The math behind multis

[Dawggin It]

Ok so maybe this isn't the best thread for this question but I'll give it a go.

Even as a mug punter, I still have some limited understanding of how utilising a Kelly model can be a mathematical approach to gambling when your evaluation of a game differs to the bookie. What I can't figure out, and would like someone to explain to me, is how the hell do you convert an opinion about the likely outcome of an AFL match into a percentage chance? It still seems to be applying supposedly objective equations to a largely subjective opinion about a game.

Take north vs crows tmrw as an example. I think Adelaide will win, but ask me what percentage chance they are of winning.... How do ppl even come up with a ball-park figure let alone percentage chance of covering a line, then using that guestimation to guide your stake.

I get that that is exactly what bookies are doing when they set the odds, but to arrive at a figure of 62% vs 70%.... How do you do it?

 

[brett128]

Something I kacked up a few years ago playing around with my first model:

For home team:

=((NORMDIST(0.5,PREDICTED MOV,999,TRUE) - NORMDIST(0.5, PREDICTED MOV, 37.85,TRUE))

For away team:

1-home team

<=== The above is a rudimentary points to probability converter for excel. Converting points to % would be a start to finding bets that you think are value.

Predicted MOV is your estimated margin of victory. Home will be a + value, the away team will be a -value.

So Melbourne vs GWS today will be say (+)10 points, whilst North vs Adelaide would say, have a predicted MOV of -10 points (an away victory).

37.85 is the standard deviation of MOV in the AFL.

Hope this helps.

 

[Donkson]

why is a guy with a .2% edge handing out advice like its gospel?

 

[Borgsta]

Cos he has a model and has been to uni and can use "yank"terminology

 

[Daytripper]

I do use Kelly but only use it as a guide for what my stake could be.

As pointed out earlier an extended run of outs can wipe your bank completely out and every punter has at least 2 or 3 runs of extended outs during any given year. Anyone who says otherwise is lying.

Brett - I would be wary of using lines exclusively for sports betting. AFL teams can and do 'pull the pin' when a long way in front by taking their best players off for a rest. If you are going to bet lines then in my opinion its best to restrict to lines 15 and under.

Betting lines for NBA, NFL is even riskier is there is no percentage component-ie there is no incentive for a team to win by as much as they can. Hence you can see occasions when the entire starting 5 (NBA) is rested for the last quarter playing havoc with any line betting.

 

[madpunter]

Something I kacked up a few years ago playing around with my first model:

For home team:

=((NORMDIST(0.5,PREDICTED MOV,999,TRUE) - NORMDIST(0.5, PREDICTED MOV, 37.85,TRUE))

For away team:

1-home team

<=== The above is a rudimentary points to probability converter for excel. Converting points to % would be a start to finding bets that you think are value.

Predicted MOV is your estimated margin of victory. Home will be a + value, the away team will be a -value.

So Melbourne vs GWS today will be say (+)10 points, whilst North vs Adelaide would say, have a predicted MOV of -10 points (an away victory).

37.85 is the standard deviation of MOV in the AFL.

Hope this helps.

This formula doesn't work as is. For the home team, wouldn't it be 0.5 - (your formula)?

 

[madpunter]

There's another problem with that formula... it can't handle large predicted margins, like > 100 points.

 

[brett128]

There's another problem with that formula... it can't handle large predicted margins, like > 100 points.

I'm assuming you've got it to work now.

The Gaussian distribution breaks down at the tails because of the game's standard deviation. Trust me, this formula is fine for 99.9% of games. I used it until I completely did away with converting points to probabilities.

If you're estimating MOV's at >100 points, your AFL modelling has problems anyway.

 

[Kynge of Begrem]

Hey everyone, I've only really scanned this board the last few days, but it seems multis are all the rage here. Yes, I use multis too, but they are only the small part of a kelly optimal solution in maximization of expected bankroll growth.

If you're a zero edge sports bettor (and tbh, if you haven't come across terms like the 'kelly criterion' in your sports betting forays before, you most likely have zero edge);

Your goal should be to:

To lose money as slowly as possible whilst maximising your expected utility (entertainment/contentment).

That's the only way to rationalize in my own mind why zero edged punters bet.


Ganchrow at SBR probably best described the math behind multis with this article:

http://forum.sbrforum.com/handicapper-think-tank/2071-parlays.html

Notice how a 50% ATS (zero edge) bettor betting at -110 ($1.91) compounds 3 individual bets at negative edge of -4.5% on each individual bet to a multi bet with an edge of -12.5%.

Once you get into 4 + leg multis, the vast majority of people are better off going to their local tabaret and losing their money more slowly playing the pokies (whilst hopefully deriving more entertainment).

Anyone curious about the Kelly Criterion should read this:

http://en.wikipedia.org/wiki/Kelly_criterion

Cheers

Ah, good times

 

[KardiniaPark]

 

[Lensen]

Can not believe how many noobs post on here and other punting boards across the web. Anyone that relies on a model in sport is in for a heartache as sport does not follow cyclical variations. Teams do what they wanna do and will continue to do so and the kelly model doesn't take into account injuries. I have a couple of rules when it comes to multis:

1. If a game is 50/50 have it as your first leg and the near certainties last. This saves you a lot of heartache as shit does happen which i have learnt.
2. Always hedge on your last leg NO MATTER WHAT.

3. $1.01 bets are always certainties. Include them for that extra value

 

[Nothing]

3. $1.01 bets are always certainties. Include them for that extra value

I like the always hedge rule. It's a great way of ensuring the bookies never win. Except for the multis that don't make it to the last leg.

 

[Lensen]

$1.20-$1.30 bets are near certainties also and so are HT/FT if you believe you have a sure thing and want extra value. I have a system in place since i started punting properly and it's served me well the past two months increasing my bankroll by 300%.

two months is certainly a statistically significant timeframe to now believe you are an expert and your method is foolproof

 

[Kynge of Begrem]

I liked it when he called brett128 a noob

 

[Kynge of Begrem]

$1.20-$1.30 bets are near certainties also and so are HT/FT if you believe you have a sure thing and want extra value. Games like that Freo V Adelaide one are a case in point. As someone said they should have been $1.08. I got them at $1.36 ditto the hawks over the bombers which they were never going to lose. I have a system in place since i started punting properly and it's served me well the past two months increasing my bankroll by 300%.

Please tell me that you have a website where I can buy your system.

 

[Lensen]

Well it is what it is. I've been punting off and on a lot longer than that with honestly more failure than success with a horror run between rounds 7-10 this year with desperate 1-39 and other margin bets to chase losses and it's only since i started with near certain multis that i've started to have success. No betting strategy is fool proof but i'd like to hear yours and your success rate.

KardiniaPark is a massive advocate for my punting successes, i'm sure he can say what i am too bashful to say

 

[Kynge of Begrem]

Ah Harry, where were you a couple of hours ago when I was trying to kill some time

 

[Watts4Dinner]

Hi user Harry_The_Horse . Loving your work Pal. I'll make it as easy as possible.

- You should have something called 'PrtScn' button on the top of your keyboard. Press that.
- Open paint and press PASTE, screen grab will appear.
- Press crop and surround the area you would like to show.
- Right click CUT
-Open a new window and don't save the old picture you have cut up.
- Paste again and VOILA.
- Save to your desktop for easy access

Go to www.imgur.com and upload :

Eg1

 

[Kynge of Begrem]

No betting strategy is fool proof but i'd like to hear yours and your success rate.

betting on Melbourne

 

[asanque]

I'd appreciate it if somebody taught me how to take a screen shot using paint. I hate computers and technology.

 

[Watts4Dinner]

betting on Melbourne

Yeah shocking punter but very easy on the eye.

 

[DragzLS1]

Can not believe how many noobs post on here and other punting boards across the web. Anyone that relies on a model in sport is in for a heartache as sport does not follow cyclical variations. Teams do what they wanna do and will continue to do so and the kelly model doesn't take into account injuries. I have a couple of rules when it comes to multis:

1. If a game is 50/50 have it as your first leg and the near certainties last. This saves you a lot of heartache as shit does happen which i have learnt.
2. Always hedge on your last leg NO MATTER WHAT.

This is exactly right. I leave my last leg as a head to head bet. This weekend it was Essendon in a 5 leg multi that included north head to head (the 50/50 you spoke of as your first bet) then into west coast line, St kilda super line, Freemantle and dogs line.

Lucky I hedged by bet Sunday morning with a bet on the dogs at the super line +52.5 into the pies head to head.

I know it was risky but I was confident.

Always hedge your bet kids or you will regret it, or you will be one of those people who always talks about the 1 leg/bet that they missed "my last leg stuffed my multi, I was 2 points off ect ect" I have been there done that many times.

Good luck and happy punting.

 

[X_box_X]

This is exactly right. I leave my last leg as a head to head bet. This weekend it was Essendon in a 5 leg multi that included north head to head (the 50/50 you spoke of as your first bet) then into west coast line, St kilda super line, Freemantle and dogs line.

Lucky I hedged by bet Sunday morning with a bet on the dogs at the super line +52.5 into the pies head to head.

I know it was risky but I was confident.

Always hedge your bet kids or you will regret it, or you will be one of those people who always talks about the 1 leg/bet that they missed "my last leg stuffed my multi, I was 2 points off ect ect" I have been there done that many times.

Good luck and happy punting.

Rubbish.

In normal circumstances, you shouldn't be hedging any head to head bets in the final leg, unless:

1) Your payout is going to be substantial to you;

2) You can get a return where the mark up for the last leg is in your favour. I.E. Initially your final leg was on a team to win at odds of $1.87. You decide to hedge your final leg either before the event begins, or during the event (if live betting is available), as long as the odds are $2.16 or greater (this would guarantee you of at least a very small profit), as the mark up for the bookies is under 100%;

or 3) In some cases, you may find a good situation in a line bet. This can sometimes be the case if the final leg of your multi involves a favourite to win (paying $1.50, for example). If you are anticipating a close game, you may wish to back the underdog at the line (say +15.5 in an Aussie Rules game, paying $1.92). This at the very least will guarantee you make just shy of half your payout, but of course if the favourite wins by fewer than 15.5 points, your profit obviously increases greatly, as both your multi and hedging bet are both successful.

 

[Dominion]

Can not believe how many noobs post on here and other punting boards across the web. Anyone that relies on a model in sport is in for a heartache as sport does not follow cyclical variations. Teams do what they wanna do and will continue to do so and the kelly model doesn't take into account injuries. I have a couple of rules when it comes to multis:

1. If a game is 50/50 have it as your first leg and the near certainties last. This saves you a lot of heartache as shit does happen which i have learnt.
2. Always hedge on your last leg NO MATTER WHAT.

1. Remember to ensure the AFL schedules the games to your liking.
2. Use a strategy that is strictly worse than "Always leave out your last leg NO MATTER WHAT"


Solid.

 

[laam]

2. for single bets, in theory this is the last leg of your multi, therefore you should hedge any single bets straight away. This will guarantee you get money back on every bet!

Hopefully this extrapolation shows how dumb ALWAYS hedging the last leg of a multi is. If not, then punting might not be for you.