Increase or Decrease 90% Reliable Prediction Tip

[Alcymart]

What I found by experimenting with my 6/49 Lottery is that the Value Predictor Plug-in with certain settings yields 90% success rate on predicting if the value will increase or decrease.

Some of you may want to experiment. You need to be a registered member to use the Value prediction Plug-in.

As for the predicted Value it predicts, it's not accurate. But knowing if it will increase or decrease means much less bad tickets in the package so this should help out a few.

Experiment with your own Lotteries and see if you acheive 90%+ success rate with your lotteries you have, and I would appreciate any feedback from those who will experiment if we are on the right track.

I attached a screenshot of the settings you must put after importing the data:

[jimbo]

Alcymart
what stats are you importing? some of those settings didnt work here in ny, but wasnt sure of data to use.

i'll try anything to hit 3 out a 5

have fun
jim

[Alcymart]

Stts are taken from Quebec 6/49. I tried it with other lotteries and it doesn't work out as well. However I did see that when you play with the settings, you will eventually find good predictions coming out. I did find that the pattern size from 2 to 3 is the best prediction in most all lotteries. I don't think it's the ultimate way to win this way, however if you want to win some smaller prizes, it's not that hard. Filter 1 column at a time until there is about 12 - 20 tickets left. At about column 3. If you filter column 4 too, you end up with 1 ticket which is the jackpot, but at 3 columns you win multiple prizes as numbers are all similar. For this you must predict the 3 or 4 first columns right on. You need to predict the exact value that will come out in all 3 columns or 4 columns if you are looking for the jackpot. Basically, you don't need to predict 11 columns to win. Hope this helps.

[Midas]

Bernard,

Pleased to make your acquaintance. Appreciate your efforts, you are the first EL user to review value predictor!.

Without putting to much emphasis on prediction capabilities, each user should develop their own strategies for their gameplay. Consider value predictor like all the other great functions within EL as an experience, it's a matter of personal play. The product is constantly evolving and there is always good dialogue behind the scenes in bringing out the best .

I encourage all to comprehensively explore every feature, develop strategies, tabulate and share their findings. Here are some tips, which I have found similar to yourself when using value predictor.

(1) Weight Size = 1,with steps (0.1-0.3) gives good results.
(2) Vary pattern sizes around the best predictor pattern size.
(3) Best to set Last Rows to Ignore = Rows for Validation. I regularly set Last Rows to Ignore one greater than Rows for Validation to get a prediction for the trend. Values are useful for step ahead predict.
(4) Try this stack attack for consistency. Take similar databases from your nearest state/country. Stack them one upon each other with your lottery as the latest forming one large database. This provides good database quantity for value predictor to chew. In my case I loaded the following (6/49) databases, Loto France then Spanish Primitiva then our UK National Lottery. Total (5000+) records!.

Note for the first (200-250) records in any lottery should be deleted as the lottery is climbing to it's normal range. Check out the results for your lotto. Interesting ?.

There is plenty to explore. Good Luck.

Best Regards
George

[stan]

[quote=George]
(3) Best to set Last Rows to Ignore = Rows for Validation. I regularly set Last Rows to Ignore one greater than Rows for Validation to get a prediction for the trend. Values are useful for step ahead predict.
[/quote]
there might be a slight misunderstanding on how the input parameters are interpreted. say your input data have 100 rows, if you set last rows to ignore=10 and rows for validation=30 then the predictor will treat the data as if the last row was row number 90 (the first data row being row #1) so you should compare the prediction result with the 'actual' values from row #91.
rows for validation=30 that the predictor will first make prediction as if row #60 was the last row in your input page, then it will make prediction for row #61, then #62 etc up to row #89. then the best pattern size is chosen from these 30 prediction sets (based either on lowest mse or on most valid trend predictions) and the winning pattern size is used to make the actual prediction for row #90

i hope it's a bit clearer now...

[Midas]

Appreciate the correction Stan,

This is a great pinpoint of where I was going wrong, it finally concludes our long running debate on how I was getting strange trend prediction results.

Thank you
George

[scenium]

Hi all,


Nice to be part of Expert Lotto community.


Stan, is there an option to uncheck skip winning numbers at one go,
instead of unchecking one by one. Quite troublesome if I do a lot of skipping for testing.

Regarding the predictor, if we are able to get the best pattern sizes from
let say the past 30 validation. Does it mean that the pattern size(let say pattern 7) give the correct sum value for a particular columm for that day?
Is this 100% true based on the past data provided?
Is it possible to provide a statistic on the hit pattern size for the past drawings?

Thanks!

[stan]

[quote=scenium]
Stan, is there an option to uncheck skip winning numbers at one go,
instead of unchecking one by one. Quite troublesome if I do a lot of skipping for testing.
[/quote]
there isn't. but i was thinking of extending the winning numbers selection panel to add a new option "all draws but the last one" or "all draws but the last xx draws"...

Regarding the predictor, if we are able to get the best pattern sizes from
let say the past 30 validation. Does it mean that the pattern size(let say pattern 7) give the correct sum value for a particular columm for that day?
Is this 100% true based on the past data provided?

the resulting "best pattern size" means that it provides the most valid trend predictions for the xx select validation rows. so it's only an indicator that the predicted value/trend might be valid in the actual coming draw. i'm sorry but there's never 100% of anything when playing a lottery
so even your results happen to show 100% success on past validation rows the predicted value may be wrong for the next draw.

Is it possible to provide a statistic on the hit pattern size for the past drawings?

don't know what you mean. but pls keep in mind that the predictor plugin is just an experiment at the moment. when/if i'm satisfied with its results i'll integrate it fully into expert lotto (so you'll have predicted drawn numbers and predicted statistics values in winning numbers stats)

[scenium]

the resulting "best pattern size" means that it provides the most valid trend predictions for the xx select validation rows. so it's only an indicator that the predicted value/trend might be valid in the actual coming draw. i'm sorry but there's never 100% of anything when playing a lottery
so even your results happen to show 100% success on past validation rows the predicted value may be wrong for the next draw

Hi Stan,

I am actually referring to the pattern size for the past draw. Is it the best pattern size is use for future draw, because during the prediction calculation of the past draw, that pattern give a correct sum value?
I just would like to know that if you are able to get an exact matching pattern & sum value for the past data.



Thanks!

[stan]

[quote=scenium]

the resulting "best pattern size" means that it provides the most valid trend predictions for the xx select validation rows. so it's only an indicator that the predicted value/trend might be valid in the actual coming draw. i'm sorry but there's never 100% of anything when playing a lottery
so even your results happen to show 100% success on past validation rows the predicted value may be wrong for the next draw

Hi Stan,

I am actually referring to the pattern size for the past draw. Is it the best pattern size is use for future draw, because during the prediction calculation of the past draw, that pattern give a correct sum value?
[/quote]
the pattern size is chosen because it has the lowest mse error (the first set of results) or because it lead to most valid *trend* predictions (the second set of results) when applied to validation rows

I just would like to know that if you are able to get an exact matching pattern & sum value for the past data.

this is already in progress, it will be in the next plugin version (by the end of this week)