fill rates

[burdo77]

Im wondering what is the best way to calulate the fill rate, initail thought would of been to compare the WOM% with the back and lay side, for example if the back represented 50% and lay represented 50% there would be a good fill rate. However, im thinking there is more to it, please help?

[to75ne]

when you say fill rate do you mean how long it takes for your complete order to get matched at a particular price, or do you mean all orders qued at the price to get filled/matched?

[burdo77]

The rate in which all orders at current price get matched.

[LeTiss]

I have absolutely no idea how to calculate the fill rate, especially across markets that I'm not actively watching

However, if someone has the answers I'd be interested in seeing them, because the fill rate is the biggest killer for me. Time is money, and it's definitely taking longer for my bets to get matched, which in turn has an impact on the bottom line if I'm not trading as many markets as before

[to75ne]

personally i think it would be almost impossible as i assume you would need to take the amount on the back side qued up at a given instance, compare to the amount matched on the layed side, at the same instance, then compare the back side to see if more as been added, or how much as been pulled, and compare that with how much as been macthed on the layside again. and on and on and on, many times every second, for every runner. and thats before you calculate any meaningful number to represent a "rate".

reckon excel would a heart attack and die.

personally i judge by "eye" along with the average amount matched thats already caculated.

[burdo77]

I appreciate your answer to75ne however, im reording data therefore calculating by eye does is not possible. There is alot talked about fill rates on many forums, however nobody goes into specifics on how....

I think finding a measurable way of calculating the fill rates would give a greater understanding to what makes the price fluctuate.

[to75ne]

like letiss i would be interested in a meaningful measurement but the problem seems to me to be, how do you filter out spoof money from real money, and money that gets pulled.

i cant see how you could even meaningfully estimate the proportion of real money compared to how much spoof money there is qued at any given price , especially in the last 3 mins or so (horse racing).

Postby burdo77 » Mon Jul 28, 2014 3:16 pm

There is alot talked about fill rates on many forums, however nobody goes into specifics on how....

burdo77

maybe why nobody goes into specifics on said forums is because nobody as come up with a viable/accurate method --- my opinion.

[LinusP]

Why not just measure the average amount being matched per second?

Using the API you can pull the amount matched per odds and compare that per 0.2second to gauge the amount being matched.

[burdo77]

Thanks Linus, your answer makes sense. I'll think I'll look at incorporating something like this.

Why doesn't bet angel incorporate something like this, unless they believe that fill rate is not that important?

[to75ne]

linus idea is good but you will probably need to refine it to make it more meaningful - have ref/scale/baseline.

for instance £5000 matched divided by .2 = 25000

£500 matched divided by .2 = 2500

£50 matched divided by .2 = 250

you going to get results jumping from small numbers to large numbers all over the place.
it might be an idea to convert the results to logs and calibrating a scale so to speak.

have look how they would do this sort of thing in financial markets, i assume this would be something they measure.

[LinusP]

Agreed, you have so many variables, countdown, matched amount, racing grade, favourite odds, selection odds, day.....

I know Peter mentions fill rate during the large meetings but I think he puts money in to calculate it.