What Data Is Available With A Delayed Key?

[tommo]

Hey all, (NEW GUY HERE!)

I'm currently attempting to analyse as much data as possible using the delayed key. This is to train a LightGBM model.

I've spent the last couple hours racking my brains as to why i'm not able to pick up some of the data!

(I believe this is because i'm using a delayed app key - but i can't find any documentation as to what I can and can't use other than):

-

The Delayed App Key does not return traded volume data 'totalMatched' or EX_ALL_OFFERS via listMarketBook.

For instance I want to access the Runnerbook SP values. i'm coding in the same way i access the EX values:

Examples below (Python betfairlightweight)

BELOW WORKS FINE

best_back_prices = [runner_book.ex.available_to_back[0].price
if runner_book.ex.available_to_back
else 1.01
for runner_book in runner_books]

BELOW ALWAYS RETURNS "NONE"

back_stake_taken_price = [runner_book.sp.back_stake_taken[0].price
if runner_book.sp.back_stake_taken
else None
for runner_book in runner_books]



What I would like to know is if anybody can give me a list of values available via the delayed key so I don't spend any more time down rabbit holes.

If anyone has any tips on the data indicators for price direction as well - this would massively help with my tree based model. (i'm currently 1% off breaking even with a limited data model!)


Regards,

Tommo

[bfexplorer]

You are so wrong. Delayed data are random data, having nothing to do with real one. Buy your full subscription for api access.

[LetsGo]

Hey all, (NEW GUY HERE!)


If anyone has any tips on the data indicators for price direction as well - this would massively help with my tree based model. (i'm currently 1% off breaking even with a limited data model!)

The exchange book is about "1% off breaking even" before the start of a race, so just backing every horse in a race would give you that result.
Maybe you can't see the wood for the trees.

[tommo]

You are so wrong. Delayed data are random data, having nothing to do with real one. Buy your full subscription for api access.

I'm not sure this is true ... can anyone confirm? - the documents say it's live data which is delayed 1-180 seconds. this is fine for me for testing purposes as i sweep every 180 seconds anyway.

The exchange book is about "1% off breaking even" before the start of a race, so just backing every horse in a race would give you that result.
Maybe you can't see the wood for the trees.

You're not wrong .... i'm actually trying to scale up! so i have confusion matrix on profitable model - This has high precision and low recall:

| .....................True
| ..................up ....equal ....down
| ..........up ....75..... 564 ......35
|Pred .equal 37 ......50k+ ....20
| .........down 24 ......486 ......176
|

This gives a percentage of 68% correct and profitable margin. On a £2 stake daily return predicated at £1.35 LOL - not the best!

When I scale up - (decrease precision and increase recall) my percentages drop to 52% correct. which is actually not profitable! if you do the maths due to the up/down predictions when it's actually "equal" will lose 1 tick when you green. i just need to hit 53% but the scale is much much bigger and would increase profit to about £3 (Lol - Still low but it's a start and proof of ML concept working before I invest)

Anyway ...

Anybody able to answer my original question?

What I would like to know is if anybody can give me a list of values available via the delayed key

Cheers community