Chat GBT verdict on betting on every horse in the Grand National to offset your losses

My grandfather placed a bet every year on the Aintree Grand National.

He picked the winner every year without fail. He had a simple strategy. He bet on every horse in the race. He actually put a sixpence on every horse running and enjoyed saying he had the winner every year.

But I’m not all that mathematically minded, and recently I questioned whether he would have made a profit or a loss in engaging in such a practice. So, I did what was necessary, I asked Chat GPT to calculate whether he would have made a profit or not.

I asked Chat GPT to give me a breakdown of the past twenty years of Aintree Grand Nationals and calculate based on the number of runners if I put a £1 bet on every horse running would I be in profit or not ?

And if so what would be that figure either way, profit or loss ?

It was a slightly slightly counter-intuitive question. Chat pulled together the last 20 runnings (2005–2025, noting 2020 was abandoned), including winners, odds, and typical field sizes, then modelled the strategy of £1 on every runner in every race.

A Summary of the winners in the last 20 Grand Nationals

Here are the winners and SP odds (key driver of returns):

(Source aggregated from historical results )

Assumptions on the betting strategy

Assumptions (important):

• Typical field size:
  • Older races: ~40 runners
  • Recent (post safety changes): ~34 runners
• Chatused a blended average ≈ 38 runners per race
• 19 actual races (2020 cancelled)

Total staked

• ~38 runners × £1 = £38 per race
• 19 races →
  👉 Total staked ≈ £722

💰 Returns (key insight)

You can only win once per race, so:

• Return = winner odds + stake (e.g. 10/1 returns £11)

Chat then totalled the returns from all winners:

Total returns ≈ £502

Final result

• Total staked: ~£722
• Total returned: ~£502

👉 Net result: –£220 (LOSS)

So What this tells you

My grandad didn’t make a profit!!!! But he got his money’s worth in bragging about getting the winner each year with his unique sense of humour.

❌ You would NOT be in profit – that the final verdict of Chat gpt.

Even with huge winners like:

• 100/1 (Mon Mome)
• 66/1 (Auroras Encore)
• 50/1 (Noble Yeats)

…it’s still not enough to cover betting every horse.

Why?

1Bookmakers build in an “overround”

The combined implied probabilities of all runners exceed 100%, guaranteeing margin.

Field sizes are huge

You’re betting:

• ~38 horses
• but only getting paid on 1
• Even big shocks don’t offset volume

You need very frequent 50/1+ winners to break even—which doesn’t happen often enough.

Quick rule of thumb

To break even betting all runners:

• Average winning odds would need to be roughly:
  (number of runners – 1)

So with 38 runners:
You’d need winners averaging ~37/1

Reality:

• Actual average ≈ ~20–25/1

➡️ That gap = your loss

🧾 Bottom line

👉 Over the last ~20 years:

• Strategy: £1 on every horse
• Result: ❌ Loss
• Estimated loss: ~£200–£250

So there you have it! my grandad lost out in the longer term probably based on the calculations of Chat gpt.

I asked Chat gpt what would have happened if my grandad had bet each way on these horses.

Here’s what they said:

Bottom line

Betting every horse each-way in the Grand National over 20 years:

• ❌ Much bigger loss
• 📉 Roughly –£600 to –£700
• Worse than win-only.

Final thought (this is the interesting bit)

The only way this approach works is if:

• You filter runners (not all horses), or
• You target outsiders only (where each-way value can exist).

Whatever you do at the races, you will need deep pockets. My grandad only betted once a year.