Forums

General Betting

There is currently 1 person viewing this thread.
These 379 comments are related to the topic:
Betting for a living

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
Page 4 of 10  •  Previous | 1 | 2 | 3 | 4 | 5 | 6 | ... | 10 | Next
sort by:
Show
per page
Replies: 379
By:
Dr Crippen
When: 16 Jul 19 11:11
''See the light'' DFCIRONMAN?

Yet you can't even explain why you don't simply back horses that drift in the market when these types obviously provide the best value.
By:
aye robot
When: 16 Jul 19 11:25
Ok DC I'll take you at face value... here you go:

First off I have to make clear again that what I do is mostly about looking for untrue market conditions. In other words I don't worry about Form or whatever, I just look for signals that the market is behaving irrationally. Once I see those signals I re-price the market using some special sauce of my own recipe. Working out edges is as simple as 1/market-price - 1/my-price.

But I think that's not what you want to know about - you want to know how you would use form or other data to calculate value pre-off. Although that's not what I do but I don't mind having a stab at how I might do it on the understanding that it's not my area of best expertise:

First let's clarify terms: there is 'value' and there is 'edge' and they're not the same but they always come together. 'Edge' is the difference between the rate at which your selections win and the implied probability of the price at which your bets are matched.'Value' is more properly called 'expected value' and is the amount of money that your edge will get you on average. 3% edge at 100 (backing) is much more value than 3% edge at 2. Rather than talking about how to calculate value I'm going to talk about how to calculate edge because that's the hard bit.

The process:

First off I'd get a whole lot of historical data including all the form and breeding data for the horses in my study, all the market data about what prices matched around them and anything else I could think of. I'd also want subtle ways of objectively assessing how well each horses did in each race so I'd get winning and losing distances, shortest prices matched in running and so on.

With that in place I wouldn't start by trying to price up whole races, I'd just go for the easy win: Value is a always contrarian so I'd use market prices as a benchmark and pick out features that everyone knows about to see if they led to poor pricing because of over confidence - for example I'd identify the 5% of horses that had the best breeding and then compare their mean prices matched pre-off to their rate of winning to see if they were over-backed (ergo showing value to lay). Lets say they were - I'd want to check that using the next best 5% - If they are also over-backed but by less then I'd probably start to think that I was in business. I'd also check the converse – so I'd look at horses with poorer breeding and see if they showed value to back.

I might do the same for horses that had won four consecutive races - or I might look to see if horses with a good overall record of winning but a few 'recent' losses are generally over layed and so on through all the features. I'd also look to see if any features were generally over-looked by the market.

Once I'd found some indicative features I'd look at how they interact – if a horse is very well bred and has won four consecutive races does it get 'super' over backed? This part can get fiendishly complicated. All the while I'd check my reasoning, there's no point in looking for value in horses who's names begin with a 'Q' running on Wednesdays.

For complete clarity: I wouldn't expect any of the above ideas to work because they're too obvious, this is just a way of explaining the process. Also, I wouldn't be doing it on an envelope, I'd use a computer because that's what they're for. Machine learning is very accessible these days and if you don't use it someone else will and they'll eat your lunch so that's how I'd go about it.

Assuming that I found some feature that indicated edge, quantifying the 'trailing' edge is easy as calculating the mean difference between the rate at which the horses that I identified won and the prices at which they matched. I'd generate a function for edge relative to the strength of the feature. However it would be naive to imagine that I'd get that edge going forward because the methodology used would tend to maximise the appearance of value that's really just noise (randomness) so I'd need a way to discount that trailing edge to get a realistic figure for future edge. That's tricky but it can be done. There are a bunch of statistical and heuristic tests that you can use to see how predictive a model is likely to be and I'd look hard at your 'edge finding' methodology to see how likely it was to find 'phantom' edges. Essentially the more complex and specific your methodology the less likely it is to be predictive.

These methods would give me a 'confidence' figure with which to discount my trailing edge to get a 'forwards' edge. My confidence would be expressed as a number between 0 (no confidence) and 1 (certainty). All of that gives me a simple function for assumed edge going forwards: my trailing edge times my confidence. Once I started betting I'd update my confidence using to my results and adjust the function accordingly. There you are, quantified edge and no need to 'guess' anything at all.

I'll come back later and explain how I might do a whole book from scratch because I think that's part of your question too but I'm finished my coffee and it's sunny outside so I'm off to ride my bike.
By:
aye robot
When: 16 Jul 19 11:30
DC Drifters don't necessarily show value because they may be drifting for a reason - the drift may reflect information like the horse not looking as good as was expected or whatever, or they may be drifting away from a price that was value (to lay) towards a neutral price.

It's overly simplistic to think that drifted horses must be showing value, that's not what value is. There is no inherent relationship between drifting and value - that's just some b0ll0cks that someone who doesn't understand value has told you.

HTH.
By:
Dr Crippen
When: 16 Jul 19 11:56
Well they will be drifting for a reason won't they.

Either weight of money on another runner or because the market knows or has seen something you don't know about.

So what do you do, you intended backing your horse because you think it is value at the price on offer, and it starts to drift.
So it's getting better value all time by your reckoning.

But you say not. So has its value become too good?

Do you back it or not?
By:
DFCIRONMAN
When: 16 Jul 19 12:36
Dr C - you say :-

Yet you can't even explain why you don't simply back horses that drift in the market when these types obviously provide the best value.


---------------------------------------------------------------------

Once a horse race has been RATED......the position of a horse in the ratings DOES NOT SHIFT due to odds drifting out in the market. The RATING of that race remains the same.

As long as the RANK position and the odds indicate VALUE or NO VALUE for any  FAVOURITES (top 5 say) in top 5 rated horses, then selections generally stand out on taking "other factors" into account.

AR's post at 11.30 am is clear re "DRIFTERS" and I agree with the 2nd paragraph!
By:
jamesdean
When: 16 Jul 19 13:27
Agree with ayerobot, the management 100%, well maybe 99%. They are obviously spot on.


Just a small curve ball as to being able to win long term without betting value.


example -

Two boxers having a fight.
Pre fight boxer b is 5/1, his true price should be 10/1, only has a punchers chance.

Boxer A is a specialist, all round superb boxer but his only flaw is he starts slowly.
Boxer B is all guts and glory, comes out firing from the first bell but no stamina to last 12 rounds and has a glass chin.

Now that 5/1 on boxer b represents bad value BUT as he will come out like a raging bull to a red rag and boxer A starts slowly,
it may look like to some that boxer b is doing well and has a better chance than 5/1 so he gets backed into 5/2. After 4 rounds he's blowing
out of his backside and boxer a is toying with him. The prices are now 25/1 for boxer b.

If a trader pre fight has backed boxer b at 5/1 (a true 10/1 shot) with the view of greening up in round 2 when boxer b is on top at say 5/2.
Is the trader then winning by taking a poor value price i.e the 5/1 on a 10/1 shot.


The last time I can think of a similar situation is when Connor Mgregor fought Floyd Mayweather, yet it probably happens all the time in other sports, horse racing with front runners that blast off but dont get the distance, golfers starting on the harder back nine than the easier
front 9 etc in bigger field competions.


It's the old adage of if you're out on a hike in the safari with your hiking partner and bump into a Lion, you don't have to outrun the Lion
just your hiking partner.
By:
G Hall
When: 16 Jul 19 13:44
Magic analogy jamesdean
By:
Dr Crippen
When: 16 Jul 19 14:11
At least DFCIRONMAN has explained where his view of value comes from.

But jamesdean's post adds nothing despite his obvious writing talents.

While the rest seem clueless in answering what is a very simple question.
How do you calculate accurately whether a bet is value or not?
By:
aye robot
When: 16 Jul 19 14:33
While the rest seem clueless in answering what is a very simple question.
How do you calculate accurately whether a bet is value or not?


Are you sh1tt1ng me? I've answered this to an almost absurdly comprehensive degree in various posts through this thread. Did you even read them?

Whatever. I hope it's useful to someone (I never really thought that person would be DC).

I wish you luck DC - but I'm done answering your questions. I might as well go and teach the neighbour's cat to play Tennis.
By:
aye robot
When: 16 Jul 19 14:41
Jamesdean - in your example the player is winning by taking the value lay at 2, his back at 5 is still poor value and is costing him. He's not winning by taking poor value - he's losing by taking poor value but coming out ahead by offsetting it with a good value bet that more than makes up for it.

Over a series of identical events he would come out way ahead by ditching the poor value back.

Hope that takes care of the 1%.
By:
jamesdean
When: 16 Jul 19 14:57
Haha yes I agree ar. One could argue though that the trader made his green by backing the 5/1 as that gave them the green to trade.

Just think traders and outright backers/layers may have different views. Maybe the trader sees the 5/1 as a value back as that gives him/her the position to green up at a later stage where the pre race backer/layer would just lay the 5/1 pre fight, just laying the 5/2..2/1 after 2 rounds is more the gamblers mindset that might not suit the trader.

Both the trader and the backer taking the 5/1, one backing, one laying and both believing they got the value.
If after years of betting, trading in 000's of markets both are showing healthy profits, it would be hard to argue otherwise.

p.s I know you and TM are correct in what you say and not trying to suggest otherwise, just a different slant on it.
It was suggested that no one can win longterm betting bad value, traders may see it differently even if they would be better off being outright gamblers/value takers.
By:
jamesdean
When: 16 Jul 19 15:10
I'll (try) back it up with another example.

The boxer is 5/1 (should be 10/1) The trader thinks it will go evens. If the trader was to risk £100 at 5/1 with the view
of greening up at evens, he'd be able to lay £300 back at evens giving him £200 green whatever happens for his £100 risk.


If he was just an outright backer/layer and waited until the boxer went evens, for his £100 lay risk at evens he'd win £100 yet he'd win £200
for his bad value 5/1 back then trade

Over a series of identical events he would come out way ahead by ditching the poor value back.


Why would he be better off ditching the poor value back when it's making him more money?
By:
aye robot
When: 16 Jul 19 15:19
You're saying that if he layed for £300 then he'd win more than if he layed for £100?
Er - yeah, he would.
By:
Dr Crippen
When: 16 Jul 19 15:20
Are you sh1tt1ng me?

Excellent! And that after posting this below:

These methods would give me a 'confidence' figure with which to discount my trailing edge to get a 'forwards' edge. My confidence would be expressed as a number between 0 (no confidence) and 1 (certainty). All of that gives me a simple function for assumed edge going forwards: my trailing edge times my confidence. Once I started betting I'd update my confidence using to my results and adjust the function accordingly. There you are, quantified edge and no need to 'guess' anything at all.

Can't knock your sense of humour.
By:
aye robot
When: 16 Jul 19 15:21
If the boxer is trading at 5 and you price him at 10 then why not just lay him at 5? If he goes to evens and you're still pricing him at longer than evens then lay him again.

Am I missing something?
By:
aye robot
When: 16 Jul 19 15:22
^ for jd
By:
jamesdean
When: 16 Jul 19 15:24
You're missing my point.

If he just waits and lays £300 at evens then he's risking losing £300.
If he has £100 on at 5/1 then trades £300 at evens, he's risked £100.
The trade works out at 2/1, the outright lay is an even money chance.
By:
The Management
When: 16 Jul 19 15:26
Dr Crippen    16 Jul 19 14:11 
While the rest seem clueless in answering what is a very simple question.
How do you calculate accurately whether a bet is value or not?

Your balance will tell you this or alternatively you can record various details about your BF activity on a spreadsheet.

My favoured measure pre-event (on some markets) is the price I entered the market at versus the Betfair SP. In other markets I might use the price I entered the market at versus the price I exited the market at. There are other measures you can use but as I have already stated - the ultimate measure over time and a volume of plays will be your balance!

Today (after a long patient wait), I entered one of the very few markets that interested me today at an average of about 1.78 because I calculated the price should have been around 1.45 (my expected SP). I could have traded out of my 1.78 bet (my pre-event entry point) at a pre-event low of 1.40. I did not and the bet went on to win quite comfortably at a generous (imo) SP of around 1.53.

I could tell you how I calculated my expected SP at 1.45 but I would then have to kill you. The reality is that whilst I use a lot of data to calculate my expected starting prices - I can also do it instinctively. I could have looked at (studied) this particular event literally for seconds and told you the SP should be in the range 1.40-1.50.

History, my balance and my spreadsheets tell me that I am right a lot more often than I am wrong - so I have a potential lay or back bet in this market if the price moves significantly (either way) out of my forecast range. This level of movement from my tissue price doesn't happen all the time and it certainly doesn't happen to the kind of volume that it used to - but it does happen.

I don't know what sport you follow DC (if any) but most punters can do this (instinctively calculate a price tissue) to an extent - Men's Tennis isn't my thing but I could have told you the Betfair SP's of the Wimbledon Finalists (ND v RF) within 2 ticks without having to look anywhere and without even switching my computer on. But like MOST markets I look at there was no opportunity there for me. Knowing the "right" price doesn't always result in being able to get a bet on at the "wrong" price.

I'm happy to admit I would struggle to price up a 30 runner 5F handicap - so I don't even try! - but just looking at the graphs tells me that (knowingly or not) some people have struck value bets and some people have not. The ones that do it persistently have a chance - the others do not.

I go back to what I think was the first statement I made on this thread - Price is Everything.
By:
jamesdean
When: 16 Jul 19 15:27
Why would you lay him at 5/1 as a trader if your view was he was a speedy starter and likely to trade much shorter, even if you made
him a 10/1 shot pre fight?
By:
jamesdean
When: 16 Jul 19 15:30
I agree wholeheartedly btw that price is everything.

Just in my opinion, the same price could be seen as both a value back and a value lay depending on whether you're a trader or a more
traditional gambler.
By:
Latalomne
When: 16 Jul 19 15:33
I think you're both right in the boxing example.  One is value and one is even better value.  Like AR says, in those circumstances, lay twice!
By:
aye robot
When: 16 Jul 19 15:45
If he has £100 on at 5/1 then trades £300 at evens, he's risked £100.

He's only risked £100 on his back bet - but it's not the same because he's risked it on a different outcome at a different time. After all there was no guarantee that the boxer would shorten as you expected. To work out the expected value of that whole proposition you'd have to account for the probability of the price shortening as you expected (or not). You can't just take the value of a trade as being its total value when it succeeds, that's like saying that bets are only good value when they win.
By:
aye robot
When: 16 Jul 19 16:09
...depending on whether you're a trader or a more traditional gambler.

From a maths perspective there's no difference between a trader and any other gambler. Both place a bet, wait for an uncertain outcome and then win or lose. That's it. Personally I don't much like the term 'Trader' because I think people hide behind it and it can sometimes lead to a degree of denial, which can cause trouble.

I'm not saying that's your case though and I don't want to pick a fight with you though - you obviously do get the fundamental concept of value.
By:
jamesdean
When: 16 Jul 19 16:09
ofc but by the same token as saying he may not go evens, he may also not lose the fight so you'd also
lose as an outright layer just laying the evens or even worse laying the 5/1.

The trader is always only risking £100 to win 200 though (providing he goes evens) the outright gambler is risking £200
to win his £200 if/when he goes evens.

The trader isn't interested in whether he wins or loses the fight, he's just interested in him starting well and that's where
he see his value at backing 5/1 pre fight just as the outright layer isn't interested in how well the boxer starts as long as he loses
and hes got his value laying the 5/1
By:
jamesdean
When: 16 Jul 19 16:14
I'm not picking a fight either, as said I agree with you but I do think the same price can be value as both a back and a lay
depending on your m/o (if you don't like the term trader, will go for m/o)

It's not a fight or argument btw, I'll leave that to those who don't think value matters. Just a different point of view
By:
ericster
When: 16 Jul 19 16:28

Jul 16, 2019 -- 4:14PM, jamesdean wrote:


I'm not picking a fight either, as said I agree with you but I do think the same price can be value as both a back and a laydepending on your m/o (if you don't like the term trader, will go for m/o)It's not a fight or argument btw, I'll leave that to those who don't think value matters. Just a different point of view


"Just a different point, or different points, of view. Isn't that, in essence, what a betting market is?
As with any other market, we're all looking for a bargain. A freebie even.
And no jd it's not a fight, I congratulate you guys for keeping this thread as it was intended, a friendly exchange of views.

By:
aye robot
When: 16 Jul 19 17:03
The trader is always only risking £100 to win 200 though (providing he goes evens) the outright gambler is risking £200 to win his £200 if/when he goes evens.

The clue is in this sentence - the trader is ALWAYS risking £100, the outright gambler is only SOMETIMES risking £200. his average risk is less.

Take all the qualifying markets in which you might place this trade:
In your scenario the trader is risking £100 on every market. His expected return is not £200, it's less than that because his trade doesn't always work out. On those occasions when the trade doesn't work out the outright gambler's risk is zero, so his average risk is on all these markets is less than £200.

If you want to work out the expected value of the trade then you COULD assign probabilities to all the outcomes - fighter winning, trade succeeding etc and work it through like that but there's no point - whatever numbers you plug in the expected value of a trade is always reducible to the sum of the expected values of the two bets. If one is negative then that drags it down, no ifs no buts.
By:
jamesdean
When: 16 Jul 19 17:22
Again agree, however again you seem to be missing my point.

Last time from me because although we seem to agree mostly we're going around in circles.

My point - The same price could be value to both a backer and a layer depending on whether theyre a trader or outright gambler.

The trader may think (only time and 000's of bets would tell) that there is more value in backing the 5/1 and the bet going evens
than laying the 5/1 even if he thinks outright, the 5/1 would be poor value. The trader does not have to beat the true odds of the outcome
just the market odds at the first bell to the end of the 2nd round.

Just as the hiker does not need to out run the lion, just his hiking partner.
By:
aye robot
When: 16 Jul 19 17:33
Sorry JD but If a selection at is trading at it 5 and you're pricing it at 10 (and you're right) then you can never make that into a value back whatever you do. It will still only win 1 time in 10 and only pay out at 5 when it does. There's just no sequence of bets that will ever change that. The fact that might lay the same selection at a different time is neither here nor there - you still lose your money on that bet. Nor does that bet decrease your liability on a subsequent lay bet because that lay bet will sometimes not come - your overall liability against profit is increased not decreased.

Value is maths and maths never compromises. You can 'see' a bet as value but that changes nothing. If it doesn't show a profit on average then see it how you like it's not value.
By:
aye robot
When: 16 Jul 19 17:41
The trader does not have to beat the true odds of the outcome

that's your mistake - he does. Those bets where he fails to do so reduce his profits - which is the same as any other kind of losing.
By:
jamesdean
When: 16 Jul 19 17:45
It will still only win 1 time in 10 and only pay out at 5 when it does.

Ofc it will only 'win' 1/10 but the trader ain't bothered about how many times
it 'wins' just how many times it goes evens, after that it doesn't matter to him.


Agree value is the only way but one mans junk is another mans treasure.
By:
aye robot
When: 16 Jul 19 17:47
Will the backs he places at 5 show a profit? No. They'll show a loss.

It's that simple.
By:
unitedbiscuits
When: 16 Jul 19 19:45
Don't try to be a professional gambler - you won't gain any advantage from giving up your job, your decisions won't improve by going back on the same form a dozen times. Aim for a useful second income outside of working hours and, if it overtakes your salary to a great extent, be a lazy b**tard, and spend the time you would have been working sunbathing, gardening and making love.

Full time gamblers on betfair tend to be crabby individuals complaing about the premium charge, the price of petrol or piles. Don't be like them.
By:
Dr Crippen
When: 17 Jul 19 13:34
Don't try to be a professional gambler -

Well none of them know half enough for that anyway from what I've read here.
They're still at the novice stage of backing a horse because they think its a good price.
Or traders who simply ignore the form altogether.   

And when they come out with stuff like ''the price is everything'' it just makes me laugh.
If they can't see the blindingly obvious that it has to win first, then there's no hope for them at all.
By:
jamesdean
When: 17 Jul 19 14:01
I'm going to give you the benefit of the doubt and say you are a master fisherman, Dr.C.
The alternative is much worse.
By:
Dr Crippen
When: 17 Jul 19 20:20
Yes jamesdean but it were you who engaged on a pointless argument with aye robot. And you're on the same side.
Which shows how value in is the eye of the beholder, but winners are winners and there's no arguing about that.
By:
peckerdunne
When: 17 Jul 19 22:07
Anyone can find winners, no magic in that,however your'e strike rate is crucial and the return from it plus or minus will be totally dependent on price.

I know plenty of form students who can never make it pay but could tell you the history of every horse in the race.

I was that person for years, working in the industry and barely breaking even at times.

Form is not everything and neither is picking winners or losers for that matter.

Assessing value is subjective, its not value as a stand alone in a single market.

Value is the big picture.
By:
DFCIRONMAN
When: 18 Jul 19 13:18
DFCIRONMAN    13 Jul 19 13:50 
If you rate every horse in a race, then their RANK position can indicate "VALUE" when you check their ODDS against RANK position.

Of course you have to use MANY FACTORS to rate each horse. You also have to be "consistent" with points you allocate to each horse. There are "subjective" decisions made as well............and you have to "good" at making them, otherwise the RATINGS would not be good enough to provide consistency of HIGH STANDARD ratings.
-----------------------------------------------------------------

"subjective" input as well is a key word to enable arriving at RATINGS
that produce a high standard  where "VALUE" can be identified .
By:
DFCIRONMAN
When: 18 Jul 19 13:29
"Flatstats" site have  useful information on "VALUE BETTING"( worth a read).......though not convinced re how they use "probability" %s etc to arrive at VALUE selections.

Using crude RANK position and the ODDS to indicate VALUE re the top 5 favourites in market and where they are in RANKINGS for a race seems to be easier to "see" VALUE. Perhaps I'm wrong thoughTongue Out
By:
aye robot
When: 18 Jul 19 14:15
Perhaps I'm wrong though

Yeah 'fraid so but I've gotten bored of this.

That's not to say that your methodology couldn't find value bets, it could and I think your approach is perfectly reasonable, but your maths is overly simplistic to be technically correct.

More broadly:

This is maths, and talking about maths isn't like talking about whether you like Jazz or not; it doesn't matter how you see it or what you're opinion is - you're just either right or wrong.

The 'debate' was settled hundreds of years ago by Bernoulli, Pascal and Huygens. Their insights are now axiomatic and that's that - it's over. Different perspective? Wrong. I see it like this..? Wrong. I don't like it...I don't believe it.. sometimes there's an exception? Wrong, wrong, wrong.

Don't like it? Tough.
Page 4 of 10  •  Previous | 1 | 2 | 3 | 4 | 5 | 6 | ... | 10 | Next
sort by:
Show
per page

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
‹ back to topics
www.betfair.com