NEW! ParlayMaker Version 1.4.0

Questions, suggestion and tips regarding the use of the ProlineParlayMaker program
ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:14 am

Well I've been burning the midnight oil but here it is at last.

As of a couple of years back the OLG started using some even number spreads on their pointspread offerings. With football and basketball I was aware that this introduced a small margin of error into the values both on my site and in the ParlayMaker program, but these were not (imo) significant enough to warrant any changes.

But now we have baseball and even hockey back on pointspread. In the NHL with 1 goal hockey spreads, where a push will hit at rate of about 25% this changes things and we now have to consider the effect this has on our plays.

And that's what Version 1.4 takes care of.

I've spent the last several days trying to eliminate any bugs. I've one problem left that I'll mention later but on the whole it seems to be functioning as planned. I'm sure there's still a few bugs in there but we'll deal with them as they show up.

Also trying to verify the results the program produces was not easy to say the least. I think I've got everything covered but if anyone out there has done programming for the same (Dilbert?, Matty? Leo?) could you let me know if your numbers match up (or not).

Also when running the program for the first time duoble check that your settings are as you expect them. Although I've tried to maintain the user settings from the previous version, the way some things are handled has been changed and some settings might be lost.

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:14 am

All changes in this version relate to the Ontario (OLG) Lottery only. The proline tab now accurately handles totals with a push line. Aside from that everything else is for the Pointspread plays and parlays.

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:15 am

The first set of changes are with how the program relates to the web site. Recently a few changes where made re OLG pointspread displays. Instead of 4 teams the site now uses the payout level of a 5 team parlay (1.82) as the basis of all expected win calculations. Also instead of converting even number spreads to straight win or loss probabilities the site now shows the probabilities for all possible outcomes - Visit / Push / Home.

The program has been change to match these. The text box displaying each game's win probability will, if a push is possible, show the probability of both the win and the push with the two values seperated by a '/'. Also both values are used to calculate the edge at a default payout level of 1.82.

Another change is that when you select add tab/pointspread, 'Nhl' has been added to the sports options list.

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:17 am

How pointspread parlay edges are worked out!

This is where the major upgrade has been done!

In the previous version for each parlay the program calculated the probability of a win. But rather than considering pushes it just used an adjusted win and loss value - same as the values on the site.

But now each parlay is broken down into all possible combinations of wins and pushes. In some parlays there could be hundreds of possible sub-parlays that are analyzed for probabilities and payout levels. For each parlay the program cycles through all the possible variations (or sub-parlays as I call then) and then combines these for a final parlay value.
If you look at the parlay list you'll notice that the values of the Payout and Prob columns seem inconsistent with the value in the Edge column. How the edge was determined can be better seen by looking at the PayAvg column (newly added). As the name implies this is the average payout of a winning ticket factoring in all possible combinations of wins and pushes.

The above principles have also been applied to the analysis of 10-11-12 team parlays. In the past some short cuts were used here due to the complexity involved in getting the correct values. These are also calculated precisely now using all possible paying combinations of wins, losses and pushes. (In case you were wondering the short cuts did prove in comparison to be very solid - there is almost no difference in these values as compared to the previous version's).

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:18 am

Boxing Additional Games

This has been slightly changed again so I'm going to cover all of it once more.

First you need to understand why we might want to box in an additional game. This is especially true for both 2 team parlays and for 6 team parlays. Both of these payout at a higher level with a boxed game. An extension of this is that there may also, because of possible pushes, be a benefit in boxing in an additional game for 3,4,7,8 team parlays. If you don't understand this concept then I suggest reading the article 'Pointspread Basics' on this site.

First there is the 'Box Games' check box. You can consider this as sort of a master switch to quickly turn of or on all boxing. If this is not checked then no parlays will have a box game added.

So after checking the 'Box Games' option we can move on to the 'Box Options' drop down list. Whether or not you want to print the extra tickets required to box games for each parlay size can be set here.

Note that checking the option does not mean that an additional will be automatically boxed in. What it means is that the program will check to see if there is a benefit in boxing in a game, and if yes it will.
For example if you check 'Box Up 7 Teams' then the program will check each 7 team parlay played and determine if a box helps. The key being can this end up as a 6 team parlay? If yes then a box is needed to increase the payout. So a 7 team parlay with no even number spreads would not ever have an additional game added. But if there is at least one possible push play in the parlay then a boxed game will be added. For 'Box Up 8 Teams' it is the same except now there must be at least two possible push plays for the boxed game to be added.
By the way a 6 teamer will always have the game boxed in as it could of course end up being a 6 teamer.

The same as above applies to both 3 and 4 team parlays with the key now being can it end up a 2 teamer.

In previous versions the box games feature was only used for the actual printing of tickets. In V1.4 the box options are also factored into the creation of the parlay list with regard to calculating edges.
Let's say you're playing a set of 7 team parlays. The program calculates the various outcomes for a parlay and finds a possible winning combination of 6 wins and 1 push. What should the program use as a payout on this parlay - 35-1 or 37.5-1? The answer will depend on whether or not the 'Box Up 7 Teams' is checked.

I had a bit of fun playing around with different setting to see how much of a gain there was in boxing in games. The panel on the right with 'Print/Summary/LLE' is useful for this. Click on the 'Summary' tab so that you get an immediate overview of the different outcomes. Try running of a set of 7 team parlays with the boxup7 checked and also unchecked. The 2nd line in the summary panel shows the difference in net expected win.
I found myself that for football there was a enough of a gain with 7 teamers to make in worth while but I would not bother with the extra printing on the 8 teamers. In hockey even for 8 team parlays there was a significant gain for boxing in an additional game.

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Thu Oct 17, 2013 11:32 am

I've been testing this for about 5 days and been comparing the results to running the same lists through the old version.

For football it was much as we all expected. The edge values were just a shade lower. No real difference after setting Kelly bet sizes.

More interesting was comparing the result for NHL pointspread.
Again as expected the edges were much lower than the previous version. Also as expected the Kelly bet fraction went up because of the higher win probability.
What surprised me (but hats off to MattyKGB again!) was that in the end the expected win with the pushes factored in actually went up a little.
It's true that the earlier version over estimated edges. But it also under estimated the win probability resulting in the Kelly factor being set too low.

PLP

Kaboshed
Posts: 94
Joined: Fri Aug 02, 2013 1:26 pm

Re: NEW! ParlayMaker Version 1.4.0

Post by Kaboshed » Fri Oct 18, 2013 11:07 am

When trying to generate proline plays the program is encountering an error and closing.

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Fri Oct 18, 2013 11:48 am

Kaboshed wrote:When trying to generate proline plays the program is encountering an error and closing.

It may be reading a file with an outdated format.

Try this. Go to your documents folder and delete the ProlinePlayer folder. Then try again and let me know what happens.

PLP

Kaboshed
Posts: 94
Joined: Fri Aug 02, 2013 1:26 pm

Re: NEW! ParlayMaker Version 1.4.0

Post by Kaboshed » Fri Oct 18, 2013 12:00 pm

Still getting the error, proline list is populated with -10 min edge, parlay list with min 5.

"Collection index must be in the range 1 to the size of the collection"

ProlinePlayer
Site Admin
Posts: 2076
Joined: Sun Mar 31, 2013 6:33 am

Re: NEW! ParlayMaker Version 1.4.0

Post by ProlinePlayer » Fri Oct 18, 2013 1:10 pm

Kaboshed wrote:Still getting the error, proline list is populated with -10 min edge, parlay list with min 5.

"Collection index must be in the range 1 to the size of the collection"
Thanks,

see the problem
will look try to find the cause

PLP

Post Reply