How many records is considered reliable

Hi!

Checking some greyhound and horse strategies.

If I want to bet x rank with odds from y to z and historical results shows over 10% profit/value/edge in a long run, how many historical results do I need to be quite reliable?

If I have records from 1500 races and in these 1500 races there have been 450 contestants which meet the criteria (where I get the stats), is it enough? How much is enough? How many dogs/horses with same criteria?

hi
the minimum amount of records should be 12 months,
this is to cover the 4 seasons, so you then have a reasonable amount.
some horses/dogs like the cold winter, some dont like it,others like the hot summer, others hate it.
some like the rainy season and others grind to a halt.


mcbee

hi
the minimum amount of records should be 12 months,
this is to cover the 4 seasons, so you then have a reasonable amount.
some horses/dogs like the cold winter, some dont like it,others like the hot summer, others hate it.
some like the rainy season and others grind to a halt.


mcbee

Thanks for this. I'm just thinking this about marging of error point of view. For example in surveys you get certain margin of error with certain sample size. Sample size of 384 gets you 5% margin of error. Is it the same in betting (if you compare selections with same attributes)? Or does it have to be 1 year or more? 1 year sample size will be about 10 000. Or if I have sample size of 10 000, will next 10 000 sample be totally different?

hi
the problem that we face is, that 12 months would be good enough,BUT we now have another problem horses and dogs are now 12 months older, so some will be better and others will be slowing down.
you can at the end of the day, only take an average then hope for the best.


mcbee

BUT we now have another problem horses and dogs are now 12 months older

A no brainer really, I am not sure the OP is actually measuring animals, sounds more like just reading the stats for the traps in basic form, meaning he will back a trap that has for example

bet x rank with odds from y to z

Trap 1 could fall into that criteria in one race but trap 4 in the next irrelevant of the animals name or previous history.

GoalScorer, is that what you mean?

A no brainer really, I am not sure the OP is actually measuring animals, sounds more like just reading the stats for the traps in basic form, meaning he will back a trap that has for example

Trap 1 could fall into that criteria in one race but trap 4 in the next irrelevant of the animals name or previous history.

GoalScorer, is that what you mean?

Yes, correct.

If for example I notice that the x ranked (lets say 2nd favourite) when odds between y and z, makes 10% profit after 450 races (ALL races with dog/horse with same rank and odds range), how much can I trust it?

And lets say that the favourite with any odds have won 31 % of races after 1000 races, will favourite win closely to 31 % in next 1000 races or anything between 0 and 100 %?

hi
i have spent years trying that between x and y, it worked for a few weeks and then bang goes may bank.
i have used over 12 months of greyhound results to find the trap with x wins then a loss and x losses then a win, failed.
i even designed triggers to record every winning trap with prices and fav pos, then the triggers would use the recorded info and back or lay after x wins/losses,it got to a point where even with over 12 months of data, the triggers would place the bet and every now again there would be an extra few wins/losses in a row, in the end i gave up because no bets were been placed, due to the maximum wins/losses in a row never happening again.
i had triggers getting data for every single greyhound track, so using just track records, but still the same happend.
i thought that i had it sorted at one time, money was coming in very nicely,because i had the maximum amount of losses in a row before placing the bet,but i used recovery, so the maximum losses in a row was 22, i set the triggers to place the bet at 15 losses in a row after calculating the recovery cylce, allowed 24 losses in a row, then bang 26 losses in a row wiped me out.
bin there done that and could'nt afford the T shirt after.

mcbee

hi GoalScorer
can i ask how you took the results with the prices.
are they recorded sp prices etc

mcbee

hi
i have spent years trying that between x and y, it worked for a few weeks and then bang goes may bank.
i have used over 12 months of greyhound results to find the trap with x wins then a loss and x losses then a win, failed.
i even designed triggers to record every winning trap with prices and fav pos, then the triggers would use the recorded info and back or lay after x wins/losses,it got to a point where even with over 12 months of data, the triggers would place the bet and every now again there would be an extra few wins/losses in a row, in the end i gave up because no bets were been placed, due to the maximum wins/losses in a row never happening again.
i had triggers getting data for every single greyhound track, so using just track records, but still the same happend.
i thought that i had it sorted at one time, money was coming in very nicely,because i had the maximum amount of losses in a row before placing the bet,but i used recovery, so the maximum losses in a row was 22, i set the triggers to place the bet at 15 losses in a row after calculating the recovery cylce, allowed 24 losses in a row, then bang 26 losses in a row wiped me out.
bin there done that and could'nt afford the T shirt after.

mcbee

Okay so it might not be reliable at all...  

Thanks for sharing this info and sorry to hear that.

Did you use lay bet to lay the selection if either rank or odds drifted too far? Cause when bet gets placed, it could be something else when market eventually gets suspended...

I will try couple of strategies. Just BAGS or Televised market (depends which works best where), certain rank with certain odds and that's that. Could be different trap in every race!

I will research this a bit still before putting real money. More races to the database.

And yes they are recorded Betfair SP prices which I try to match in my future strategy. Put huge amount data to Excel and then some searching with formulas.

hi
now there is another problem.
if you record all prices at sp, then you must us sp for your future betting.
when i recorded my data it was set at 1 second before the start time and my bets were placed at exactly 1 second before the start time.
nearly all punters get the results from the racing post or the greyhound board, then use the info to make a system on betfair, big mistake, because when you use a bot or do manual bets, you place your bet at the start time, then the prices alter many times and the favourite position changes many times.
the only way is to record the prices and favourite position at the normal time that you would place your bet, then you are on the correct side for bet placement compared to results.


mcbee

hi
now there is another problem.
if you record all prices at sp, then you must us sp for your future betting.
when i recorded my data it was set at 1 second before the start time and my bets were placed at exactly 1 second before the start time.
nearly all punters get the results from the racing post or the greyhound board, then use the info to make a system on betfair, big mistake, because when you use a bot or do manual bets, you place your bet at the start time, then the prices alter many times and the favourite position changes many times.
the only way is to record the prices and favourite position at the normal time that you would place your bet, then you are on the correct side for bet placement compared to results.


mcbee

Nope mcbee you are incorrect. I have made a fail safe solution for that. I have betted greyhounds even in minus one minute. That is one whole minute after "the start". Would be easy if approximated matched volume for certain racecircuits is known.

And if the price drifts too much, you can lay the selection with very small loss. Small enough that profit is still made if value is large enough.

And the odds range will be so vague that it doesn't matter if the price drifts a bit or a bit more...

And you could also try dutching greyhounds if you get over 97% strike rate.. That's one solution..

But testing is still needed.

could be why i lost.
been to strict with the rules.
i never used test mode for recording the data, always used small stakes.
maybe i should start again.

mcbee