Charting from Excel in BA

[ShaunWhite]

Now let's ffwd 6 weeks and see how you all did.....

mmmm not so good is it, i wonder if they'll take the new car back?

But this chart is entirely predictable, laying the 4th fav on dogs has no 'substance', no fundamental reason why it should work. So what we get is a reversion to mean, just as you'd expect. By all means try to make hay while the sun shines on a strategy, but unless you know why it's working, you'll probably need your brolly sooner rather than later.

[Atho55]

Why did you feel the need to introduce a strategy on dogs that does not resemble the gist of the thread. I think most would have found it more constructive to take the criteria, run it through the dataset I know you have from 1st Feb 18 to 19th Mar 19 then we can all look at a historical v current chart.

I don`t think anyone is saying it`s a get rich scheme. It worked for a few days, failed for a lot more days and has just worked for a few more days. It can just as easily start to fail again.

If we can find some clues why it worked when it did then we might be able to find an entry point with less risk.

[wearthefoxhat]

Now let's ffwd 6 weeks and see how you all did.....

mmmm not so good is it, i wonder if they'll take the new car back?

But this chart is entirely predictable, laying the 4th fav on dogs has no 'substance', no fundamental reason why it should work. So what we get is a reversion to mean, just as you'd expect. By all means try to make hay while the sun shines on a strategy, but unless you know why it's working, you'll probably need your brolly sooner rather than later.

What could also have happened was that the 4th ranked dog became the 5th ranked dog, after the trend had been spotted by systemites, cos everyone that layed the 4th ranked dog then pushed the price out. In that case you'd need to be flexible with your approach. Good graphs though.

In this thread, at least the range of ranks are covered so offers a little more cover.

[BetScalper]

ok stats fans, here's a free edge for you if you like following trends and favouritism rankings. 4,200 dogs so it's a decent sample.

Chart shows £1 level stake commision paid. 300pts profit in a month, easy money.

Lay the dog ranked 4th fav....that's it.
Should be easy in Guardian, £20 a race, 6 grand a month. Set it up and phone the car dealership. Who says nobody ever gives away an edge on here.

Hey Shaun,

What would the profit line look like if the 4th favourite was in Trap 1 at Nottingham, which only has a 14.3% win rate regardless of Ranking ?

[Valda]

This can be checked using the HorseRaceBase data in its system builder (not the dogs obviously).
Will try to run all the runner/rank combinations and let you know - might not be today tho.

[Atho55]

These are the results as collated for yesterday. The 2 winners mentioned yesterday by Valda look to be there.

Strat A 2003.png

Noticed yesterday that there may be opportunites to BTL the selections IP so added my attempt to get the ticks between BSP & IPMIN

[Valda]

B2L results are impressive.
Can’t check that in HRB but can check main criteria by year/month/week.

[wearthefoxhat]

B2L results are impressive.
Can’t check that in HRB but can check main criteria by year/month/week.

I appreciate the OP's examples are shown over the jumps. However I sense due the unpredictable nature of All Weather Racing, it maybe worth a second look there too.

Quick review of Class 5&6 on the AWR courses, for 2017/2018/2019, and Jan/Feb/Mar combined. (Adrian Massey Site - Approx guide only)

@0% Comms (with the bookmakers)

0%.PNG

@2% Comms (If you have the special offer)

2%.PNG

@5% Comms (standard betfair fodder)

5%.PNG

Show's how big an impact the comms has at each stage!

[ShaunWhite]

Why did you feel the need to introduce a strategy on dogs that does not resemble the gist of the thread.

Because it perfectly exemplifies the pitfalls of backfitting and with dogs you trade as many in a month as you do horses in a year so the consequences show up faster. What I'm saying isn't exactly controversial.....

********************
Re: Strategy Development: Modelling
Post by Euler » Sat Jan 06, 2018 2:11 pm

Data mining can be statistical deja vu.
...
But that is only the first step as I then need to describe what is happening and why.
********************

If you disgree, take it up with the teacher because I'm just a student.

[Valda]

Bots running - Shaun’s dog one performing like shit so turned it off - see how the pivots do

[Atho55]

As you appear to be the "chart guru" would you be interested in producing one for the strategy criteria to the dates I mention above. I have about 20 days worth of data missing at the front end so can`t do it.

Interested to see how it compares to same time last year.

[Atho55]

You on that one Valda?

[Valda]

Yep

[wearthefoxhat]

You on that one Valda?

On that one Atho55? Class 5 AWR...


Algorithm from Racing Post gave confirmation...

Ling455.PNG

[Atho55]

Yes it was selected... has been a few today