Straight Value bets - Pre play markets in main leagues around the world

[Nickse34]

Hi,

I have spent the last few months really looking into pre play value in the FH and 1.5 goal markets. What I’ve found is that there isn’t much left for someone like me to take. I’ve been betting those markets for 2 years but the more I learn the harder it seems to get. I’ve got a few niche systems that yield between 2 and 4% ROI but I can't stake big on them and the volume of bets is limited.

I do not have data no else has and no insider info. My models are decent but I cant beat the market or even use it to anticipate an early mis pricing. I don’t even think modelling pre play markets is smart, syndicates , sharps and bots are just in a league on their own it seems. It’s really tight and sharp. As I said I did find a bit of dumb money here and there but I don’t think I can make sustainable income from it. Can’t even get matched most of the time, or not at a value price!

Any wisdom from anyone here on whether I still need to dig deeper or just look at inplay or even pre off racing!

Disheartening business that! I remember when I first started and some online guru told me that pre play football betting would be a passive income requiring minimal efforts. Make money while you sleep…

[jamesedwards]

Hi,

I have spent the last few months really looking into pre play value in the FH and 1.5 goal markets. What I’ve found is that there isn’t much left for someone like me to take. I’ve betting betting those markets for 2 years but the more I learn the harder it seems to get.

I do not have data no else has and no insider info. My models are decent but I cant beat the market or even use it to anticipate an early mis pricing. I don’t even think modelling pre play market is smart, syndicates , sharps and bots are just in a league on their own it seems. It’s really tight and sharp. I did find a bit of dumb money here and there but I don’t think I can make sustainable income from it. Can’t even get matched most of the time, or not at a value price!

Any wisdom from anyone here on whether I still need to dig deeper or just look at inplay or even pre off racing!

Disheartening business that!

The markets are sharp. 25 years of fat has been trimmed leaving only the very leanest of cuts remaining.

Horse racing is my game and I've never been able to make football work. But if I was going to put some real effort behind it then I would pick an obscure league somewhere and pile all my effort into becoming an expert on that.

[Nickse34]

Hi,

I have spent the last few months really looking into pre play value in the FH and 1.5 goal markets. What I’ve found is that there isn’t much left for someone like me to take. I’ve betting betting those markets for 2 years but the more I learn the harder it seems to get.

I do not have data no else has and no insider info. My models are decent but I cant beat the market or even use it to anticipate an early mis pricing. I don’t even think modelling pre play market is smart, syndicates , sharps and bots are just in a league on their own it seems. It’s really tight and sharp. I did find a bit of dumb money here and there but I don’t think I can make sustainable income from it. Can’t even get matched most of the time, or not at a value price!

Any wisdom from anyone here on whether I still need to dig deeper or just look at inplay or even pre off racing!

Disheartening business that!

The markets are sharp. 25 years of fat has been trimmed leaving only the very leanest of cuts remaining.

Horse racing is my game and I've never been able to make football work. But if I was going to put some real effort behind it then I would pick an obscure league somewhere and pile all my effort into becoming an expert on that.

Yes, I did look at more obscure leagues and markets but the lack of liquidity makes things hard. Money comes in at the last second before KO and will make the price move quite dramatically and unpredictably. Also quite hard to rely on CLV as a good indicator. I don't know...

[ShaunWhite]

Hi,
My models are decent but I cant beat the market

Hi,
Rhetorical questions really but I'd say consider...

What modelling software are you using and would changing help.
Are you happy with the number of data features you're passing with each team/game or could you suppliment that.
Have you got enough (and relevent) data history.
Have you exhausted your model training hyperparameters, small tweaks can help a lot.
Is it a static pricing model or do you also feed it live market info.

[wearthefoxhat]

My models are decent but I cant beat the market or even use it to anticipate an early mis pricing. I don’t even think modelling pre play markets is smart, syndicates , sharps and bots are just in a league on their own it seems.

FWIW, taking a view pre-off, and seeing how it evolves in-play, is the way I approach football overall. I don't tend to look at the obscure leagues, mainly due to the standard of play and potential match rigging. Preference is for the bigger leagues, not necessarily the lower divisions, due to the same reasons mentioned.

There is a perception that all models are wrong, not saying I'd agree with that, but through testing and review, you'll get it to the stage you're content with it, until there's a rule change that could impact your model. ie: VAR or no VAR (cup games, lower divisions), 5 subs allowed (timing strategy), drink breaks (stops momentum).

In the future there may be no linesman, changes to the marginal offside rule, or only applicable up to and beyond the 18 yard line, amber cards - sin bins. Stopping the clock when there's an infringement, then on the 45th/90th minute, the game stops when the ball is out of play. (rugby style).

Any of these would have an impact, will they create less or more goals?

[MrJoeBlack]

Two responses to this FWIW - I tend to trade football as I would Texas Hold'em i.e. I identify 'betting rounds' and determine whether my position has improved, worsened, or is much the same. I think it's important not to ignore context and that evolves from the initial offer in the market, team news, kick off, first half, HT, etc.

And then, this is a summary of what I'm doing at the moment in case there are any useful insights.

I wanted to create a data-driven representation of how football teams actually play — and how their style changes depending on the opponent.

This lets me:

• Compare styles between teams
  Understand what a team actually does tactically
  Predict how two teams will interact
  Identify advantage/weakness patterns before a match

It is not just a strength rating.
It is a style identity model.

1. Build Player Identity Vectors (per game)

From each match report, every player gets a feature vector describing how they played:

I then converts each player’s numbers into 6 performance characteristics:

PC1 = volume, involvement, ball usage
PC2 = progression vs safety
PC3 = defensive intensity vs passive play
PC4 = direct vs slow buildup
PC5 = risk vs control
PC6 = dynamic movements vs static roles

(These exact meanings evolve with more data — but that’s the core pattern emerging so far.)

2. Combine Player Vectors Into a Team Identity Vector (per game)

3. Contextualise Each Identity Based on Opponent

A team’s identity in raw form reflects both:

what the team chose to do
what the opponent forced them to do

4. Use Team Identity Vectors to predict games. Use Player Identity Vectors to assess key team news.

[Nickse34]

Hi,
My models are decent but I cant beat the market

Hi,
Rhetorical questions really but I'd say consider...

What modelling software are you using and would changing help.
Are you happy with the number of data features you're passing with each team/game or could you suppliment that.
Have you got enough (and relevent) data history.
Have you exhausted your model training hyperparameters, small tweaks can help a lot.
Is it a static pricing model or do you also feed it live market info.

I do everything in Python and excel.

I am happy with the features I use but It is a static model - I try to enter the market before team news and smart money start to come in and let the market either confirm or refute my hypothesis. This is how I compensate for the lack of live information - position management and optimal entry/exit execution. Could I add more features, potentially but nothing the market doesn't already know about and I refrain from just taking shots in the dark. Unless I have an hypothesis I want to explore I don't add for the sake of it, hoping it'll improve what I do.

I have a lot of historical data but mainly use what i have collected since the added time rule change in 2022.

Have I exhausted everything? I don't know is the honest answer but probably not - there are a lot of things I just don't know about.

[Nickse34]

My models are decent but I cant beat the market or even use it to anticipate an early mis pricing. I don’t even think modelling pre play markets is smart, syndicates , sharps and bots are just in a league on their own it seems.

FWIW, taking a view pre-off, and seeing how it evolves in-play, is the way I approach football overall. I don't tend to look at the obscure leagues, mainly due to the standard of play and potential match rigging. Preference is for the bigger leagues, not necessarily the lower divisions, due to the same reasons mentioned.

There is a perception that all models are wrong, not saying I'd agree with that, but through testing and review, you'll get it to the stage you're content with it, until there's a rule change that could impact your model. ie: VAR or no VAR (cup games, lower divisions), 5 subs allowed (timing strategy), drink breaks (stops momentum).

In the future there may be no linesman, changes to the marginal offside rule, or only applicable up to and beyond the 18 yard line, amber cards - sin bins. Stopping the clock when there's an infringement, then on the 45th/90th minute, the game stops when the ball is out of play. (rugby style).

Any of these would have an impact, will they create less or more goals?

Yes, I don't look at how things develop in-play but I have considered it. I would require a complete change of approach and I may well look into it more around Xmas time.

I don't know if a model can't be right or wrong and I don't expect mine to be right. What I have helps me with making informed decisions but it hasn't allowed me to find my space in the pre play markets. I still feel that I am getting somewhat bullied and I am wondering if going deeper with the whole modelling thing is actually beneficial or not, especially if there isn't anything fundamentally unique with what I can feed into it.

A lot of people exploit these rule changes and things like break in play, cards, injuries but I haven't got the stats to really look into this at the moment. Again, I would probably need to shift my attention to in-play a bit more and build an in-play database.

Thanks

[Nickse34]

Two responses to this FWIW - I tend to trade football as I would Texas Hold'em i.e. I identify 'betting rounds' and determine whether my position has improved, worsened, or is much the same. I think it's important not to ignore context and that evolves from the initial offer in the market, team news, kick off, first half, HT, etc.

And then, this is a summary of what I'm doing at the moment in case there are any useful insights.

I wanted to create a data-driven representation of how football teams actually play — and how their style changes depending on the opponent.

This lets me:

• Compare styles between teams
  Understand what a team actually does tactically
  Predict how two teams will interact
  Identify advantage/weakness patterns before a match

It is not just a strength rating.
It is a style identity model.

1. Build Player Identity Vectors (per game)

From each match report, every player gets a feature vector describing how they played:

I then converts each player’s numbers into 6 performance characteristics:

PC1 = volume, involvement, ball usage
PC2 = progression vs safety
PC3 = defensive intensity vs passive play
PC4 = direct vs slow buildup
PC5 = risk vs control
PC6 = dynamic movements vs static roles

(These exact meanings evolve with more data — but that’s the core pattern emerging so far.)

2. Combine Player Vectors Into a Team Identity Vector (per game)

3. Contextualise Each Identity Based on Opponent

A team’s identity in raw form reflects both:

what the team chose to do
what the opponent forced them to do

4. Use Team Identity Vectors to predict games. Use Player Identity Vectors to assess key team news.

I have read about this quite a lot (David Sumpter) Very interesting approach but the data I'd need won't be cheap. Do you stream an API? Howe many leagues do you operate in?