Greyhound BSP betting strategies

[tico]

Hi,
Looks good but remember lack of liquidity .Did you do all day racing or just concentrate on the afternoon when liquidity is at it's best ?
Kinders
Tico

[Mike Oxlong]

Hey @tc2580, I am running a similar greyhound BSP strategy that should be +ve EV in theory. However in practice, I am losing some edge due to 'missed' bets, either not getting money on runners I should be, or ending up taking BSP when it is below my predicted odds.

I have automated to take BSP under certain conditions, but haven't found the optimal way to ensure all selections are taken at correct prices. What is your strategy for ensuring that all correct bets are taken?

Cheers

[tico]

Hi Folks
Only my humble opinion but automated strategies usually end in tears .Semi automate and keep your eye on the ball (or horse )>
Kinders
Tico

[LinusP]

Hey @tc2580, I am running a similar greyhound BSP strategy that should be +ve EV in theory. However in practice, I am losing some edge due to 'missed' bets, either not getting money on runners I should be, or ending up taking BSP when it is below my predicted odds.

I have automated to take BSP under certain conditions, but haven't found the optimal way to ensure all selections are taken at correct prices. What is your strategy for ensuring that all correct bets are taken?

You want to use a LimitOnCloseOrder where the 'Bets are matched if, and only if, the returned starting price is better than a specified price', I assume BA can do this?

Hi Folks
Only my humble opinion but automated strategies usually end in tears .Semi automate and keep your eye on the ball (or horse )>
Kinders
Tico

Tears of joy in my humble opinion.

[Mike Oxlong]

You want to use a LimitOnCloseOrder where the 'Bets are matched if, and only if, the returned starting price is better than a specified price', I assume BA can do this?

Ah yes, I can see this in the Betfair API, will have a dig around in BetAngel to see if it is surfaced anywhere, cheers!

Edit: For anyone reading with similar interests, I believe you need to use the SP Bets Limits parameters.

[jediwolf1]

Over the past few months I’ve been experimenting with two automated greyhound racing betting strategies — one focused on backing runners and the other on laying them. Both are fully data-driven and built around the idea of identifying dogs that are either significantly stronger or clearly weaker than the rest of the field based on a set of parameters.

I won't be sharing exact mechanics behind the models, as that would pretty much reveal the edge (if there is one at all), but I will share performance stats, Montecarlo outputs, and some observations. I'm genuinely interested in discussion, critique, and seeing how others approach similar concepts. Is it sustainable? Am I not seeing something obvious that will break the strategy soon?

Over the last 90 days, I’ve put both strategies through testing using £1 stakes. Below are the high-level results before compounding, staking plans, or anything fancy — just flat stakes applied to raw model selections.

BACK STRATEGY:

Bets: 2984
Total P&L: £222.57 (based on £1 stakes)
Mean per bet: 0.0746
Average odds: 3.74
Win rate: 38%
Loss rate: 62%



Montecarlo output:

12 month (£1 stakes)




P&L distribution, 10000 12-month simulations:




This is very much a high-volume, low-yield approach, chipping away with small but consistent gains. The strike rate isn’t anything spectacular on the surface, but the pricing efficiency seems to be doing the heavy lifting — the model appears to pick dogs that are regularly mispriced at the BSP. Over the 90-day test window, the strategy produced a £222 profit from flat £1 stakes, which isn’t life-changing, but it’s steady and importantly, easily scalable, since it's betting on BSP.


LAY STRATEGY:

Bets: 17827
Total P&L: £1,056.53 (based on £1 stakes, not liability)
Mean per bet: 0.0593
Average odds: 17.8
Win rate: 88.65%
Loss rate: 11.35%



Montecarlo output:

12 month (£1 stakes)



P&L distribution, 1000 12-month simulations:



This strategy operates at the extreme end of the high-volume, low-yield spectrum. On paper, it absolutely dwarfs the Back Strategy in raw profit, returning just over £1,000 from flat £1 stakes across nearly 18,000 bets in 90 days. The hit rate is great — close to 89%. And both the equity curve and Montecarlo projections suggest a very robust upward trajectory over a 12-month period.

However, the apparent profitability needs to be viewed in context. The average lay odds are 17.8, which is frankly uncomfortable territory. Yes, the strike rate keeps the model afloat, but the occasional loser is catastrophic in relative size. When calculating ROI using liability (not the stake), the percentage return collapses to something only marginally above breakeven. That means the strategy isn’t generating large edges per bet — it's simply grinding out thousands of tiny edges and relying on volume to make them visible.

This introduces a practical problem: bankroll requirements. To survive the inevitable losing runs when laying at these odds, you need a very deep wallet and the psychological tolerance to watch large liabilities materialise without flinching.



Final Thoughts / Discussion

One thing I’ve always believed — long before I even wrote a line of code — is that the greyhound markets are incredibly efficient. BSP feels like a brutally accurate reflection of a dog’s true winning chances. And maybe it still is. But these results have left me wondering whether there are pockets of inefficiency that can be systematically exploited… or whether I’ve just been lucky enough to stumble into one during a favourable period.

Both strategies suggest (if that’s the right word) that some kind of edge exists. I’m not claiming I’ve cracked the code or found a magic formula — far from it — but the numbers do raise the question: is this sustainable, or is it simply variance masquerading as skill? The sample sizes are large enough that it feels meaningful, yet part of me keeps thinking I’m missing something obvious that more experienced people take for granted.

So I guess that’s what I’m hoping to understand here. Is this the kind of thing that burns bright for a few months before collapsing? Or is it possible that a casual programmer with a bit of domain knowledge really can uncover inefficiencies that the market hasn’t fully priced in? Can I reasonably expect this to keep working for another three months? A year? Three years? Or am I already bumping into the ceiling and shouldn’t be too proud of myself?

Don’t worry — I’m not quitting my job, mortgaging the house, and buying a yacht called “BSP Beater”. I’m just genuinely curious whether what I’m seeing is:
1. a legitimate, scalable edge,
2. a temporary quirk that will get arbitraged away, or
3. nothing special at all and I’ve just been playing in a statistical sweet spot.

I’d really appreciate honest thoughts from anyone who’s been around this space longer than I have.
Does this look promising, naïve, dangerous, or boringly normal?

Hi,

Just stumbled across this now. Have you continued to have success with this method?

I think that BSP for greyhound races isn't that efficient and that there is value left on the table, im currently running my own strategy with greyhounds and seeing consistent success with it at BSP.