1
US Open thi...(6) Svetlana Kuznetsova (RUS) - Anastasia Rodionova (RUS)...{96}...{4}...[1] ...1,063...1...0,06
US Open thi...(19) Jelena Jankovic (SCG) - (9) Nicole Vaidisova (CZE)...{18}...{82}...[2] ...1,592...1...-1
US Open thi...(14) Francesca Schiavone (ITA) - (21) Shahar Peer (ISR)...{49}...{51}...[2] ...2,15...1...1,15

+366-215=30
Avr. coef 1.835
+3966.9$

[ZaletoFF]

Originally Posted by bez007

thank you, i'm now following your selections lets hope for some good wins

Yes

[itooam]

Looks like a good system - have been interested myself in developing a neural network (for horse/football) - had a look at "fann" a couple of months ago which would suit my development language of choice c#.net. What neural network software do you use?
Also, do you know of any good internet resources to help learn how to use them effectively I.e., deciding on appropriate range to use for variables etc? I have only read limited material on neural networks so any guidelines would be very helpful. Hope your success continues - I will be watching with interest.

[ZaletoFF]

In the Internet it is possible to find much neural packages. Input data for my neural network it is calculated in my program created by me in Delphi. Training and forecasting - in freeware russian neural package (this package is very simple to use).
There are no rules, which specify how to use input variables. Efficiency of a neural network can be estimated only experimentally...
Unfortunately I know few Internet - resources about neural networks, the best, in my opinion - http://www.calsci.com/ I received all knowledge from books...

Thank for good words!

[London_Calling]

Hi Zaleto, It's interesting but I'm a bit disillusioned at the moment with NN's.

In my limited experiments with Neural Networks, all I ever seem to identify is the market Fav.

This guy's system http://www.mymait.com/- despite his excitement - also only produces the market favourites.

What percentage of favourites does your network pick ?

[itooam]

Thanks for advice ZalettoFF - I will have to have a look on Amazon and get me some "NN for Dummies" books

London Calling - your system may be identifying favourites but have you looked at the general difference in odds to your neural network output probabilities (I am presuming it will return a floating point number as the success rate for each horse in the race - each one tested individually) and then success rate? You only need a very small edge to make a profit (that elusive value - whether fav or not). Though, this is how I presume you would use NN's and I don't wish to pretend to know much about them in the real world - only the case studies and examples I have read.

[itooam]

Just had a look at that link - shame he doesn't publish odds too? Success rate (in my opinion) is worth jack if the odds you are getting are less than the probabilites you have estimated.