The wisdom of the crowd...

[jamesedwards]

Rather than measuring it on an outright basis, try measuring against return on stake or liability. That's how I determine if my strategy is working or getting better or worse.

Do you mean cumulative P&L as a % of cumulative stake?

Like this...

z309.jpg

Interesting, never thought of measuring this before.

[jamesedwards]

...sorry to ask what will no doubt be a dumb question James, but is your latest graph suggesting that in the 988 matches monitored there have been 78 more draws than non-draws?

ie 533 draws and 455 non-draws? or is it a percentage graph?

The graph was cumulative profit over time. Number of markets on the X-axis against cumulative P&L on the Y-axis.

1007 markets from which 289 were draws.

[Tuco]

thanks James

are you backing the draw at the very start of the match, before the start or during the match?

[jamesedwards]

thanks James

are you backing the draw at the very start of the match, before the start or during the match?

Just before KO. Offering, not taking, and only when the market meets certain maturity thresholds.

[jamesedwards]

Still waiting for Mr Mean Reversion to stomp all over this.

Perhaps this weekend?

[Stomp stomp stomp!]

z313.jpg

[Euler]

Are you trading out or just placing a bet. As football can go long periods above or below average.

I haven't run my data so far this season but can do at some point this week.

[ruthlessimon]

As football can go long periods above or below average.

Out of interest (question to anyone btw)

How do you personally tell the difference between a real pattern that’s short-lived and pure randomness?

[jamesedwards]

Are you trading out or just placing a bet. As football can go long periods above or below average.

I haven't run my data so far this season but can do at some point this week.

I'm just letting the bets ride.

At least this way once it flatlines I will have at least contributed slightly towards Expert Fee.

[jamesedwards]

As football can go long periods above or below average.

Out of interest (question to anyone btw)

How do you personally tell the difference between a real pattern that’s short-lived and pure randomness?

There's the r² https://en.wikipedia.org/wiki/Coefficie ... ermination
Right click on a line of data on an excel graph, select 'add trendline', then tick the 'display r² value on graph' checkbox.

Also Peter's suggestion to plot cumulative profit over cumulative stake over time makes sense as this shows the trend of your EV.

[ruthlessimon]

As football can go long periods above or below average.

Out of interest (question to anyone btw)

How do you personally tell the difference between a real pattern that’s short-lived and pure randomness?

There's the r² https://en.wikipedia.org/wiki/Coefficie ... ermination
Right click on a line of data on an excel graph, select 'add trendline', then tick the 'display r² value on graph' checkbox.

Also Peter's suggestion to plot cumulative profit over cumulative stake over time makes sense as this shows the trend of your EV.

Did you backtest the strategy before taking it live — or just start based on the underlying logic and let the r² play out?

[jamesedwards]

Out of interest (question to anyone btw)

How do you personally tell the difference between a real pattern that’s short-lived and pure randomness?

There's the r² https://en.wikipedia.org/wiki/Coefficie ... ermination
Right click on a line of data on an excel graph, select 'add trendline', then tick the 'display r² value on graph' checkbox.

Also Peter's suggestion to plot cumulative profit over cumulative stake over time makes sense as this shows the trend of your EV.

Did you backtest the strategy before taking it live — or just start based on the underlying logic and let the r² play out?

Ner, I literally am only running it for the craic.

Bored trading 3m slog horse races all day.

[ruthlessimon]

There's the r² https://en.wikipedia.org/wiki/Coefficie ... ermination
Right click on a line of data on an excel graph, select 'add trendline', then tick the 'display r² value on graph' checkbox.

Also Peter's suggestion to plot cumulative profit over cumulative stake over time makes sense as this shows the trend of your EV.

Did you backtest the strategy before taking it live — or just start based on the underlying logic and let the r² play out?

Ner, I literally am only running it for the craic.

Bored trading 3m slog horse races all day.

Fair enough Might be worth a thread of its own. I’m really interested to see how this plays out; i.e. proof why backtesting isn’t always necessary.

[jamesedwards]

Did you backtest the strategy before taking it live — or just start based on the underlying logic and let the r² play out?

Ner, I literally am only running it for the craic.

Bored trading 3m slog horse races all day.

Fair enough Might be worth a thread of its own. I’m really interested to see how this plays out; i.e. proof why backtesting isn’t always necessary.

I'm sure it will eventually return to mean. I only started it for a laugh.

Although I'm a data driven guy I have no programming ability, and prefer the trial and error approach to trading. I've never back tested anything, but have a few very successful horse racing automations that are just ideas I set running, which were tweaked and refined over time.

I've never had a long-term successful football automation though. I've had some success at a net net level (ie contribution to expert fee > loss after commission), but Betfair didn't like that and put a stop to it.

[Naffman]

Thought this was interesting

IMG_6371.png