Correct Score - Odd probability

[KThom]

Hi all,
I have been working on a system/ database that produces a projected scoreline based on several factors.

Not wanting to go too much into the calculations, but basically using previous h2h data (weighted in favour of most recent meeting) using current form (weighted in favour of most recent, and also overall league trends (low scoring, high scoring, draw percentages etc).

This has obviously been a major project, and one that has taken several months to perfect. I actually use the score projection side of things to place over / under goal bets with quite success.

I have a spreadsheet with over 7750 games worth of data (I am not knowing how to upload a sheet because it says I can not post attachments? I have tried to paste, I will see how it looks). The games range over the past 3 seasons and take into account a variety of leagues (French 1&2, All 4 English pro leagues, Belgian, Dutch, German 1&2, Turkish, Spanish, Italian, Swedish) – You get the idea.

I do not factor in the first 6, or last 3 score projections from any league as these tend to produce inaccurate readings – particularly the first 6 games where a newly promoted or relegated team has little or no h2h stats or form.

I have converted the number of projected scores into an odds probability to see if I can find value betting on a fixed score line.

Taking 0-0 for example, the database generated 288 0-0 predictions, of which 39 where correct to which I calculate a percentage odds value of 7.38. (1/(39/288)).

This is where my mind begins to ache. Unfortunately I don’t have, (and there is no way of getting) correct score odds for all of these games. Using “average” correct score odds is dangerous as they obviously vary between game to game, and even between bookie to bookie.

So I do not know, that if I had back every 0-0 projection at odds of 7.38 or above I would have been profitable. It could have been that of the 39 correct 0-0 projections, the actual odds of a 0-0 were much lower. I also do not know that if I had back all of the projected 0-0’s (288) at average odds above 7.38 I would have at least broken even – again, I do not know what odds were available for all of these games.

I guess, the back testing stops before it begins as far as working out probable profit / loss, all that can be done is to take the system forwards.

The big question is, how do I use the data to make bets/ betting profitable?

I think I rule out any projected scoreline with less than 5%? worth of data? Is there a test that can be done to calculate how reliable my data is based on my overall sample size (7794 games)?
If I use the 5% rule, any projected scoreline with less than 390 predictions can be ruled out as “insufficient data”. Gain, I am guessing at 5% and 390 games. Perhaps someone can tell me a more robust way to give data accuracy. (P and T testing I know some about but not my string point).

Using the highest projected scoreline (1-1) it generates an overall probability of 7.10 on an actual 1-1 draw. Again, as above with the 0-0, I can not back test results against actual odds as I do not have this information. If I assume that the 1-1 projection of 7.10 is fairly robust (nearly 1,300 projections), what should I be looking for going forward in projections?

There are after all 2 1-1 projections in next weekends Norwegian league ( Tromso vs Sandefjord and Lillestrom vs Fredrikstad). Should I place for a 1-1 regardless of odds? Should I only place if I can obtain 7.10 or over? This is the part that I need to convert past probability – into future chance.

I hope you are still reading, any thoughts?

											Actual
Pred	0-0	0-1	0-2	0-3	0-4	1-0	1-1	1-2	1-3	1-4	2-0	2-1	2-2	2-3	2-4	3-0	3-1	3-2	3-3	4-0	4-1	4-2	4-3	5-0	5-1	Games
0-0	7.38	14.40	19.20	57.60	288.00	7.38	8.23	18.00	32.00	288.00	11.08	18.00	16.94	57.60	288.00	41.14	36.00	36.00		96.00	72.00	96.00	288.00	288.00	288.00	288
0-1	9.65	9.25	19.30	26.12	148.00	8.88	8.54	12.69	34.15	148.00	13.88	13.06	23.37	63.43	148.00	40.36	27.75	40.36	88.80	148.00	111.00			222.00	148.00	444
0-2	15.67	9.17	10.16	47.00	125.33	11.06	11.39	9.89	23.50	75.20	14.46	12.13	26.86	28.92	53.71	37.60	34.18	62.67	376.00	94.00	125.33	188.00		376.00	376.00	376
0-3	38.00	5.43	12.67	9.50	38.00	38.00	12.67	4.75	19.00			12.67	19.00													38
0-4		4.00	2.00											4.00												4
1-0	12.12	19.69	24.23	67.50	315.00	8.22	7.21	17.83	33.75	94.50	11.12	10.50	21.98	63.00	315.00	21.98	18.90	41.09	157.50	105.00	45.00	118.13	472.50	236.25	105.00	945
1-1	10.68	11.23	21.53	44.55	258.40	8.13	7.10	14.20	34.92	129.20	13.89	12.07	21.18	92.29	258.40	29.36	24.85	51.68	86.13	184.57	61.52	99.38	646.00	646.00	184.57	1292
1-2	10.40	9.27	17.33	33.43	104.00	10.52	7.93	14.18	31.20	133.71	17.33	11.85	22.83	55.06	133.71	28.36	26.00	62.40	85.09	133.71	104.00	117.00	312.00	936.00	234.00	936
1-3	8.25	9.43	8.25	33.00		22.00	11.00	9.43		33.00	13.20	22.00	11.00	66.00		66.00	33.00		66.00			66.00	66.00			66
1-4							3.00		3.00								3.00									3
2-0	12.91	14.88	36.58	292.67	219.50	8.36	7.26	24.39	79.82	219.50	10.58	9.44	24.39	97.56	878.00	15.14	22.51	35.12	125.43	30.28	39.91	125.43	125.43	219.50	175.60	878
2-1	11.24	14.09	29.50	104.89	314.67	9.08	7.49	18.51	49.68	118.00	9.94	11.51	18.88	62.93	236.00	18.88	23.02	47.20	72.62	59.00	62.93	157.33	472.00	134.86	118.00	944
2-2	12.86	12.86	17.73	65.60	656.00	8.75	7.72	14.58	38.59	131.20	11.31	13.67	15.26	65.60	131.20	32.80	21.87	54.67	72.89	72.89	72.89	131.20	328.00	109.33	656.00	656
2-3	16.50	16.50	16.50		33.00	16.50	8.25		16.50		16.50	8.25	16.50	11.00			16.50	16.50				33.00	33.00			33
2-4									1.00																	1
3-0	11.45	36.89	166.00	110.67		7.55	12.77	20.75	166.00		5.63	8.74	30.18	83.00	166.00	15.81	23.71	47.43	110.67	18.44	25.54	166.00	332.00	166.00	166.00	332
3-1	10.78	14.55	72.75	97.00		9.09	8.56	22.38	41.57	145.50	9.09	7.86	36.38	97.00		29.10	14.55	18.19	58.20	145.50	58.20	145.50	291.00		72.75	291
3-2	24.71	17.30	21.63	34.60		7.21	9.61	15.73	57.67		8.65	15.73	19.22	28.83	173.00	24.71	24.71	21.63	86.50	57.67	57.67	57.67	86.50	86.50	173.00	173
3-3		12.00	12.00			12.00	12.00	12.00	12.00		12.00	12.00								12.00	12.00	12.00	12.00			12
4-0	7.75	15.50				10.33	31.00	15.50	31.00		10.33	15.50				6.20	15.50	31.00		10.33				31.00		31
4-1	14.00	28.00				9.33	9.33	28.00			9.33	14.00	9.33			9.33	28.00	28.00	28.00					14.00		28
4-2	15.00	15.00					15.00	5.00	15.00		5.00			15.00		15.00			15.00		15.00					15
4-3											3.00	3.00										3.00				3
5-0																								2.00		2
5-1			3.00								3.00														3.00	3
																										7794

[MetallicIan]

Hi KThom,

Looking at your chart and personally speaking I think that your figures look very good.

From the 288 games sample for example, your 0-0's look to have a very high accuracy. I would think that even betting "blind" at odds of 7.38 or above on 0-0's that those predictions would be profitable.

It looks like the hit rate for all your predictions down the diagonal are exceptional, as I would expect the average odds for all the corresponding odds to be higher.

[KThom]

Hello metalician, thanks for the feedback - appreciated

An unexpected problem has occured before I begin. My work pc now blocks this site under the terms "gambling" whereas It didn't yesterday!! Has anything in the site make up changed? Or perhaps an eagle eye in my IT dept?
Either way, my only means of posting is via my iPhone - so apologies for any text typos. This also makes it more difficult fir me to upload tables or images

Anyway I digress

To go live with this system I would indeed be looking to bet on a projected scores actual outcome. For instance a projected 1-0, I would bet on a 1-0 - even though 1-1 shows a greater hit rate. I will therefore be using the diagonal boxed data only

As mentioned before I would like to discover a correct way of measuring the accuracy of my sample sizes based on the overall amount of games. I will keep searching the net as I know I have stumbled across this formula somewhere before

It will basically tell me how truly accurate 945 selections out of a total (population) of 7794 games is - using 1-1 as the example here. So anyone reading this that can help with this formula I'd be greatful

I currently have 4 live leagues in play that I can post selections for over the summer. They are swedish Norwegian Brazilian and Japanese top flight leagues I am reluctant to include any leagues with less than 16 teams as this tends to restrict the parameters I set. I let you know that i know that I know nthing about any of these leagues which means I can't be objective about any of my selections - they will be based on data driven selections

I have discover the ansell staking plan on another thread and this is exactly the type of staking formula I have been looking for. It will not allow you to stake on a selection with odds lower than my projections


I have to say now that there was a slight errror on the table above as I unintentionally doubled up a few leagues. It doesn't affec the figures greatly and I will post the "true" odds for the selections before hand

So befre I start posting selections I need to resolve some factors (not least an Internet acess).

1) I really need to establish a formula that shows how accurate my sample sizes are fir each projected scoreline, and ideally how accurate the odds are within that projection

2) once this is established I can chose which scorelne(s) I want to go live with

Hopefully I can calculate my accuracy and start posting selections in time for next weekends games

[KThom]

Further to my last I have now re-discovered the sampling website. It's www.surveysystem.com/sscalc.htm#one

I now need to try and get my head around what data I need to pump in

[KThom]

Ok, I have managed to log onto a relatives PC - and although it may seem (and it feels) like I am talking to myself I am sure many of you are reading to see what happens when i "go live".

Before I do, I would still like clarity over the sample size calculation method, so if anyone can assist with regards to the correct formula(s)??

For now, I have uploaded the new table. At the bottom I have added 2 rows. The first is the odds (as a decimal) based on betting blind on EVERY game for each respective scoreline. Below that odd, is a percentage. This represents the difference between my projected odds, and the actual probability for each scoreline.

It shows that 0-0, 0-2 and 3-1 have the greatest difference, so I am thinking about using these 3 scores for the live leagues.

There are now games this weekend due to the 4 leagues either being on break or not reaching game 6 stage yet.

I will hopefully have selections ready for next weekend, but I really want to get the criteria fixed BEFORE I start posting selections. I have already determined the Ansell staking plan, I just need clarification over the sample size (at the moment I am working with 230 games plus (3.5%).

So if anyone is avidly reading, but disagreeing, or has alternate views or comments on my data please feel free to suggest before I attempt to go live. It may well be that I have made a gross oversight, or made an inaccurate calculation. - but for now, I have attached the new table.

Here's hoping we can achieve a long term profitable correct scoring system. Patience is most definitely the key because the strike rate will be a lot lower than other "normal" backing systems. I fully expect games and weeks of no winners, the key is that when we do hit a winner it more than pays off.

[KThom]

Ok, I have used the Ladbrokes correct score vs match odds table to give me a rough understanding of the actual odds for each of the games observed in my table.

Of course, using this table only gives a vague idea of what correct score odds were available, but its better than nothing.

Below is a table of data with the AVERAGE odds for each scoreline.

The headers are explained as the following:

Score – This is the actual outcome of the game
Proj – This is the projected odds from my table, using each projected score vs actual (Projected 0-0 vs actual 0-0 etc)
Act 1 – This is the average actual obtainable odds for the projected games. I have used the match odds to lookup the correct score odds using a Ladbrokes domestic correct score win odds sheet.
Tot – This is the total odds for each scoreline assuming EVERY game (6,581) was bet upon that particular scoreline occurring
Act 2 – This is the average odds for all games ending in each particular scoreline (Again the Ladbrokes scoresheet was used to generate probable odds)
% - This is the percentage of projected scores for all games. I have used 3.5% as a cut off – assuming anything below 230 games is insufficient data (I still would like to find the “official” way to test the accuracy of each scoreline as a sample size)

Score	Proj	Act 1	Tot	Act 2	%
0-0	7.03	9.00	11.73	9.67	3.63%
0-1	9.07	8.18	12.66	8.62	6.06%
0-2	13.17	9.32	22.16	13.36	4.60%
0-3					0.47%
0-4					0.05%
1-0	7.87	7.59	8.61	7.77	12.43%
1-1	7.34	6.57	7.85	6.76	17.50%
1-2	13.75	10.24	15.56	12.49	11.70%
1-3					0.79%
1-4					0.03%
2-0	10.66	7.81	11.75	9.07	11.34%
2-1	10.89	8.29	11.39	9.01	11.91%
2-2	14.73	15.57	20.83	16.31	8.28%
2-3					0.36%
2-4					0.02%
3-0	17.06	10.25	23.25	17.36	4.15%
3-1	14.44	13.00	23.59	17.53	3.51%
3-2					1.99%
3-3					0.18%
4-0					0.40%
4-1					0.36%
4-2					0.17%
4-3					0.03%
5-0					0.02%
5-1					0.02%

From this data I can conclude that only 1 projected scoreline shows a greater chance of hitting, than the odds provided.

The 0-0.

0-0shows a 14.23% chance of hitting, with an average payout of 9.00. (giving a 42.3 point profit) over 100 level bets.

Of course this data is raw, but it is the best I have to work with.

I will start posting any projected 0-0’s from this weekend onwards. I will also be using the Ansell staking plan, and will only post selections once I have a matched bet. I will then post with which bookmaker or exchange I have placed the bet.

I may now focus on trying to calculate any “lay” betting opportunities based on the opposite of the above.

Again, any help with T testing, Chi Squared or any other sample reliability test to see how robust my data is would be much appreciated.

[MetallicIan]

KThom,

It's well known that the overround on correct scores at traditional bookmakers like Ladbrokes are extremely large and even if you have an edge would be unlikely to make a profit in the long term (In my opinion).

I still think however that if you were to use your system on Betfair that you would find that your expected odds are below those usually seen there.

Unfortunately there isn't an easy way to backtest your system using Betfair, experience has taught me that even using betfairs data download site that the information is not formatted particularly well and difficult to extract the exact information you need quickly enough.

I would suggest that you paper test your system using Betfair CS odds (remembering to deduct commission from any winning bets).
It may seem a bit labourious but could well be worth the effort financially.


As an aside and maybe a topic to start another thread, but it has occurred to me before now that if there are enough people interested on here in helping out trialling someones system, that it may save time and effort on the part of an individual if people are willing to pool their time and resources either in data mining or paper testing certain matches allocated by the person trialling their new system idea.

The person wouldn't necessarily have to divulge any system criteria to those "helpers" but if successful they could be rewarded either by having the system criteria made known to them or free selections in the future.

We are Borg - You will be assimilated - Resistance is futile.

[KThom]

Hi again Ian, firstly I've somehow made an over sight and not included the 2-2 draw in surmising a profitable scoreline.

The vast majority of my correct score (and indeed over / under scores) were placed on betfair for the reasons you have given. But for analysis purpose it is not easy to back test unless using standard bookies odds.

I will use the projected odds as my guide going forward (for 0-0 and 2-2) and be vigilant over where the best prices lay.

I have had some tremendous help on Chi squared and T testing and am in the process of running my results through the mill. I will publish any findings before the weekends games.

I think your pooling idea is a good one, fortunately the actual data collecting isn't that much of an issue for me as I can do this mostly whilst in "work"

[KThom]

Ok, as we approach the weekend, and the re-commencement of some of the leagues its about time I “firm up” my selections and staking process.

I have been busy running each scoreline through the Chi Test and T Tests to check robustability and yield outcome. I have also fed in games played to date in the Japanese, Swedish and Norwegian leagues to bolster the data.

My findings are published in the table below.

Predicted scoreline	The break even odds projected by betting on each scoreline	Actual average odds available for these predictions	The actual likelihood of this scoreline hitting in general	Overall average odds for ALL games that produced this scoreline	Frequency of hitting based on all predictions being bet upon	Assumed frequency of hitting based on average odds available	Overall frequency of hitting based upon ALL games being bet upon	Assumed probability of this scoreline happening	Shows the difference between implied probability and actual as a percentage	The percentage probability that the results were NOT down to chance	The probability of achieving a negative yield	The achieved yield	The percentage of each prediction from ALL games
Score	Proj	Act 1	Tot	Act 2	Proj Cha	Act Cha	Tot Cha	Act Cha	Gain	Chi Sq	t Test	Yield	% of All
0-0	7.03	8.97	11.73	9.28	14.23%	11.15%	8.52%	10.78%	3.08%	96.22%	8.87%	27.62%	3.63%
0-1	9.07	8.14	12.63	8.60	11.03%	12.29%	7.92%	11.63%	-1.26%	6.53%	78.83%	-10.28%	6.06%
0-2	13.17	8.83	22.19	13.00	7.59%	11.33%	4.51%	7.69%	-3.74%	77.43%	98.88%	-33.00%	4.60%
0-3	6.75	8.00	59.18	26.26	14.81%	12.50%	1.69%	3.81%	2.31%	86.73%	36.88%	18.52%	0.47%
0-4	0.00	0.00	234.61	18.82	0.00%	0.00%	0.43%	5.31%	0.00%	26.12%	95.84%	-100.00%	0.05%
1-0	7.86	7.52	8.61	7.72	12.73%	13.30%	11.62%	12.95%	-0.57%	61.31%	68.70%	-4.28%	12.43%
1-1	7.32	6.54	7.85	6.71	13.65%	15.29%	12.74%	14.91%	-1.64%	5.69%	94.74%	-10.74%	17.50%
1-2	13.75	10.00	15.53	12.39	7.27%	10.00%	6.44%	8.07%	-2.73%	87.70%	99.79%	-27.27%	11.70%
1-3	0.00	0.00	39.34	26.41	0.00%	0.00%	2.54%	3.79%	0.00%	91.23%	100.00%	-100.00%	0.79%
1-4	0.00	0.00	142.80	39.09	0.00%	0.00%	0.70%	2.56%	0.00%	21.73%	88.97%	-100.00%	0.03%
2-0	10.66	7.50	11.75	8.70	9.38%	13.33%	8.51%	11.49%	-3.95%	96.54%	99.99%	-29.62%	11.34%
2-1	10.85	8.28	11.37	8.84	9.22%	12.08%	8.80%	11.32%	-2.86%	82.47%	99.70%	-23.69%	11.91%
2-2	14.68	15.27	20.79	15.98	6.81%	6.55%	4.81%	6.26%	0.27%	69.99%	40.32%	4.05%	8.28%
2-3	12.00	46.50	61.39	32.23	8.33%	2.15%	1.63%	3.10%	6.18%	86.94%	15.98%	287.50%	0.36%
2-4	0.00	0.00	193.21	30.71	0.00%	0.00%	0.52%	3.26%	0.00%	N/A	N/A	N/A	0.02%
3-0	17.06	9.56	23.29	16.25	5.86%	10.46%	4.29%	6.15%	-4.60%	93.83%	99.91%	-43.96%	4.15%
3-1	14.44	12.19	23.54	16.42	6.93%	8.21%	4.25%	6.09%	-1.28%	6.84%	77.35%	-15.58%	3.51%
3-2	26.20	26.00	43.50	26.97	3.82%	3.85%	2.30%	3.71%	-0.03%	22.99%	50.70%	-0.76%	1.99%
3-3	0.00	0.00	92.52	54.65	0.00%	0.00%	1.08%	1.83%	0.00%	34.25%	100.00%	-100.00%	0.18%
4-0	8.67	22.33	69.88	29.97	11.54%	4.48%	1.43%	3.34%	7.06%	82.08%	15.36%	157.69%	0.40%
4-1	0.00	0.00	60.82	32.24	0.00%	0.00%	1.64%	3.10%	0.00%	67.56%	100.00%	-100.00%	0.36%
4-2	0.00	0.00	136.85	44.25	0.00%	0.00%	0.73%	2.26%	0.00%	34.61%	100.00%	-100.00%	0.17%
4-3	0.00	0.00	345.74	0.00	0.00%	0.00%	0.29%	0.00%	0.00%	N/A	N/A	N/A	0.03%
5-0	1.00	15.00	205.28	28.47	100.00%	6.67%	0.49%	3.51%	93.33%	N/A	N/A	N/A	0.02%
5-1	0.00	0.00	168.44	25.97	0.00%	0.00%	0.59%	3.85%	0.00%	N/A	N/A	N/A	0.02%

My only concern (and still untested theory) is the sample size for each prediction, and that of the overall selections.

I have used a cut off of 3.5% - so any game with less than this amount of data is excluded from the selections process.

From here I am left with 11 scorelines. Of which only 2 show a positive yield from the t-testing. (I have used the average Ladbrokes odds for this testing, so better odds may well have been – and will be available). It is always better to “play safe” when it comes to testing your models anyway.
The Chi sq or Chitest shows the percentage chance that the results for each selection were NOT down to luck. Anything above 95% is an acceptable rate for this kind of testing. Again I have used the average Ladbrokes prices and even divided them down by 12.5% to take into account their overround. So this is clearly a safety first test.
From this testing you can see only 1 scoreline (the 0-0) is above the recommended Chi test level. It also shows a t test score of 8.87% which is higher than I would like, but by far the best score of all my selections. I am confident that by “shopping” for the best prices I can overcome many of the small questions around the 0-0 reliability. After all, finding the best price is a big chunk of what aids a winning system.

My only concern is still that my sample size for 0-0’s is too small (less than 250 out of a total of over 6,500 games). But as I have not found a definitive way of testing this particular aspect I will have to play on for now.

As for staking, I will be placing these bets with money myself. But for the benefit of system testing I will assume a bank of 500 points. I will be using the Ansell staking method to further enhance my (hopeful) profitability with a decided confidence level of 4.
Rules for bet placing: I will not match any bet under the current break even point + 10% (currently 7.73).
I will only post my selections once I have a match, to protect any possibly market liquidity shift. If there was no match, I will identify this after kick off.
I will use the best bookmaker available to me (this may not necessarily mean it is THE best price obtainable on the net)
I will only bet on games identified as a potential 0-0 by my system. (currently 4 leagues in play).

To let you know, there are no selections identified for this weekend in either of the qualifying leagues (Norway or Japan). I will aim to identify games as soon as I have a match, and will only post un-matched games after kick off – these will obviously not be included in my staking results. With a projected hit rate of around 14.2% I am obviously aware that this may provide long weeks of no winnings – but hopefully the testing will go well over the summer and I can then get stuck back into 20+ leagues in September time which will obviously bring more games to chose.

[KThom]

Ok, first selection this summer to be published.

Just a quick re-cap and definition for the rules of the 0-0 system.

*All odds must be at 10/1 or higher (11.00)

*There will be no trading in running - so a game is a loss if it loses 1-0 in the 94th minute the same as if it ends 6-4

*I am aiming for around an 11% hit rate, and a yield rate of around 17.5%

*I am using the Ansell staking plan, with a confidence level of 4, and a 500 point starting bank

*I will post a running bank after each games outcome.

Game 1

Date	Home	Away	League	Home Pred	Away Pred	Stake	Odds	Bookmaker
21-Jun-09	Corinthians	São Paulo	Brazil L1	0	0	13.24	14.5	William Hill

[MetallicIan]

Good Luck KThom

[KThom]

Latest stats:

Bets	1
Wins	0
Win rate	0.00%
Staked	13.24
Profit + / -	-13.24
Yield	-100.00%
Avg Odds	14.50
Starting Bank	500.00
Current Bank	486.76

Game 2:

24-Jun-09

Tromso IL

Viking Stavanger

Norway L1

0

0

8.69

13

Sky Bet

[KThom]

I thought game 2 was the one for a long time

Still, here we go with latest results

Bets	2
Wins	0
Win rate	0.00%
Staked	21.93
Profit + / -	-21.93
Yield	-100.00%
Avg Odds	13.75
Starting Bank	500.00
Current Bank	478.07

Game 3:

28-Jun-09

Sport Recife

Grêmio

Brazil L1

0

0

13.24

14.50

William Hill

[muppet77]

What was the result bud?

[KThom]

0-1, still a loss is a loss

[muppet77]

Unlucky. And match one? Keep up the effort.

[KThom]

Sorry, match 1 was 3-1 so a non starter although did take 38 mins for a goal.

I will endeavor to post scores now also as not everyone checks on these leagues.

Thanks for the wishes, I hope to hit one soon

[KThom]

Game 3 ends 3-1

Latest results

Bets	3
Wins	0
Win rate	0.00%
Staked	51.61
Profit + / -	-51.61
Yield	-100.00%
Avg Odds	13.3
Starting Bank	500.00
Current Bank	448.39

Game 4:

04-Jul-09

Yamagata

Urawa

Japan L1

0

0

3.2

10.00

Bet 365

Game 5:

05-Jul-09

Náutico

Internacional

Brazil L1

0

0

13.24

14.50

William Hill

[KThom]

Game 4: 2-3
Game 5: 0-2

No selections for this weekend as yet - will post up if anything changes. Still after the illusive first!!

Bets	5
Wins	0
Win rate	0.00%
Staked	51.61
Profit + / -	-51.61
Yield	-100.00%
Avg Odds	13.3
Starting Bank	500.00
Current Bank	448.39

[KThom]

There is one selection for Sunday - I'm awaiting a match on Bet fair so will post once matched.