2023 Anytime Goalscorer System

[harry_rag]

12 points on Vlahovic at 17/10 with Hills

[harry_rag]

6 from 14 since the last update (Sadiq, Olmo, Sterling, Fernandes, Emegha and Vlahovic).

68 winners from 191 bets so far for a profit of 227.55 points with an ROI of 11.67%.

30 points lost on 3 bets posted in error.

Current "target" ROI based on overall sample is 12.28% (now based on 578 bets from 3275 selections).

[harry_rag]

12 points on Dia (Salernitana) at 12/5 with Hills

[harry_rag]

12 points on Bailey and Kellyman (A Villa) at 2/1 and 11/5 with Hills

[harry_rag]

1 from 3 since the last update (Dia and Bailey).

70 winners from 194 bets so far for a profit of 268.35 points with an ROI of 13.51%.

30 points lost on 3 bets posted in error.

Current "target" ROI based on overall sample is 13.25% (now based on 581 bets from 3294 selections).

[harry_rag]

Been a bad month overall for the anytime goalscorer betting (singles anyway). I lost 330 points overall but it's fairly stark where the blame lies.

No issues with this system, monthly ROI was 14.92% so very much "as you were".

No real issue with the "studied" bets; there was a loss of 85.49 points which saw the year to date ROI drop to 4.78% but I've had one much worse month and one that was similarly bad. Carry on, trying to improve the decision making and be more selective and see how it stands at the end of the year.

The real shocker was the "B" system bets where I lost 292.53 points as a result of only finding 2 winners from 34 bets. Whilst that's freakishly bad it's enough for me to pull the plug on that system as it currently stands, and having looked at the data, revamp it so that the selection process becomes aligned with this, the "A" system. Not that anyone will be that interested, I'll explain the change in a separate post.

[harry_rag]

Quick summary of the differences between the "A" and "B" systems.

1. Spread price range: The original A system relates to players whose goal minutes are priced in the range 20-35 (basically likely scorers who won't usually be odds on). The B system relates to players priced between 15-19.5 for whom I subsequently started gathering data. Less likely scorers who show a poorer return to best bookies odds and my take on "fair" odds.
2. How I calculate "fair" odds: At one point I changed the approach for the A bets but carried on with the existing method for the B bets as I didn't feel I'd gathered enough data to apply the same change. It's not a huge difference but it does mean I've been using two slightly different formulas until now.
3. For the A system it's strictly no bet where the best bookies odds falls short of my minimum back price but, for the B bets, I carried on with the approach of trying to get matched at my desired odds on the exchange if the bookies price was too short.

After the poor performance this month I'm 231.3 points down this year from 289 B bets with an ROI of -8.26%.

If I simply applied the A system approach to the B system data (setting the minimum spread price at 16.5) I'd have had much better returns. Overall the return would've been +23.3 points from 248 bets with an ROI of 9.39%. This year there would've been 158 bets with a return of 35.31 points and an ROI of 22.35%.

So the changes are use the same approach for calculating fair and target odds, no bet where the bookies odds are too short and set the minimum spread price at 16.5. It seems an absolute no-brainer to use an approach that has already proved itself with the A data and provides such positive returns from the B data. Effectively I'm only running one system now and have just extended the spread price range from 20-35 to 16.5 to 35 but I'll keep it separate for the rest of this year to see if it carries on performing at the same level.

(Just checked and I'd have still had a poor month last month even with the new approach; -11.65 points from 19 bets with an ROI of -61.3%. It doesn't alter the decision though as the current approach has never looked better than borderline profitable so I'll trust the overall data.)

[harry_rag]

12 points on Aubameyang (Marseille) at 9/5 with Hills

[harry_rag]

12 points on Oyarzabal (R Sociedad) at 19/10 with Lads

[harry_rag]

12 points on Sterling at 7/4 with 365

[harry_rag]

12 points on Son at 2/1 Lads or Coral

[harry_rag]

12 points on Aboukhlal (Toulouse) at 13/5 with Lads

[harry_rag]

12 points on Moffi (Nice) at 2/1 with Hills

[harry_rag]

12 points on Zapata (Torino) at 9/4 Sky Bet

[harry_rag]

3 from 7 since the last update (Son, Aboukhlal and Moffi).

71 winners from 197 bets so far for a profit of 288.15 points with an ROI of 14.22%.

18.6 points lost on 7 bets posted in error.*

Current "target" ROI based on overall sample is 13.52% (now based on 588 bets from 3326 selections).

* Spotted 2 more recent bets that had been posted in error so did an audit and found another 2. There were 2 winners in among those bets so they made a small profit, hence the reduction in the net loss for all bets posted in error. There's a flag in the spreadsheet that occasionally makes me think that a player is a system pick when they aren't. I'll try and switch it off to minimise the risk of it happening again. 

[Torque]

On 9/1/2023 at 2:29 PM, harry_rag said:

Not that anyone will be that interested, I'll explain the change in a separate post.

Don't be so defeatist. You know what it's like on here. Back in the old days, you could see at the bottom of any forum page how many registered users and guests were looking at a thread and if it was still like that I'm sure you'd find plenty are interested.

[harry_rag]

2 minutes ago, Torque said:

Don't be so defeatist. You know what it's like on here. Back in the old days, you could see at the bottom of any forum page how many registered users and guests were looking at a thread and if it was still like that I'm sure you'd find plenty are interested.

All very tongue in cheek, though I appreciate that most people won't be as interested as me in the minute details of the systems; sometimes it's good just to get your thinking down and preserved for posterity.

This year I'll be seeing how my anytime TD scorer system does when the NFL starts and I've got a very experimental forecasting system to try out on the Rugby World Cup. I'm turning into some sort of hipper betting version of Magnus Pyke! 

[harry_rag]

12 points on Weghorst at 11/5 with Uni

[harry_rag]

12 points on Calhanoglu (Turkey) at 3.8 on BF (13/5 best bookies price is fine)

[harry_rag]

12 points on Johnson at 21/10 with Uni

[harry_rag]

I've been asked a few questions about the system via PM but thought there'd be more value in posting the reply on the thread in case anyone else has similar questions in future. I'm not going full chapter and verse, just trying to stick to what I've been asked.

• Who are "the spread betting firms" - for the purpose of this thread, Sporting Index and Spreadex. Both must be quoting the player goal minutes market for a game for there to be a potential system bet. If you're unsure about how spread betting works (buying and selling, the more right you are the more you win but the more wrong you are the more you lose) then check out the two firms' websites for more information but it doesn't really matter in terms of this system as I'm not betting with them, just using their prices as input data.
• What are the player goal minutes prices? I'll use tomorrow's West Ham v Man City game as an example. I usually wait until the line-ups are confirmed but let's assume these 3 players all start at the stated prices:
  
  Haaland and Foden are outside the range of 20-35 that applies for this system, Haaland being too high and Foden too low. Alvarez is bang in the sweet spot so would be considered as a potential bet. Based on the rule of thumb that the average time a goal is scored is 50 minutes then we can use these prices to arrive at a goal expectation for each player. A price of 40 would denote 0.8 of a goal, a price of 25 would be 0.5 and 18 would be 0.36 etc. etc. Alvarez is priced at 18-21 with Spreadex which gives him a market price of 20-21 (the best prices we can sell or buy at). That gives a midpoint of 20.5 which is the value I use when considering the 20-35 criteria. 20.5 would equate to 0.41 of a goal. Again, I'll refrain from explaining how a bet on this market works because I'm just using those prices as data; the information is out there if required.
• What next? Having identified Alvarez as a potential bet I enter his details in a spreadsheet along with the best available anytime goalscorer price for him as displayed on Oddschecker (which is currently 2/1). The spreadsheet calculates fair odds for him to score (I make it 3.03) and the odds at which I'd be willing to back him with a big enough margin added (3.28). So it's no bet as he's not available at what I regard as fair odds, let alone my target back odds. Had he been 3.3 or bigger I'd be backing him as a selection for this thread.
• How do you calculate the fair odds? Well I'd prefer not to give the precise recipe away but it's largely based on using the poisson distribution and applying it to the goals expectation arrived at as detailed above.

Having read that back I can imagine it being as clear as mud to anyone not already familiar with some of the concepts! 

I'll give a couple of worked examples from recent bets that might make it a little bit clearer.

[harry_rag]

On 9/11/2023 at 7:18 PM, harry_rag said:

12 points on Johnson at 21/10 with Uni

His goal minutes were priced at 22-24 giving a midpoint of 23. The best available odds of 3.1 exceeded my target odds so I backed. He didn't score so it was a losing bet.

Anyone who bought his goal minutes at 24 would have lost 24 x their stake because he didn't score but anyone who sold would have won 22 x their stake.

On 9/3/2023 at 11:34 AM, harry_rag said:

12 points on Aboukhlal (Toulouse) at 13/5 with Lads

He was priced at 20-22 (midpoint 21) and was a bet because the price exceeded my target odds. He scored in the 8th minute so I won my bet.

Anyone who bought his goal minutes at 22 would have lost 14 x their stake but anyone who sold at 20 would have made a 12 point profit. If you buy you want the player to score as many goals as possible, the later the better. If they score too early (as in this case) then you would still lose money as a buyer.

On 9/2/2023 at 2:47 PM, harry_rag said:

12 points on Son at 2/1 Lads or Coral

He was priced at 25-24 (midpoint 24.5) and was a bet at the price as per the previous examples. He scored a hat trick in the 16th, 63rd and 66th minutes so I won my bet, though the 2nd and 3rd goals made no difference to the return.

His goal minutes made up at 145 (the aggregate time of all the goals he scored) so anyone who bought at 24 won 121 x their stake while anyone who sold at 25 lost 120 x their stake.

Hopefully that brings it to life a bit more than the dry theory in the previous post!

[Torque]

1 hour ago, harry_rag said:

The spreadsheet calculates fair odds for him to score (I make it 3.03) and the odds at which I'd be willing to back him with a big enough margin added (3.28).

If you're coming up with your own odds and margin, and providing you have enough faith in those numbers, you should consider mild kelly staking. It should make what you're doing more efficient.

[harry_rag]

15 minutes ago, Torque said:

If you're coming up with your own odds and margin, and providing you have enough faith in those numbers, you should consider mild kelly staking. It should make what you're doing more efficient.

Have you got a suggested formula in mind? I could run it against my results. @Smiles Tennis kindly ran some analysis for me on various staking plans and Kelly did ok on that (in terms of plans that never bust the bank) but the ROI wasn't that much better than level stakes and the staking was quite volatile. One advantage with level stakes is I can pitch it at a level I know I can get on more often than not. The trouble with any staking plan (especially if the stakes cover quite a wide range) is I might find myself trying to get more on than I'm able to, thus diminishing some of the advantage from using the plan.

[Torque]

52 minutes ago, harry_rag said:

Have you got a suggested formula in mind? I could run it against my results. @Smiles Tennis kindly ran some analysis for me on various staking plans and Kelly did ok on that (in terms of plans that never bust the bank) but the ROI wasn't that much better than level stakes and the staking was quite volatile. One advantage with level stakes is I can pitch it at a level I know I can get on more often than not. The trouble with any staking plan (especially if the stakes cover quite a wide range) is I might find myself trying to get more on than I'm able to, thus diminishing some of the advantage from using the plan.

I've seen 50% used quite widely, but you could be really conservative and use 25% for example, and include a conservative loss recovery element - for example 25% kelly stakes relative to bank high unless the bank drops by 25% when you would recalculate the 25% kelly stakes to the reduced bank and continue with loss recovery in mind.

If your average odds are around evens then 25% kelly should be safe enough and maybe too safe, but even so in my experience over thousands and thousands of bets it will outperform level staking. And that makes sense, since some even money bets for example will contain more value than others. 

[Smiles Tennis]

3 hours ago, Torque said:

I've seen 50% used quite widely, but you could be really conservative and use 25% for example, and include a conservative loss recovery element - for example 25% kelly stakes relative to bank high unless the bank drops by 25% when you would recalculate the 25% kelly stakes to the reduced bank and continue with loss recovery in mind.

If your average odds are around evens then 25% kelly should be safe enough and maybe too safe, but even so in my experience over thousands and thousands of bets it will outperform level staking. And that makes sense, since some even money bets for example will contain more value than others. 

During covid lockdowns I signed up to and used Trademate sports. It was excellent until I got limited at all the bookies (which they say will happen in time as you're betting value prices before they can adjust them down to the sharps), but they suggest 30% Kelly which worked very well for that betting. Doubled a bank in a few months I think it was. 

[harry_rag]

12 points on Vlahovic (juventus) at 21/10 with 365

[alexcaruso808]

Harry how do you select what games to bet on?  On a Saturday there are so many high quality games, it's hard to pick!

[harry_rag]

49 minutes ago, alexcaruso808 said:

Harry how do you select what games to bet on?  On a Saturday there are so many high quality games, it's hard to pick!

Any game where both spread firms are quoting the player goal minutes market first and foremost (a pre-requisite for a system bet). Pretty much all Prem, most La Liga and Serie A, televised games in France, Germany and other UK leagues. I might look at other games where one firm is quoting the goal minutes but it would be a "studied" bet rather than an automatic one. When I'm "on duty" I just plough through the games as the line-ups come through. Some players stick in your mind so I'll always back them if the price is in the ballpark of my minimum odds. I'd hardly ever even look at any game that didn't have at least one of the spread firms quoting prices. I'd have nothing to orientate myself against!

[harry_rag]

12 points on Wilson at 8/5 with 365